@article{
 title = {Object Classifications by Image Super-Resolution Preprocessing for Convolutional Neural Networks},
 type = {article},
 year = {2020},
 pages = {476-483},
 volume = {5},
 publisher = {ASTES Journal},
 id = {baa13eb8-e869-37bf-a2a2-59f4df2a37f5},
 created = {2020-04-21T18:31:47.228Z},
 accessed = {2020-04-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T18:31:47.310Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 bibtype = {article},
 author = {Na, Bokyoon and Fox, Geoffrey C},
 doi = {10.25046/aj050261},
 journal = {Advances in Science, Technology and Engineering Systems Journal},
 number = {2}
}

@inproceedings{
 title = {Understanding ML Driven HPC: Applications and Infrastructure},
 type = {inproceedings},
 year = {2020},
 pages = {421-427},
 websites = {http://arxiv.org/abs/1909.02363},
 month = {3},
 publisher = {Institute of Electrical and Electronics Engineers (IEEE)},
 day = {20},
 id = {08bdabc6-5175-3e1f-9ec4-8322eb27dfdf},
 created = {2020-04-21T19:52:56.824Z},
 accessed = {2020-04-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T19:52:56.896Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We recently outlined the vision of "Learning Everywhere" which captures the possibility and impact of how learning methods and traditional HPC methods can be coupled together. A primary driver of such coupling is the promise that Machine Learning (ML) will give major performance improvements for traditional HPC simulations. Motivated by this potential, the ML around HPC class of integration is of particular significance. In a related follow-up paper, we provided an initial taxonomy for integrating learning around HPC methods. In this paper, which is part of the Learning Everywhere series, we discuss "how" learning methods and HPC simulations are being integrated to enhance effective performance of computations. This paper identifies several modes --- substitution, assimilation, and control, in which learning methods integrate with HPC simulations and provide representative applications in each mode. This paper discusses some open research questions and we hope will motivate and clear the ground for MLaroundHPC benchmarks.},
 bibtype = {inproceedings},
 author = {Jha, Shantenu and Fox, Geoffrey},
 doi = {10.1109/escience.2019.00054}
}

We recently outlined the vision of "Learning Everywhere" which captures the possibility and impact of how learning methods and traditional HPC methods can be coupled together. A primary driver of such coupling is the promise that Machine Learning (ML) will give major performance improvements for traditional HPC simulations. Motivated by this potential, the ML around HPC class of integration is of particular significance. In a related follow-up paper, we provided an initial taxonomy for integrating learning around HPC methods. In this paper, which is part of the Learning Everywhere series, we discuss "how" learning methods and HPC simulations are being integrated to enhance effective performance of computations. This paper identifies several modes --- substitution, assimilation, and control, in which learning methods integrate with HPC simulations and provide representative applications in each mode. This paper discusses some open research questions and we hope will motivate and clear the ground for MLaroundHPC benchmarks.

@inproceedings{
 title = {Learning Everywhere: A Taxonomy for the Integration of Machine Learning and Simulations},
 type = {inproceedings},
 year = {2020},
 pages = {439-448},
 websites = {http://arxiv.org/abs/1909.13340},
 month = {3},
 publisher = {Institute of Electrical and Electronics Engineers (IEEE)},
 day = {20},
 id = {f16ae0e7-93f8-394e-9c0e-fd24cdc1c82e},
 created = {2020-04-21T19:54:31.915Z},
 accessed = {2020-04-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T19:54:32.002Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We present a taxonomy of research on Machine Learning (ML) applied to enhance simulations together with a catalog of some activities. We cover eight patterns for the link of ML to the simulations or systems plus three algorithmic areas: particle dynamics, agent-based models and partial differential equations. The patterns are further divided into three action areas: Improving simulation with Configurations and Integration of Data, Learn Structure, Theory and Model for Simulation, and Learn to make Surrogates.},
 bibtype = {inproceedings},
 author = {Fox, Geoffrey and Jha, Shantenu},
 doi = {10.1109/escience.2019.00057}
}

We present a taxonomy of research on Machine Learning (ML) applied to enhance simulations together with a catalog of some activities. We cover eight patterns for the link of ML to the simulations or systems plus three algorithmic areas: particle dynamics, agent-based models and partial differential equations. The patterns are further divided into three action areas: Improving simulation with Configurations and Integration of Data, Learn Structure, Theory and Model for Simulation, and Learn to make Surrogates.

@techreport{
 title = {Learning Everywhere Resource for BDEC},
 type = {techreport},
 year = {2019},
 id = {7a76b2e9-6fb2-323d-8aef-bdae4a7d5f8d},
 created = {2019-08-14T17:21:05.823Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-14T17:21:05.823Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 bibtype = {techreport},
 author = {Fox, Geoffrey and July, Updated}
}

@inproceedings{
 title = {Big Data Benchmarks of High-Performance Storage Systems on Commercial Bare Metal Clouds},
 type = {inproceedings},
 year = {2019},
 websites = {https://conferences.computer.org/serviceswp/2019/pdfs/CLOUD2019-5XaiJ82Ya5AyIoh2T1E5Bm/6ex2QOZUKAReRoHl2O8jiP/5Fkht00rYCKTzblzYTjaNc.pdf},
 month = {5},
 day = {27},
 id = {d6d07a06-6477-3069-bf2d-97b8a86748dd},
 created = {2019-08-21T18:14:53.512Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T18:16:39.646Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Bare metal servers are widely available on public clouds to provide direct access to hardware and the system configuration with high performance storage and network devices are well suited for big data applications. Highly-optimized server with additional CPU core count and dense storage may lead to better performance in certain workloads and to ensure responsiveness of deployed services. Recent work on Hadoop ecosystems has addressed the performance improvement of scale-up machines configured with SSD storage and increased network bandwidth. The paper evaluates big data processing on dedicated clusters and provides the performance analysis of NVMe devices and SSD block storage options available on Amazon, Google, Microsoft, and Oracle Clouds. We show the benchmark results along with the system performance tests as we want to demonstrate the compute resource requirements for large-scale applications. The system capacity and limits for the underlying servers are described along with the cost analysis of scaling workloads on these platforms.},
 bibtype = {inproceedings},
 author = {Lee, Hyungro and Fox, Geoffrey},
 doi = {10.1109/CLOUD.2019.00014},
 booktitle = {IEEE International Conference on Cloud Computing (IEEE CLOUD 2019)}
}

Bare metal servers are widely available on public clouds to provide direct access to hardware and the system configuration with high performance storage and network devices are well suited for big data applications. Highly-optimized server with additional CPU core count and dense storage may lead to better performance in certain workloads and to ensure responsiveness of deployed services. Recent work on Hadoop ecosystems has addressed the performance improvement of scale-up machines configured with SSD storage and increased network bandwidth. The paper evaluates big data processing on dedicated clusters and provides the performance analysis of NVMe devices and SSD block storage options available on Amazon, Google, Microsoft, and Oracle Clouds. We show the benchmark results along with the system performance tests as we want to demonstrate the compute resource requirements for large-scale applications. The system capacity and limits for the underlying servers are described along with the cost analysis of scaling workloads on these platforms.

@inproceedings{
 title = {Perspectives on High-Performance Computing in a Big Data World},
 type = {inproceedings},
 year = {2019},
 pages = {145-145},
 publisher = {Association for Computing Machinery (ACM)},
 id = {a2bd0311-1c81-34a9-a379-0cabd2eb0ea1},
 created = {2019-08-21T18:21:30.779Z},
 accessed = {2019-08-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T18:29:44.793Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {High-Performance Computing (HPC) and Cyberinfrastructure have played a leadership role in computational science even since the start of the NSF computing centers program. Thirty years ago parallel computing was a centerpiece of computer science research. Naively Big Data surely requires HPC to be processed, and transformational Big Data technology such as Hadoop and Spark exploit parallelism to success. Nevertheless, the HPC community does not appear to be thriving as a leader in Data Science while parallel computing is no longer a centerpiece. Some reasons for this are the dominant presence of Industry in technology futures and the universal fascination with Artificial Intelligence and Machine Learning. Maybe the pendulum will swing back a bit, but I expect the "AI first" philosophy to dominate in the foreseeable future. Thus I describe a future where HPC thrives in collaboration with Industry and AI. In particular, I discuss the promise of MLforHPC (AI for systems) and HPCforML (systems for AI).},
 bibtype = {inproceedings},
 author = {Fox, Geoffrey C.},
 doi = {10.1145/3307681.3325410},
 booktitle = {Proceedings of the 28th International Symposium on High-Performance Parallel and Distributed Computing (HPDC '19)}
}

High-Performance Computing (HPC) and Cyberinfrastructure have played a leadership role in computational science even since the start of the NSF computing centers program. Thirty years ago parallel computing was a centerpiece of computer science research. Naively Big Data surely requires HPC to be processed, and transformational Big Data technology such as Hadoop and Spark exploit parallelism to success. Nevertheless, the HPC community does not appear to be thriving as a leader in Data Science while parallel computing is no longer a centerpiece. Some reasons for this are the dominant presence of Industry in technology futures and the universal fascination with Artificial Intelligence and Machine Learning. Maybe the pendulum will swing back a bit, but I expect the "AI first" philosophy to dominate in the foreseeable future. Thus I describe a future where HPC thrives in collaboration with Industry and AI. In particular, I discuss the promise of MLforHPC (AI for systems) and HPCforML (systems for AI).

@inproceedings{
 title = {Twister2: TSet high-performance iterative dataflow},
 type = {inproceedings},
 year = {2019},
 keywords = {Batch,Big data,Dataflow,Iterative,Mapreduce,Parallel programming,Stream},
 pages = {55-60},
 month = {5},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {1},
 id = {001d38a7-43fd-3f24-aebd-c65f989e0cfc},
 created = {2019-08-21T18:30:51.173Z},
 accessed = {2019-08-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T18:30:51.263Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {The dataflow model is slowly becoming the de facto standard for big data applications. While many popular frameworks are built around the dataflow model, very little research has been done on understanding the inner workings of the dataflow model. This has led to many inefficiencies in existing frameworks. It is important to note that understanding the relation between dataflow and HPC building blocks allows us to address and alleviate many of the fundamental inefficiencies in dataflow by learning from the extensive research literature in the HPC community. In this paper, we present TSet's, the dataflow abstraction of Twister2, which is a big data framework designed for high-performance dataflow and iterative computations. We discuss the dataflow model adopted by TSet's and the rationale behind implementing iteration handling at the worker level. Finally, we evaluate TSet's to show the performance of the framework.},
 bibtype = {inproceedings},
 author = {Wickramasinghe, Pulasthi and Kamburugamuve, Supun and Govindarajan, Kannan and Abeykoon, Vibhatha and Widanage, Chathura and Perera, Niranda and Uyar, Ahmet and Gunduz, Gurhan and Akkas, Selahattin and Fox, Geoffrey},
 doi = {10.1109/HPBDIS.2019.8735495},
 booktitle = {2019 International Conference on High Performance Big Data and Intelligent Systems, HPBD and IS 2019}
}

The dataflow model is slowly becoming the de facto standard for big data applications. While many popular frameworks are built around the dataflow model, very little research has been done on understanding the inner workings of the dataflow model. This has led to many inefficiencies in existing frameworks. It is important to note that understanding the relation between dataflow and HPC building blocks allows us to address and alleviate many of the fundamental inefficiencies in dataflow by learning from the extensive research literature in the HPC community. In this paper, we present TSet's, the dataflow abstraction of Twister2, which is a big data framework designed for high-performance dataflow and iterative computations. We discuss the dataflow model adopted by TSet's and the rationale behind implementing iteration handling at the worker level. Finally, we evaluate TSet's to show the performance of the framework.

@techreport{
 title = {Parallel Performance of Molecular Dynamics Trajectory Analysis},
 type = {techreport},
 year = {2019},
 keywords = {Big Data,Global Arrays,HDF5,HPC,MDAnalysis,MPI,MPI I/O,Molecular Dynamics,Python,Straggler,Trajectory Analysis},
 pages = {1-60},
 id = {dcf9957b-d904-3177-9e2c-1017d45f8ae4},
 created = {2019-08-21T18:57:48.492Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T18:57:48.492Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {The performance of biomolecular molecular dynamics (MD) simulations has steadily increased on modern high performance computing (HPC) resources but acceleration of the analysis of the output trajectories has lagged behind so that analyzing simulations is increasingly becoming a bottleneck. To close this gap, we studied the performance of parallel trajectory analysis with MPI and the Python MDAnalysis library on three different XSEDE supercomputers where trajectories were read from a Lustre parallel file system. We found that strong scaling performance was impeded by stragglers, MPI processes that were slower than the typical process and that therefore dominated the overall run time. Stragglers were less prevalent for compute-bound workloads, thus pointing to file reading as a crucial bottleneck for scaling. However, a more complicated picture emerged in which both the computation and the ingestion of data exhibited close to ideal strong scaling behavior whereas stragglers were primarily caused by either large MPI communication costs or long times to open the single shared trajectory file. We improved overall strong scaling performance by two different approaches to file access, namely subfiling (splitting the trajectory into as many trajectory segments as number of processes) and MPI-IO with Parallel HDF5 trajectory files. Applying these strategies, we obtained near ideal strong scaling on up to 384 cores (16 nodes). We summarize our lessons-learned in guidelines and strategies on how to take advantage of the available HPC resources to gain good scalability and potentially reduce trajectory analysis times by two orders of magnitude compared to the prevalent serial approach.},
 bibtype = {techreport},
 author = {Khoshlessan, Mahzad and Paraskevakos, Ioannis and Fox, Geoffrey C and Jha, Shantenu and Beckstein, Oliver}
}

The performance of biomolecular molecular dynamics (MD) simulations has steadily increased on modern high performance computing (HPC) resources but acceleration of the analysis of the output trajectories has lagged behind so that analyzing simulations is increasingly becoming a bottleneck. To close this gap, we studied the performance of parallel trajectory analysis with MPI and the Python MDAnalysis library on three different XSEDE supercomputers where trajectories were read from a Lustre parallel file system. We found that strong scaling performance was impeded by stragglers, MPI processes that were slower than the typical process and that therefore dominated the overall run time. Stragglers were less prevalent for compute-bound workloads, thus pointing to file reading as a crucial bottleneck for scaling. However, a more complicated picture emerged in which both the computation and the ingestion of data exhibited close to ideal strong scaling behavior whereas stragglers were primarily caused by either large MPI communication costs or long times to open the single shared trajectory file. We improved overall strong scaling performance by two different approaches to file access, namely subfiling (splitting the trajectory into as many trajectory segments as number of processes) and MPI-IO with Parallel HDF5 trajectory files. Applying these strategies, we obtained near ideal strong scaling on up to 384 cores (16 nodes). We summarize our lessons-learned in guidelines and strategies on how to take advantage of the available HPC resources to gain good scalability and potentially reduce trajectory analysis times by two orders of magnitude compared to the prevalent serial approach.

@techreport{
 title = {Harnessing the Computing Continuum for Programming Our World},
 type = {techreport},
 year = {2019},
 pages = {1-11},
 websites = {https://www.researchgate.net/publication/332246123_Harnessing_the_Computing_Continuum_for_Programming_Our_World},
 id = {1c720f5a-1e2c-368c-9b9e-d65e5ee5b9d8},
 created = {2019-08-21T18:58:52.171Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T19:05:56.392Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {This paper outlines a vision for how best to harness the continuum of interconnected sensors, actuators, instruments, and computing systems, from small numbers of very large devices to large numbers of very small devices. Our hypothesis is that only via a continuum perspective can we intentionally specify desired continuum actions and effectively manage outcomes and systemic properties—adaptability and homeostasis, temporal constraints and deadlines—and elevate the discourse from device programming to intellectual goals and outcomes.},
 bibtype = {techreport},
 author = {Beckman, Pete and Dongarra, Jack and Ferrier, Nicola and Fox, Geoffrey and Moore, Terry and Reed, Dan and Beck, Micah}
}

This paper outlines a vision for how best to harness the continuum of interconnected sensors, actuators, instruments, and computing systems, from small numbers of very large devices to large numbers of very small devices. Our hypothesis is that only via a continuum perspective can we intentionally specify desired continuum actions and effectively manage outcomes and systemic properties—adaptability and homeostasis, temporal constraints and deadlines—and elevate the discourse from device programming to intellectual goals and outcomes.

@misc{
 title = {Advances in big data programming, system software and HPC convergence},
 type = {misc},
 year = {2019},
 source = {Journal of Supercomputing},
 pages = {489-493},
 volume = {75},
 issue = {2},
 month = {2},
 publisher = {Springer New York LLC},
 day = {6},
 id = {7d3d61c2-968b-38bb-8a7b-95ffd9731ac1},
 created = {2019-08-21T19:10:09.489Z},
 accessed = {2019-08-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T19:10:09.489Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {In the era of big data, developing modern computing systems and system software that can scale to massive amounts of data becomes a key challenge to both researchers and practitioners. Scalability in distributed system usually means that the performance of a system should increase proportionally with the increase in resources. However, this is not sufficient in the big data era. The system should be designed in a way so that all the five Vs of big data can be tackled. Driven by this insight, this special issue aims at presenting the current state-of-the-art research and future trends on various aspects of big data programming and system software techniques for big data processing and attempts towards building highly adaptive big data systems that can automatically adapt their behaviours to the amount of available resources. The major subjects cover methodologies, modelling, analysis and newly introduced applications. Besides the latest research achievements, this special issue also covers innovative commercial data management systems, innovative commercial applications of big data technology and experience in applying recent research advances to real-world problems.},
 bibtype = {misc},
 author = {Hsu, Ching Hsien and Fox, Geoffrey and Min, Geyong and Sharma, Sugam},
 doi = {10.1007/s11227-018-2706-x}
}

In the era of big data, developing modern computing systems and system software that can scale to massive amounts of data becomes a key challenge to both researchers and practitioners. Scalability in distributed system usually means that the performance of a system should increase proportionally with the increase in resources. However, this is not sufficient in the big data era. The system should be designed in a way so that all the five Vs of big data can be tackled. Driven by this insight, this special issue aims at presenting the current state-of-the-art research and future trends on various aspects of big data programming and system software techniques for big data processing and attempts towards building highly adaptive big data systems that can automatically adapt their behaviours to the amount of available resources. The major subjects cover methodologies, modelling, analysis and newly introduced applications. Besides the latest research achievements, this special issue also covers innovative commercial data management systems, innovative commercial applications of big data technology and experience in applying recent research advances to real-world problems.

@inproceedings{
 title = {Learning Everywhere: Pervasive Machine Learning for Effective High-Performance Computation},
 type = {inproceedings},
 year = {2019},
 pages = {422-429},
 websites = {https://ieeexplore.ieee.org/document/8778333/},
 month = {5},
 publisher = {IEEE},
 id = {d9c3a14b-34ab-36f4-8a25-47a6e3e5c968},
 created = {2019-08-21T19:29:39.227Z},
 accessed = {2019-08-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T19:29:39.227Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {The convergence of HPC and data-intensive methodologies provide a promising approach to major performance improvements. This paper provides a general description of the interaction between traditional HPC and ML approaches and motivates the Learning Everywhere paradigm for HPC. We introduce the concept of effective performance that one can achieve by combining learning methodologies with simulation-based approaches, and distinguish between traditional performance as measured by benchmark scores. To support the promise of integrating HPC and learning methods, this paper examines specific examples and opportunities across a series of domains. It concludes with a series of open computer science and cyberinfrastructure questions and challenges that the Learning Everywhere paradigm presents.},
 bibtype = {inproceedings},
 author = {Fox, Geoffrey and Glazier, James and Kadupitiya, JCS and Jadhao, Vikram and Kim, Minje and Qiu, Judy and Sluka, James P. and Somogy, Endre and Marathe, Madhav and Adiga, Abhijin and Chen, Jiangzhuo and Beckstein, Oliver and Jha, Shantenu},
 doi = {10.1109/IPDPSW.2019.00081},
 booktitle = {2019 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)}
}

The convergence of HPC and data-intensive methodologies provide a promising approach to major performance improvements. This paper provides a general description of the interaction between traditional HPC and ML approaches and motivates the Learning Everywhere paradigm for HPC. We introduce the concept of effective performance that one can achieve by combining learning methodologies with simulation-based approaches, and distinguish between traditional performance as measured by benchmark scores. To support the promise of integrating HPC and learning methods, this paper examines specific examples and opportunities across a series of domains. It concludes with a series of open computer science and cyberinfrastructure questions and challenges that the Learning Everywhere paradigm presents.

@article{
 title = {Twister2: Design of a big data toolkit},
 type = {article},
 year = {2019},
 keywords = {big data,dataflow,event-driven computing,high performance computing},
 publisher = {John Wiley and Sons Ltd},
 id = {64c4c5a8-155d-3bae-8849-37cdde5f8a0f},
 created = {2019-08-21T20:01:29.562Z},
 accessed = {2019-08-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T20:01:29.668Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Summary Data-driven applications are essential to handle the ever-increasing volume, velocity, and veracity of data generated by sources such as the Web and Internet of Things (IoT) devices. Simultaneously, an event-driven computational paradigm is emerging as the core of modern systems designed for database queries, data analytics, and on-demand applications. Modern big data processing runtimes and asynchronous many task (AMT) systems from high performance computing (HPC) community have adopted dataflow event-driven model. The services are increasingly moving to an event-driven model in the form of Function as a Service (FaaS) to compose services. An event-driven runtime designed for data processing consists of well-understood components such as communication, scheduling, and fault tolerance. Different design choices adopted by these components determine the type of applications a system can support efficiently. We find that modern systems are limited to specific sets of applications because they have been designed with fixed choices that cannot be changed easily. In this paper, we present a loosely coupled component-based design of a big data toolkit where each component can have different implementations to support various applications. Such a polymorphic design would allow services and data analytics to be integrated seamlessly and expand from edge to cloud to HPC environments.},
 bibtype = {article},
 author = {Kamburugamuve, Supun and Govindarajan, Kannan and Wickramasinghe, Pulasthi and Abeykoon, Vibhatha and Fox, Geoffrey},
 doi = {10.1002/cpe.5189},
 journal = {Concurrency Computation: Practice and Experience}
}

Summary Data-driven applications are essential to handle the ever-increasing volume, velocity, and veracity of data generated by sources such as the Web and Internet of Things (IoT) devices. Simultaneously, an event-driven computational paradigm is emerging as the core of modern systems designed for database queries, data analytics, and on-demand applications. Modern big data processing runtimes and asynchronous many task (AMT) systems from high performance computing (HPC) community have adopted dataflow event-driven model. The services are increasingly moving to an event-driven model in the form of Function as a Service (FaaS) to compose services. An event-driven runtime designed for data processing consists of well-understood components such as communication, scheduling, and fault tolerance. Different design choices adopted by these components determine the type of applications a system can support efficiently. We find that modern systems are limited to specific sets of applications because they have been designed with fixed choices that cannot be changed easily. In this paper, we present a loosely coupled component-based design of a big data toolkit where each component can have different implementations to support various applications. Such a polymorphic design would allow services and data analytics to be integrated seamlessly and expand from edge to cloud to HPC environments.

@inproceedings{
 title = {Machine Learning for Performance Enhancement of Molecular Dynamics Simulations},
 type = {inproceedings},
 year = {2019},
 keywords = {Clouds,Machine learning,Molecular dynamics simulations,Parallel computing,Scientific computing},
 pages = {116-130},
 volume = {11537 LNCS},
 publisher = {Springer Verlag},
 id = {cf9b9da5-2647-3931-bb95-24275cab86b4},
 created = {2019-08-21T20:08:23.914Z},
 accessed = {2019-08-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T20:08:23.981Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We explore the idea of integrating machine learning with simulations to enhance the performance of the simulation and improve its usability for research and education. The idea is illustrated using hybrid OpenMP/MPI parallelized molecular dynamics simulations designed to extract the distribution of ions in nanoconfinement. We find that an artificial neural network based regression model successfully learns the desired features associated with the output ionic density profiles and rapidly generates predictions that are in excellent agreement with the results from explicit molecular dynamics simulations. The results demonstrate that the performance gains of parallel computing can be further enhanced by using machine learning.},
 bibtype = {inproceedings},
 author = {Kadupitiya, Jcs and Fox, Geoffrey C. and Jadhao, Vikram},
 doi = {10.1007/978-3-030-22741-8_9},
 booktitle = {International Conference on Computational Science ICCS2019}
}

We explore the idea of integrating machine learning with simulations to enhance the performance of the simulation and improve its usability for research and education. The idea is illustrated using hybrid OpenMP/MPI parallelized molecular dynamics simulations designed to extract the distribution of ions in nanoconfinement. We find that an artificial neural network based regression model successfully learns the desired features associated with the output ionic density profiles and rapidly generates predictions that are in excellent agreement with the results from explicit molecular dynamics simulations. The results demonstrate that the performance gains of parallel computing can be further enhanced by using machine learning.

@article{
 title = {Object Classification by a Super-resolution Method and a Convolutional Neural Networks},
 type = {article},
 year = {2019},
 keywords = {CNN,convolution neural networks,deep learning,machine learning,object detection,super-resolution},
 pages = {16-23},
 volume = {1},
 websites = {http://ijdat.org/index.php/ijdat/article/view/4/4},
 id = {974f9e7c-011b-3eb5-8d6d-4f3621099260},
 created = {2019-08-21T20:16:58.488Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-10-01T17:21:32.254Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Recently with many blurless or slightly blurred images, convolutional neural networks classify objects with around 90 percent classification rates, even if there are variable sized images. However, small object regions or cropping of images make object detection or classification difficult and decreases the detection rates. In many methods related to convolutional neural network (CNN), Bilinear or Bicubic algorithms are popularly used to interpolate region of interests. To overcome the limitations of these algorithms, we introduce a super-resolution method applied to the cropped regions or candidates, and this leads to improve recognition rates for object detection and classification. Large object candidates comparable in size of the full image have good results for object detections using many popular conventional methods. However, for smaller region candidates, using our super-resolution preprocessing and region candidates, allows a CNN to outperform conventional methods in the number of detected objects when tested on the VOC2007 and MSO datasets},
 bibtype = {article},
 author = {Na, Bokyoon and Fox, Geoffrey C},
 journal = {International Journal of Data Mining Science},
 number = {1}
}

Recently with many blurless or slightly blurred images, convolutional neural networks classify objects with around 90 percent classification rates, even if there are variable sized images. However, small object regions or cropping of images make object detection or classification difficult and decreases the detection rates. In many methods related to convolutional neural network (CNN), Bilinear or Bicubic algorithms are popularly used to interpolate region of interests. To overcome the limitations of these algorithms, we introduce a super-resolution method applied to the cropped regions or candidates, and this leads to improve recognition rates for object detection and classification. Large object candidates comparable in size of the full image have good results for object detections using many popular conventional methods. However, for smaller region candidates, using our super-resolution preprocessing and region candidates, allows a CNN to outperform conventional methods in the number of detected objects when tested on the VOC2007 and MSO datasets

@inproceedings{
 title = {Machine Learning for Performance Enhancement of Molecular Dynamics Simulations},
 type = {inproceedings},
 year = {2019},
 keywords = {Clouds,Machine learning,Molecular dynamics simulations,Parallel computing,Scientific computing},
 pages = {116-130},
 volume = {11537 LNCS},
 publisher = {Springer Verlag},
 id = {6a70ef6d-7f55-35cd-873f-97b71e594fd8},
 created = {2019-08-26T20:49:05.888Z},
 accessed = {2019-08-26},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-26T20:49:05.982Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Kadupitiya, Jcs and Fox, Geoffrey C. and Jadhao, Vikram},
 doi = {10.1007/978-3-030-22741-8_9},
 booktitle = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}
}

@techreport{
 title = {Learning Everywhere: Pervasive Machine Learning for Effective High-Performance Computation: Application Background},
 type = {techreport},
 year = {2019},
 pages = {33},
 websites = {http://dsc.soic.indiana.edu/publications/Learning_Everywhere.pdf},
 id = {531c1252-3443-3fc9-ab45-fc5cfa7f08ec},
 created = {2019-08-26T21:08:23.230Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-26T21:08:23.230Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 bibtype = {techreport},
 author = {Fox, Geoffrey and Glazier, James A and Kadupitiya, Jcs and Jadhao, Vikram and Kim, Minje and Qiu, Judy and Sluka, James P and Somogyi, Endre and Marathe, Madhav and Adiga, Abhijin and Chen, Jiangzhuo and Beckstein, Oliver and Jha, Shantenu}
}

@article{
 title = {Regge phenomenology of the N∗ and Δ∗ poles},
 type = {article},
 year = {2019},
 volume = {99},
 month = {2},
 publisher = {American Physical Society},
 day = {1},
 id = {ba6dec59-c388-3acd-98a1-78de31d5d5c9},
 created = {2019-09-03T19:49:53.174Z},
 accessed = {2019-09-03},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T19:49:53.284Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We use Regge phenomenology to study the structure of the poles of the $N^*$ and $\mathrmΔ^*$ spectrum. We employ the available pole extractions from partial wave analysis of meson scattering and photoproduction data. We assess the importance of the imaginary part of the poles (widths) to o...},
 bibtype = {article},
 author = {Silva-Castro, J. A. and Fernández-Ramírez, C. and Albaladejo, M. and Danilkin, I. V. and Jackura, A. and Mathieu, V. and Nys, J. and Pilloni, A. and Szczepaniak, A. P. and Fox, G.},
 doi = {10.1103/PhysRevD.99.034003},
 journal = {Physical Review D},
 number = {3}
}

We use Regge phenomenology to study the structure of the poles of the $N^*$ and $\mathrmΔ^*$ spectrum. We employ the available pole extractions from partial wave analysis of meson scattering and photoproduction data. We assess the importance of the imaginary part of the poles (widths) to o...

@inproceedings{
 title = {Benchmarking Deep Learning for Time Series: Challenges and Directions},
 type = {inproceedings},
 year = {2019},
 keywords = {benchmark,deep learning,machine learning,performance,time series},
 pages = {5679-5682},
 month = {12},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {1},
 id = {4fe41ff0-884c-3804-bb94-5d55c61eba5b},
 created = {2020-04-21T18:42:11.807Z},
 accessed = {2020-04-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T18:42:11.905Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Deep learning for time series is an emerging area with close ties to industry, yet under represented in performance benchmarks for machine learning systems. In this paper, we present a landscape of deep learning applications applied to time series, and discuss the challenges and directions towards building a robust performance benchmark of deep learning workloads for time series data.},
 bibtype = {inproceedings},
 author = {Huang, Xinyuan and Fox, Geoffrey C. and Serebryakov, Sergey and Mohan, Ankur and Morkisz, Pawel and Dutta, Debojyoti},
 doi = {10.1109/BigData47090.2019.9005496},
 booktitle = {Proceedings - 2019 IEEE International Conference on Big Data, Big Data 2019}
}

Deep learning for time series is an emerging area with close ties to industry, yet under represented in performance benchmarks for machine learning systems. In this paper, we present a landscape of deep learning applications applied to time series, and discuss the challenges and directions towards building a robust performance benchmark of deep learning workloads for time series data.

@inproceedings{
 title = {Streaming Machine Learning Algorithms with Big Data Systems},
 type = {inproceedings},
 year = {2019},
 keywords = {Big-Data,Dataflow,Streaming Machine Learning},
 pages = {5661-5666},
 month = {12},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {1},
 id = {c88b536b-6dc7-39da-9c90-de4080c21cfa},
 created = {2020-04-21T19:04:55.439Z},
 accessed = {2020-04-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T19:04:55.509Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Designing low latency applications that can process large volumes data with higher efficiency is a challenging problem. With the limited time to process data, usage of online algorithms are becoming important in the big-data applications. Stream processing is a well-known area that has been studied for a long time. In this research, our objective is to use state of the art big-data analytic engines to implement online algorithms and compare the strengths and weaknesses in each system. We use a streaming version of Support Vector Machines (SVM) and KMeans to do the analysis. Apache Flink, Apache Storm and Twister2 streaming frameworks are used to implement these algorithms. Our study focuses on the efficiency of online training of these algorithms and the results show higher performance in Twister2 framework for these algorithms.},
 bibtype = {inproceedings},
 author = {Abeykoon, Vibhatha and Laszewski, Gregor Von and Kamburugamuve, Supun and Govindrarajan, Kannan and Wickramasinghe, Pulasthi and Widanage, Chathura and Perera, Niranda and Uyar, Ahmet and Gunduz, Gurhan and Akkas, Selahattin},
 doi = {10.1109/BigData47090.2019.9006337},
 booktitle = {Proceedings - 2019 IEEE International Conference on Big Data, Big Data 2019}
}

Designing low latency applications that can process large volumes data with higher efficiency is a challenging problem. With the limited time to process data, usage of online algorithms are becoming important in the big-data applications. Stream processing is a well-known area that has been studied for a long time. In this research, our objective is to use state of the art big-data analytic engines to implement online algorithms and compare the strengths and weaknesses in each system. We use a streaming version of Support Vector Machines (SVM) and KMeans to do the analysis. Apache Flink, Apache Storm and Twister2 streaming frameworks are used to implement these algorithms. Our study focuses on the efficiency of online training of these algorithms and the results show higher performance in Twister2 framework for these algorithms.

@article{
 title = {Automated Ice-Bottom Tracking of 2D and 3D Ice Radar Imagery Using Viterbi and TRW-S},
 type = {article},
 year = {2019},
 keywords = {Feature extraction,glaciology,ice thickness,ice tracking,radar tomography},
 pages = {3272-3285},
 volume = {12},
 month = {9},
 publisher = {Institute of Electrical and Electronics Engineers},
 day = {1},
 id = {8d850227-ba1c-3de9-8d23-7ae02ba2c1e6},
 created = {2020-04-21T19:09:21.573Z},
 accessed = {2020-04-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T19:09:21.658Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Multichannel radar depth sounding systems are able to produce two-dimensional (2D) and three-dimensional (3D) imagery of the internal structure of polar ice sheets. Information such as ice thickness and surface elevation is extracted from these data and applied to research in ice flow modeling and ice mass balance calculations. Due to a large amount of data collected, we seek to automate the ice-bottom layer tracking and allow for efficient manual corrections when errors occur in the automated method. We present improvements made to previous implementations of the Viterbi and sequential tree-reweighted message passing (TRW-S) algorithms for ice-bottom extraction in 2D and 3D radar imagery. These improvements are in the form of novel cost functions that allow for the integration of further domain-specific knowledge into the cost calculations and provide additional evidence of the characteristics of the ice sheets surveyed. Along with an explanation of our modifications, we demonstrate the results obtained by our modified implementations of the two algorithms and by previously proposed solutions to this problem, when compared to manually corrected ground truth data. Furthermore, we perform a self-assessment of tracking results by analyzing differences in the estimated ice-bottom for surveyed locations where flight paths have crossed and, thus, two separate measurements have been made at the same location. Using our modified cost functions and preprocessing routines, we obtain significantly decreased mean error measurements from both algorithms, such as a 47% reduction in average tracking error in the case of 3D imagery between the original and our proposed implementation of TRW-S.},
 bibtype = {article},
 author = {Berger, Victor and Xu, Mingze and Al-Ibadi, Mohanad and Chu, Shane and Crandall, David and Paden, John and Fox, Geoffrey Charles},
 doi = {10.1109/JSTARS.2019.2930920},
 journal = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
 number = {9}
}

Multichannel radar depth sounding systems are able to produce two-dimensional (2D) and three-dimensional (3D) imagery of the internal structure of polar ice sheets. Information such as ice thickness and surface elevation is extracted from these data and applied to research in ice flow modeling and ice mass balance calculations. Due to a large amount of data collected, we seek to automate the ice-bottom layer tracking and allow for efficient manual corrections when errors occur in the automated method. We present improvements made to previous implementations of the Viterbi and sequential tree-reweighted message passing (TRW-S) algorithms for ice-bottom extraction in 2D and 3D radar imagery. These improvements are in the form of novel cost functions that allow for the integration of further domain-specific knowledge into the cost calculations and provide additional evidence of the characteristics of the ice sheets surveyed. Along with an explanation of our modifications, we demonstrate the results obtained by our modified implementations of the two algorithms and by previously proposed solutions to this problem, when compared to manually corrected ground truth data. Furthermore, we perform a self-assessment of tracking results by analyzing differences in the estimated ice-bottom for surveyed locations where flight paths have crossed and, thus, two separate measurements have been made at the same location. Using our modified cost functions and preprocessing routines, we obtain significantly decreased mean error measurements from both algorithms, such as a 47% reduction in average tracking error in the case of 3D imagery between the original and our proposed implementation of TRW-S.

@inproceedings{
 title = {Scientific image restoration anywhere},
 type = {inproceedings},
 year = {2019},
 keywords = {Deep learning,Edge computing,Image restoration,Model quantization},
 pages = {8-13},
 websites = {http://arxiv.org/abs/1911.05878},
 month = {11},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {1},
 id = {abd3b439-80ed-35cf-ab00-a2d1b63e64c4},
 created = {2020-04-21T19:19:33.486Z},
 accessed = {2020-04-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-04-21T19:19:33.589Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {The use of deep learning models within scientific experimental facilities frequently requires low-latency inference, so that, for example, quality control operations can be performed while data are being collected. Edge computing devices can be useful in this context, as their low cost and compact form factor permit them to be co-located with the experimental apparatus. Can such devices, with their limited resources, can perform neural network feed-forward computations efficiently and effectively? We explore this question by evaluating the performance and accuracy of a scientific image restoration model, for which both model input and output are images, on edge computing devices. Specifically, we evaluate deployments of TomoGAN, an image-denoising model based on generative adversarial networks developed for low-dose x-ray imaging, on the Google Edge TPU and NVIDIA Jetson. We adapt TomoGAN for edge execution, evaluate model inference performance, and propose methods to address the accuracy drop caused by model quantization. We show that these edge computing devices can deliver accuracy comparable to that of a full-fledged CPU or GPU model, at speeds that are more than adequate for use in the intended deployments, denoising a 1024x1024 image in less than a second. Our experiments also show that the Edge TPU models can provide 3x faster inference response than a CPU-based model and 1.5x faster than an edge GPU-based model. This combination of high speed and low cost permits image restoration anywhere.},
 bibtype = {inproceedings},
 author = {Abeykoon, Vibhatha and Liu, Zhengchun and Kettimuthu, Rajkumar and Fox, Geoffrey and Foster, Ian},
 doi = {10.1109/XLOOP49562.2019.00007},
 booktitle = {Proceedings of XLOOP 2019: 1st Annual Workshop on Large-Scale Experiment-in-the-Loop Computing, Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis}
}

The use of deep learning models within scientific experimental facilities frequently requires low-latency inference, so that, for example, quality control operations can be performed while data are being collected. Edge computing devices can be useful in this context, as their low cost and compact form factor permit them to be co-located with the experimental apparatus. Can such devices, with their limited resources, can perform neural network feed-forward computations efficiently and effectively? We explore this question by evaluating the performance and accuracy of a scientific image restoration model, for which both model input and output are images, on edge computing devices. Specifically, we evaluate deployments of TomoGAN, an image-denoising model based on generative adversarial networks developed for low-dose x-ray imaging, on the Google Edge TPU and NVIDIA Jetson. We adapt TomoGAN for edge execution, evaluate model inference performance, and propose methods to address the accuracy drop caused by model quantization. We show that these edge computing devices can deliver accuracy comparable to that of a full-fledged CPU or GPU model, at speeds that are more than adequate for use in the intended deployments, denoising a 1024x1024 image in less than a second. Our experiments also show that the Edge TPU models can provide 3x faster inference response than a CPU-based model and 1.5x faster than an edge GPU-based model. This combination of high speed and low cost permits image restoration anywhere.

@article{
 title = {Vector meson photoproduction with a linearly polarized beam},
 type = {article},
 year = {2018},
 volume = {97},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049067436&doi=10.1103%2FPhysRevD.97.094003&partnerID=40&md5=c3e8c7a8d17dcaf2f3e6e356802ef69c},
 publisher = {American Physical Society},
 id = {8369fc0f-8b35-302f-bf0a-6c1e37cb92e4},
 created = {2018-07-12T19:57:02.442Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-07-12T19:57:02.442Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Mathieu2018},
 source_type = {article},
 notes = {cited By 0},
 private_publication = {false},
 abstract = {We propose a model based on Regge theory to describe photoproduction of light vector mesons. We fit the SLAC data and make predictions for the energy and momentum-transfer dependence of the spin-density matrix elements in photoproduction of ω, ρ0 and φ mesons at Eγ∼8.5 GeV, which are soon to be measured at Jefferson Lab. © 2018 authors. Published by the American Physical Society. Published by the American Physical Society under the terms of the »https://creativecommons.org/licenses/by/4.0/» Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI. Funded by SCOAP3.},
 bibtype = {article},
 author = {Mathieu, V and Nys, J and Fernández-Ramírez, C and Jackura, A and Pilloni, A and Sherrill, N and Szczepaniak, A P and Fox, G},
 doi = {10.1103/PhysRevD.97.094003},
 journal = {Physical Review D},
 number = {9}
}

We propose a model based on Regge theory to describe photoproduction of light vector mesons. We fit the SLAC data and make predictions for the energy and momentum-transfer dependence of the spin-density matrix elements in photoproduction of ω, ρ0 and φ mesons at Eγ∼8.5 GeV, which are soon to be measured at Jefferson Lab. © 2018 authors. Published by the American Physical Society. Published by the American Physical Society under the terms of the »https://creativecommons.org/licenses/by/4.0/» Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI. Funded by SCOAP3.

@article{
 title = {Analyticity Constraints for Hadron Amplitudes: Going High to Heal Low Energy Issues},
 type = {article},
 year = {2018},
 pages = {41001-p1-p5},
 volume = {122},
 websites = {http://arxiv.org/abs/1708.07779},
 id = {cf78f2a9-f08b-3e09-97ee-952854251b45},
 created = {2018-08-09T16:38:13.973Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T15:22:45.991Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Analyticity constitutes a rigid constraint on hadron scattering amplitudes. This property is used to relate models in different energy regimes. Using meson photoproduction as a benchmark, we show how to test contemporary low energy models directly against high energy data. This method pinpoints deficiencies of the models and treads a path to further improvement. The implementation of this technique enables one to produce more stable and reliable partial waves for future use in hadron spectroscopy and new physics searches.},
 bibtype = {article},
 author = {Mathieu, V. and Nys, J. and Pilloni, A. and Fernández-Ramírez, C. and Jackura, A. and Mikhasenko, M. and Pauk, V. and Szczepaniak, A. P. and Fox, G.},
 journal = {Europhysics Letters},
 number = {4}
}

Analyticity constitutes a rigid constraint on hadron scattering amplitudes. This property is used to relate models in different energy regimes. Using meson photoproduction as a benchmark, we show how to test contemporary low energy models directly against high energy data. This method pinpoints deficiencies of the models and treads a path to further improvement. The implementation of this technique enables one to produce more stable and reliable partial waves for future use in hadron spectroscopy and new physics searches.

@article{
 title = {Structure of pion photoproduction amplitudes},
 type = {article},
 year = {2018},
 keywords = {doi:10.1103/PhysRevD.98.014041 url:https://doi.org},
 pages = {014041},
 volume = {98},
 websites = {https://link.aps.org/doi/10.1103/PhysRevD.98.014041},
 publisher = {American Physical Society},
 id = {02e8a170-59df-32b7-8742-df159385852b},
 created = {2018-08-09T16:38:14.086Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-08-09T16:38:14.086Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 bibtype = {article},
 author = {Mathieu, V. and Nys, J. and Fernández-Ramírez, C. and Blin, A. N. Hiller and Jackura, A. and Pilloni, A. and Szczepaniak, A. P. and Fox, G.},
 doi = {10.1103/PhysRevD.98.014041},
 journal = {Physical Review D},
 number = {1}
}

@book{
 title = {Big data and extreme-scale computing: Pathways to Convergence-Toward a shaping strategy for a future software and data ecosystem for scientific inquiry},
 type = {book},
 year = {2018},
 source = {International Journal of High Performance Computing Applications},
 keywords = {Big data,extreme-scale computing,future software,high-end data analysis,traditional HPC},
 pages = {435-479},
 volume = {32},
 issue = {4},
 id = {4ae61f3f-085a-3129-a8e2-bfa5b190274e},
 created = {2018-08-09T16:38:14.173Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-08-09T16:38:14.173Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© The Author(s) 2018. Over the past four years, the Big Data and Exascale Computing (BDEC) project organized a series of five international workshops that aimed to explore the ways in which the new forms of data-centric discovery introduced by the ongoing revolution in high-end data analysis (HDA) might be integrated with the established, simulation-centric paradigm of the high-performance computing (HPC) community. Based on those meetings, we argue that the rapid proliferation of digital data generators, the unprecedented growth in the volume and diversity of the data they generate, and the intense evolution of the methods for analyzing and using that data are radically reshaping the landscape of scientific computing. The most critical problems involve the logistics of wide-area, multistage workflows that will move back and forth across the computing continuum, between the multitude of distributed sensors, instruments and other devices at the networks edge, and the centralized resources of commercial clouds and HPC centers. We suggest that the prospects for the future integration of technological infrastructures and research ecosystems need to be considered at three different levels. First, we discuss the convergence of research applications and workflows that establish a research paradigm that combines both HPC and HDA, where ongoing progress is already motivating efforts at the other two levels. Second, we offer an account of some of the problems involved with creating a converged infrastructure for peripheral environments, that is, a shared infrastructure that can be deployed throughout the network in a scalable manner to meet the highly diverse requirements for processing, communication, and buffering/storage of massive data workflows of many different scientific domains. Third, we focus on some opportunities for software ecosystem convergence in big, logically centralized facilities that execute large-scale simulations and models and/or perform large-scale data analytics. We close by offering some conclusions and recommendations for future investment and policy review.},
 bibtype = {book},
 author = {Asch, M. and Moore, T. and Badia, R. and Beck, M. and Beckman, P. and Bidot, T. and Bodin, F. and Cappello, F. and Choudhary, A. and de Supinski, B. and Deelman, E. and Dongarra, J. and Dubey, A. and Fox, G. and Fu, H. and Girona, S. and Gropp, W. and Heroux, M. and Ishikawa, Y. and Keahey, K. and Keyes, D. and Kramer, W. and Lavignon, J. F. and Lu, Y. and Matsuoka, S. and Mohr, B. and Reed, D. and Requena, S. and Saltz, J. and Schulthess, T. and Stevens, R. and Swany, M. and Szalay, A. and Tang, W. and Varoquaux, G. and Vilotte, J. P. and Wisniewski, R. and Xu, Z. and Zacharov, I.},
 doi = {10.1177/1094342018778123}
}

© The Author(s) 2018. Over the past four years, the Big Data and Exascale Computing (BDEC) project organized a series of five international workshops that aimed to explore the ways in which the new forms of data-centric discovery introduced by the ongoing revolution in high-end data analysis (HDA) might be integrated with the established, simulation-centric paradigm of the high-performance computing (HPC) community. Based on those meetings, we argue that the rapid proliferation of digital data generators, the unprecedented growth in the volume and diversity of the data they generate, and the intense evolution of the methods for analyzing and using that data are radically reshaping the landscape of scientific computing. The most critical problems involve the logistics of wide-area, multistage workflows that will move back and forth across the computing continuum, between the multitude of distributed sensors, instruments and other devices at the networks edge, and the centralized resources of commercial clouds and HPC centers. We suggest that the prospects for the future integration of technological infrastructures and research ecosystems need to be considered at three different levels. First, we discuss the convergence of research applications and workflows that establish a research paradigm that combines both HPC and HDA, where ongoing progress is already motivating efforts at the other two levels. Second, we offer an account of some of the problems involved with creating a converged infrastructure for peripheral environments, that is, a shared infrastructure that can be deployed throughout the network in a scalable manner to meet the highly diverse requirements for processing, communication, and buffering/storage of massive data workflows of many different scientific domains. Third, we focus on some opportunities for software ecosystem convergence in big, logically centralized facilities that execute large-scale simulations and models and/or perform large-scale data analytics. We close by offering some conclusions and recommendations for future investment and policy review.

@techreport{
 title = {Big Data Benchmarks on Bare Metal Cloud},
 type = {techreport},
 year = {2018},
 pages = {1-5},
 id = {01c542a9-497f-3765-bef3-35536d0e380d},
 created = {2018-08-09T16:38:14.630Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T19:07:08.133Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {High performance computing requires to deal with a large number of applications running on different environments, and bare metal cloud is promising to enable new hardware features but in a easier way than traditional HPC systems. Data that we need to deal with is growing exponentially although many big data software support processing them at scale. We perform big data benchmark on public bare metal clouds to demonstrate computing performance with direct hardware access and block storage using up to 25000 and 32000 IOPS respectively for Oracle and Amazon. The preliminary results indicate that Amazon and Oracle are competitive supporting high throughput and low latency with operations in parallel. We investigate further on storage options available on Oracle bare metal with different data sets and anticipate to evaluate petabyte-scale workloads on cluster configurations in the future.},
 bibtype = {techreport},
 author = {Lee, Hyungro and Fox, Geoffrey C},
 doi = {10.13140/RG.2.2.12204.36486}
}

High performance computing requires to deal with a large number of applications running on different environments, and bare metal cloud is promising to enable new hardware features but in a easier way than traditional HPC systems. Data that we need to deal with is growing exponentially although many big data software support processing them at scale. We perform big data benchmark on public bare metal clouds to demonstrate computing performance with direct hardware access and block storage using up to 25000 and 32000 IOPS respectively for Oracle and Amazon. The preliminary results indicate that Amazon and Oracle are competitive supporting high throughput and low latency with operations in parallel. We investigate further on storage options available on Oracle bare metal with different data sets and anticipate to evaluate petabyte-scale workloads on cluster configurations in the future.

@inproceedings{
 title = {Performance Characterization of Multi-threaded Graph Processing Applications on Intel Many-Integrated-Core Architecture},
 type = {inproceedings},
 year = {2018},
 pages = {199–208},
 websites = {http://arxiv.org/abs/1708.04701},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 id = {61e168f8-dae8-3e59-b1a6-46f14e9f406c},
 created = {2018-08-09T16:38:14.663Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T15:26:00.755Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Intel Xeon Phi many-integrated-core (MIC) architectures usher in a new era of terascale integration. Among emerging killer applications, parallel graph processing has been a critical technique to analyze connected data. In this paper, we empirically evaluate various computing platforms including an Intel Xeon E5 CPU, a Nvidia Geforce GTX1070 GPU and an Xeon Phi 7210 processor codenamed Knights Landing (KNL) in the domain of parallel graph processing. We show that the KNL gains encouraging performance when processing graphs, so that it can become a promising solution to accelerating multi-threaded graph applications. We further characterize the impact of KNL architectural enhancements on the performance of a state-of-the art graph framework.We have four key observations: 1 Different graph applications require distinctive numbers of threads to reach the peak performance. For the same application, various datasets need even different numbers of threads to achieve the best performance. 2 Only a few graph applications benefit from the high bandwidth MCDRAM, while others favor the low latency DDR4 DRAM. 3 Vector processing units executing AVX512 SIMD instructions on KNLs are underutilized when running the state-of-the-art graph framework. 4 The sub-NUMA cache clustering mode offering the lowest local memory access latency hurts the performance of graph benchmarks that are lack of NUMA awareness. At last, We suggest future works including system auto-tuning tools and graph framework optimizations to fully exploit the potential of KNL for parallel graph processing.},
 bibtype = {inproceedings},
 author = {Liu, Xu and Chen, Langshi and Firoz, Jesun S. and Qiu, Judy and Jiang, Lei},
 doi = {10.1109/ISPASS.2018.00033},
 booktitle = {2018 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2018}
}

Intel Xeon Phi many-integrated-core (MIC) architectures usher in a new era of terascale integration. Among emerging killer applications, parallel graph processing has been a critical technique to analyze connected data. In this paper, we empirically evaluate various computing platforms including an Intel Xeon E5 CPU, a Nvidia Geforce GTX1070 GPU and an Xeon Phi 7210 processor codenamed Knights Landing (KNL) in the domain of parallel graph processing. We show that the KNL gains encouraging performance when processing graphs, so that it can become a promising solution to accelerating multi-threaded graph applications. We further characterize the impact of KNL architectural enhancements on the performance of a state-of-the art graph framework.We have four key observations: 1 Different graph applications require distinctive numbers of threads to reach the peak performance. For the same application, various datasets need even different numbers of threads to achieve the best performance. 2 Only a few graph applications benefit from the high bandwidth MCDRAM, while others favor the low latency DDR4 DRAM. 3 Vector processing units executing AVX512 SIMD instructions on KNLs are underutilized when running the state-of-the-art graph framework. 4 The sub-NUMA cache clustering mode offering the lowest local memory access latency hurts the performance of graph benchmarks that are lack of NUMA awareness. At last, We suggest future works including system auto-tuning tools and graph framework optimizations to fully exploit the potential of KNL for parallel graph processing.

@article{
 title = {Crossover analysis and automated layer-tracking assessment of the extracted DEM of the basal topography of the canadian arctic archipelago ice-cap},
 type = {article},
 year = {2018},
 keywords = {DEM,SAR,Synthetic aperture radar imaging,ice,ice-bottom tracking,tomography},
 pages = {862-867},
 id = {4b87961b-9199-3268-8c2e-f98721b40422},
 created = {2018-08-09T16:38:14.725Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-08-09T16:38:14.725Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© 2018 IEEE. In 2014, as part of the NASA Operation IceBridge project, the Center for Remote Sensing of Ice Sheets operated a multi-beam synthetic aperture radar depth sounder/imager over the Canadian Arctic Archipelago (CAA) to generate digital elevation models (DEMs) of the glacial basal topography. In this work, we briefly describe the processing steps that led to the generation of these DEMs, algorithm improvements over previously published results, and assess the results from two different perspectives. First, we evaluate the self-consistency of the DEMs where flight paths cross over each other and two measurements are made at the same location. Secondly, we compare the quality of the outputs of the ice-bottom tracker before and after applying manual corrections to the tracker results; the tracker is an algorithm that we implemented to automatically track the ice-bottom. Even though the CAA ice-caps are mountainous areas, where the scenes often have ice and no ice regions, which makes the imaging complicated, the statistical results show good tracking performance and a good match between the overlapped DEMs, where the mean error of the crossover DEMs is 37±9 m.},
 bibtype = {article},
 author = {Al-Ibadi, Mohanad and Sprick, Jordan and Athinarapu, Sravya and Berger, Victor and Stumpf, Theresa and Paden, John and Leuschen, Carl and Rodriguez, Fernando and Xu, Mingze and Crandall, David and Fox, Geoffrey and Burgess, David and Sharp, Martin and Copland, Luke and Van Wychen, Wesley},
 doi = {10.1109/RADAR.2018.8378673},
 journal = {2018 IEEE Radar Conference, RadarConf 2018}
}

© 2018 IEEE. In 2014, as part of the NASA Operation IceBridge project, the Center for Remote Sensing of Ice Sheets operated a multi-beam synthetic aperture radar depth sounder/imager over the Canadian Arctic Archipelago (CAA) to generate digital elevation models (DEMs) of the glacial basal topography. In this work, we briefly describe the processing steps that led to the generation of these DEMs, algorithm improvements over previously published results, and assess the results from two different perspectives. First, we evaluate the self-consistency of the DEMs where flight paths cross over each other and two measurements are made at the same location. Secondly, we compare the quality of the outputs of the ice-bottom tracker before and after applying manual corrections to the tracker results; the tracker is an algorithm that we implemented to automatically track the ice-bottom. Even though the CAA ice-caps are mountainous areas, where the scenes often have ice and no ice regions, which makes the imaging complicated, the statistical results show good tracking performance and a good match between the overlapped DEMs, where the mean error of the crossover DEMs is 37±9 m.

@techreport{
 title = {Detecting ice layers in radar images with deep learning},
 type = {techreport},
 year = {2018},
 pages = {2-5},
 issue = {April},
 websites = {https://pdfs.semanticscholar.org/e24d/e190e0e01e0e53003fa83daeb4557859f9f6.pdf},
 id = {25d6b7f7-80f9-348a-8419-f79c690caacf},
 created = {2018-08-09T16:38:14.850Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T20:01:27.940Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {This paper proposes a Deep Convolutional Neural Network approach to detect Ice surface and bottom layers from radar imagery. Radar images are capable to penetrate the earth surface and provide us with valuable information from the underlying layers of ice surface. In recent years, deep hierarchical learning techniques for object detection and segmentation greatly improved the performance of traditional techniques based on hand-crafted feature engineering. We designed a hybrid Deep Convolutional Network to produce the images of surface and bottom ice boundary as outputs. Our network takes advantage of undecimated wavelet transform to provide the highest level of information from radar images, as well as multi-layer and multi-scale optimized architecture. In this work, radar images from 2009-2016 NASA Operation IceBridge Mission are used to train and test the network. Our network outperformed the state-of-the art accuracy.},
 bibtype = {techreport},
 author = {Hamid Kamangir, Maryam Rahnemoonfar, Dugan Dobbs, J Paden, Geoffrey Fox}
}

This paper proposes a Deep Convolutional Neural Network approach to detect Ice surface and bottom layers from radar imagery. Radar images are capable to penetrate the earth surface and provide us with valuable information from the underlying layers of ice surface. In recent years, deep hierarchical learning techniques for object detection and segmentation greatly improved the performance of traditional techniques based on hand-crafted feature engineering. We designed a hybrid Deep Convolutional Network to produce the images of surface and bottom ice boundary as outputs. Our network takes advantage of undecimated wavelet transform to provide the highest level of information from radar images, as well as multi-layer and multi-scale optimized architecture. In this work, radar images from 2009-2016 NASA Operation IceBridge Mission are used to train and test the network. Our network outperformed the state-of-the art accuracy.

@article{
 title = {Finding and Counting Tree-Like Subgraphs Using MapReduce},
 type = {article},
 year = {2018},
 pages = {217-230},
 volume = {4},
 websites = {https://ieeexplore.ieee.org/document/8090537/},
 month = {7},
 day = {1},
 id = {ad584e09-e1ea-3c47-937b-eabd3bdda929},
 created = {2019-08-21T17:43:33.013Z},
 accessed = {2019-08-21},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T17:43:33.088Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 bibtype = {article},
 author = {Zhao, Zhao and Chen, Langshi and Avram, Mihai and Li, Meng and Wang, Guanying and Butt, Ali and Khan, Maleq and Marathe, Madhav and Qiu, Judy and Vullikanti, Anil},
 doi = {10.1109/TMSCS.2017.2768426},
 journal = {IEEE Transactions on Multi-Scale Computing Systems},
 number = {3}
}

@inproceedings{
 title = {Object Detection by a Super-Resolution Method and a Convolutional Neural Networks},
 type = {inproceedings},
 year = {2018},
 keywords = {CNN,convolution neural networks,deep learning,machine learning,object detection,super-resolution},
 pages = {2263-2269},
 month = {1},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {22},
 id = {0f6820bd-713c-3f62-8a5f-e63a49b90418},
 created = {2019-08-21T20:17:02.227Z},
 accessed = {2019-08-21},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T20:29:02.814Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Recently with many blurless or slightly blurred images, convolutional neural networks classify objects with around 90 percent classification rates, even if there are variable sized images. However, small object regions or cropping of images make object detection or classification difficult and decreases the detection rates. In many methods related to convolutional neural network (CNN), Bilinear or Bicubic algorithms are popularly used to interpolate region of interests. To overcome the limitations of these algorithms, we introduce a super-resolution method applied to the cropped regions or candidates, and this leads to improve recognition rates for object detection and classification. Large object candidates comparable in size of the full image have good results for object detections using many popular conventional methods. However, for smaller region candidates, using our super-resolution preprocessing and region candidates, allows a CNN to outperform conventional methods in the number of detected objects when tested on the VOC2007 and MSO datasets.},
 bibtype = {inproceedings},
 author = {Na, Bokyoon and Fox, Geoffrey C.},
 doi = {10.1109/BigData.2018.8622135},
 booktitle = {2018 IEEE International Conference on Big Data, Big Data}
}

Recently with many blurless or slightly blurred images, convolutional neural networks classify objects with around 90 percent classification rates, even if there are variable sized images. However, small object regions or cropping of images make object detection or classification difficult and decreases the detection rates. In many methods related to convolutional neural network (CNN), Bilinear or Bicubic algorithms are popularly used to interpolate region of interests. To overcome the limitations of these algorithms, we introduce a super-resolution method applied to the cropped regions or candidates, and this leads to improve recognition rates for object detection and classification. Large object candidates comparable in size of the full image have good results for object detections using many popular conventional methods. However, for smaller region candidates, using our super-resolution preprocessing and region candidates, allows a CNN to outperform conventional methods in the number of detected objects when tested on the VOC2007 and MSO datasets.

@techreport{
 title = {Machine Learning for Parameter Auto-tuning in Molecular Dynamics Simulations: Efficient Dynamics of Ions near Polarizable Nanoparticles},
 type = {techreport},
 year = {2018},
 keywords = {Auto-tuning,Energy Minimization,Hybrid MPI/OpenMP,Machine Learning,Nanoscale Simulations,Parallel Computing},
 pages = {15},
 websites = {www.sagepub.com/,http://dsc.soic.indiana.edu/publications/Manuscript.IJHPCA.Nov2018.pdf},
 id = {87fbf87a-5e47-371e-b171-17939c819fcc},
 created = {2019-09-03T19:11:21.598Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:00:33.304Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {true},
 private_publication = {false},
 abstract = {Simulating the dynamics of ions near polarizable nanoparticles (NPs) is extremely challenging due to the need to solve the Poisson equation at every simulation timestep. Recently, a molecular dynamics (MD) method based on a dynamical optimization framework bypassed this obstacle by representing the polarization charge density as virtual dynamic variables, and evolving them in parallel with the physical dynamics of ions. We highlight the computational gains accessible with the integration of machine learning (ML) methods for parameter prediction in MD simulations by demonstrating how they were realized in MD simulations of ions near polarizable NPs. An artificial neural network based regression model was integrated with MD and predicted the optimal simulation timestep and critical parameters characterizing the virtual system on-the-fly with 94.3% success. The integration of ML method with hybrid OpenMP/MPI parallelized MD simulations generated accurate dynamics of thousands of ions in the presence of polarizable NPs for over 10 million steps (with a maximum simulated physical time over 30 ns) while reducing the computational time from thousands of hours to tens of hours yielding a maximum speedup of ≈ 3 from ML-only acceleration and a maximum overall speedup of ≈ 600 from ML-hybrid Open/MPI combined method. Extraction of ionic structure in concentrated electrolytes near oil-water emulsions demonstrates the success of the method. The approach can be generalized to select optimal parameters in other molecular dynamics applications and energy minimization problems.},
 bibtype = {techreport},
 author = {Kadupitiya, Jcs and Fox, Geoffrey C and Jadhao, Vikram},
 doi = {10.1177/ToBeAssigned}
}

Simulating the dynamics of ions near polarizable nanoparticles (NPs) is extremely challenging due to the need to solve the Poisson equation at every simulation timestep. Recently, a molecular dynamics (MD) method based on a dynamical optimization framework bypassed this obstacle by representing the polarization charge density as virtual dynamic variables, and evolving them in parallel with the physical dynamics of ions. We highlight the computational gains accessible with the integration of machine learning (ML) methods for parameter prediction in MD simulations by demonstrating how they were realized in MD simulations of ions near polarizable NPs. An artificial neural network based regression model was integrated with MD and predicted the optimal simulation timestep and critical parameters characterizing the virtual system on-the-fly with 94.3% success. The integration of ML method with hybrid OpenMP/MPI parallelized MD simulations generated accurate dynamics of thousands of ions in the presence of polarizable NPs for over 10 million steps (with a maximum simulated physical time over 30 ns) while reducing the computational time from thousands of hours to tens of hours yielding a maximum speedup of ≈ 3 from ML-only acceleration and a maximum overall speedup of ≈ 600 from ML-hybrid Open/MPI combined method. Extraction of ionic structure in concentrated electrolytes near oil-water emulsions demonstrates the success of the method. The approach can be generalized to select optimal parameters in other molecular dynamics applications and energy minimization problems.

@techreport{
 title = {Contributions to High-Performance Big Data Computing},
 type = {techreport},
 year = {2018},
 keywords = {Big Data,Biomolecular simulations,Clouds,Graph Analytics,HPC,MIDAS,Network Science,Pathology,Polar Science,SPIDAL},
 websites = {http://dsc.soic.indiana.edu/publications/FormattedSPIDALPaperJune2019.pdf,https://www.researchgate.net/publication/328090399_Contributions_to_High-Performance_Big_Data_Computing},
 id = {b752ef45-82a3-3a91-a217-ba053612cb6a},
 created = {2019-09-03T19:32:08.432Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T19:32:08.519Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Our project is at the interface of Big Data and HPC-High-Performance Big Data computing and this paper describes a collaboration between 7 collaborating Universities at Arizona State, Indiana (lead), Kansas, Rutgers, Stony Brook, Virginia Tech, and Utah. It addresses the intersection of High-performance and Big Data computing with several different application areas or communities driving the requirements for software systems and algorithms. We describe the base architecture, including the HPC-ABDS, High-Performance Computing enhanced Apache Big Data Stack, and an application use case study identifying key features that determine software and algorithm requirements. We summarize middleware including Harp-DAAL collective communication layer, Twister2 Big Data toolkit, and pilot jobs. Then we present the SPIDAL Scalable Parallel Interoperable Data Analytics Library and our work for it in core machine-learning, image processing and the application communities, Network science, Polar Science, Biomolecular Simulations, Pathology, and Spatial systems. We describe basic algorithms and their integration in end-to-end use cases.},
 bibtype = {techreport},
 author = {Fox, Geoffrey and Qiu, Judy and Crandall, David and Laszewski, Gregor Von and Beckstein, Oliver and Paden, John and Paraskevakos, Ioannis and Jha, Shantenu and Wang, Fusheng and Marathe, Madhav and Vullikanti, Anil and Cheatham, Thomas}
}

Our project is at the interface of Big Data and HPC-High-Performance Big Data computing and this paper describes a collaboration between 7 collaborating Universities at Arizona State, Indiana (lead), Kansas, Rutgers, Stony Brook, Virginia Tech, and Utah. It addresses the intersection of High-performance and Big Data computing with several different application areas or communities driving the requirements for software systems and algorithms. We describe the base architecture, including the HPC-ABDS, High-Performance Computing enhanced Apache Big Data Stack, and an application use case study identifying key features that determine software and algorithm requirements. We summarize middleware including Harp-DAAL collective communication layer, Twister2 Big Data toolkit, and pilot jobs. Then we present the SPIDAL Scalable Parallel Interoperable Data Analytics Library and our work for it in core machine-learning, image processing and the application communities, Network science, Polar Science, Biomolecular Simulations, Pathology, and Spatial systems. We describe basic algorithms and their integration in end-to-end use cases.

@article{
 title = {Harp-DAAL for High Performance Big Data Computing},
 type = {article},
 year = {2018},
 websites = {https://software.intel.com/sites/default/files/parallel-universe-issue-32.pdf},
 month = {3},
 id = {32dca58a-56e5-3cda-82f8-523134016670},
 created = {2019-09-03T19:43:04.756Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T19:43:04.839Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Large-scale data analytics is revolutionizing many business and scientific domains. Easy-to-use scalable parallel techniques are necessary to process big data and gain meaningful insights. We introduce a novel HPC-Cloud convergence framework named Harp-DAAL and demonstrate that the combination of Big Data (Hadoop) and HPC techniques can simultaneously achieve productivity and performance. Harp is a distributed Hadoop-based framework that orchestrates efficient node synchronization [1]. Harp uses Intel ® Data Analytics Accelerator Library (DAAL) [2], for its highly optimized kernels on Intel ® Xeon and Xeon Phi architectures. This way the high-level API of Big Data tools can be combined with intra-node fine-grained parallelism that is optimized for HPC platforms. We illustrate this framework in detail with K-means clustering, a computation-bounded algorithm used in image clustering. We also show the broad applicability of Harp-DAAL by discussing the performance of three other big data algorithms: Subgraph Counting by color coding, Matrix Factorization and Latent Dirichlet Allocation. They share issues such as load imbalance, irregular structure, and communication issues that create difficult challenges. Figure 1 Cloud-HPC interoperable software for High Performance Big Data Analytics at Scale The categories in Figure 1 illustrate a classification of data intensive computation into five computation models that map into five distinct system architectures. It starts with Sequential, followed by centralized batch architectures corresponding exactly to the three forms of MapReduce: Map-Only, MapReduce and Iterative MapReduce. Category five is the classic MPI model. Harp brings Hadoop users the benefits of supporting all 5 classes of data-intensive computation, from pleasingly parallel to machine learning and simulations. We have expanded the applicability of Hadoop (with Harp plugin) for more classes of Big Data applications, especially complex data analytics such as machine learning and graph. We redesign a modular software stack with native kernels (with DAAL) to effectively utilize scale-up servers for machine learning and data analytics applications. Harp-DAAL shows how simulations and Big Data can use common programming environments with a runtime based on a rich set of collectives and libraries.},
 bibtype = {article},
 author = {Qiu, Judy},
 journal = {Parallel Universe}
}

Large-scale data analytics is revolutionizing many business and scientific domains. Easy-to-use scalable parallel techniques are necessary to process big data and gain meaningful insights. We introduce a novel HPC-Cloud convergence framework named Harp-DAAL and demonstrate that the combination of Big Data (Hadoop) and HPC techniques can simultaneously achieve productivity and performance. Harp is a distributed Hadoop-based framework that orchestrates efficient node synchronization [1]. Harp uses Intel ® Data Analytics Accelerator Library (DAAL) [2], for its highly optimized kernels on Intel ® Xeon and Xeon Phi architectures. This way the high-level API of Big Data tools can be combined with intra-node fine-grained parallelism that is optimized for HPC platforms. We illustrate this framework in detail with K-means clustering, a computation-bounded algorithm used in image clustering. We also show the broad applicability of Harp-DAAL by discussing the performance of three other big data algorithms: Subgraph Counting by color coding, Matrix Factorization and Latent Dirichlet Allocation. They share issues such as load imbalance, irregular structure, and communication issues that create difficult challenges. Figure 1 Cloud-HPC interoperable software for High Performance Big Data Analytics at Scale The categories in Figure 1 illustrate a classification of data intensive computation into five computation models that map into five distinct system architectures. It starts with Sequential, followed by centralized batch architectures corresponding exactly to the three forms of MapReduce: Map-Only, MapReduce and Iterative MapReduce. Category five is the classic MPI model. Harp brings Hadoop users the benefits of supporting all 5 classes of data-intensive computation, from pleasingly parallel to machine learning and simulations. We have expanded the applicability of Hadoop (with Harp plugin) for more classes of Big Data applications, especially complex data analytics such as machine learning and graph. We redesign a modular software stack with native kernels (with DAAL) to effectively utilize scale-up servers for machine learning and data analytics applications. Harp-DAAL shows how simulations and Big Data can use common programming environments with a runtime based on a rich set of collectives and libraries.

@article{
 title = {Global analysis of charge exchange meson production at high energies},
 type = {article},
 year = {2018},
 keywords = {doi:10.1103/PhysRevD.98.034020 url:https://doi.org/10.1103/PhysRevD.98.034020},
 id = {c048e46b-1e25-35ef-828e-bb9cb4e9f25c},
 created = {2019-09-03T20:11:06.253Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:11:06.344Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Many experiments that are conducted to study the hadron spectrum rely on peripheral resonance production. Hereby, the rapidity gap allows the process to be viewed as an independent fragmentation of the beam and the target, with the beam fragmentation dominated by production and decays of meson resonances. We test this separation by determining the kinematic regimes that are dominated by factorizable contributions, indicating the most favorable regions to perform this kind of experiments. In doing so, we use a Regge model to analyze the available world data of charge exchange meson production with beam momentum above 5 GeV in the laboratory frame that are not dominated by either pion or Pomeron exchanges. We determine the Regge residues and point out the kinematic regimes which are dominated by factorizable contributions.},
 bibtype = {article},
 author = {Nys, J and Hiller Blin, A N and Mathieu, V and Fernández-Ramírez, C and Jackura, A and Pilloni, A and Ryckebusch, J and Szczepaniak, A P and Fox, G},
 doi = {10.1103/PhysRevD.98.034020}
}

Many experiments that are conducted to study the hadron spectrum rely on peripheral resonance production. Hereby, the rapidity gap allows the process to be viewed as an independent fragmentation of the beam and the target, with the beam fragmentation dominated by production and decays of meson resonances. We test this separation by determining the kinematic regimes that are dominated by factorizable contributions, indicating the most favorable regions to perform this kind of experiments. In doing so, we use a Regge model to analyze the available world data of charge exchange meson production with beam momentum above 5 GeV in the laboratory frame that are not dominated by either pion or Pomeron exchanges. We determine the Regge residues and point out the kinematic regimes which are dominated by factorizable contributions.

@inproceedings{
 title = {Deep hybrid wavelet network for ice boundary detection in radra imagery},
 type = {inproceedings},
 year = {2018},
 keywords = {Deep learning,Holistically nested edge detection,Ice Boundary detection,Radar,Wavelet transform},
 pages = {3449-3452},
 volume = {2018-July},
 month = {10},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {31},
 id = {a27e053d-fc07-38e8-92a6-08ba7d516936},
 created = {2019-09-03T20:14:50.360Z},
 accessed = {2019-09-03},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-10-01T17:21:32.286Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {his paper proposes a deep convolutional neural network approach to detect ice surface and bottom layers from radar imagery. Radar images are capable to penetrate the ice surface and provide us with valuable information from the underlying layers of the ice surface. In recent years, deep hierarchical learning techniques for object detection and segmentation greatly improved the performance of traditional techniques based on hand-crafted feature engineering. We designed a deep convo-lutional network to produce the images of the surface and bottom ice boundary. Our network takes advantage of undecimated wavelet transform to provide the highest level of information from radar images, as well as multilayer and multi-scale optimized architecture. In this work, radar images from 2009-2016 NASA Operation IceBridge Mission are used to train and test the network. Our network outperformed the state-of-the art accuracy.},
 bibtype = {inproceedings},
 author = {Kamangir, Hamid and Rahnemoonfar, Maryam and Dobbs, Dugan and Paden, John and Fox, Geoffrey},
 doi = {10.1109/IGARSS.2018.8518617},
 booktitle = {International Geoscience and Remote Sensing Symposium (IGARSS)}
}

his paper proposes a deep convolutional neural network approach to detect ice surface and bottom layers from radar imagery. Radar images are capable to penetrate the ice surface and provide us with valuable information from the underlying layers of the ice surface. In recent years, deep hierarchical learning techniques for object detection and segmentation greatly improved the performance of traditional techniques based on hand-crafted feature engineering. We designed a deep convo-lutional network to produce the images of the surface and bottom ice boundary. Our network takes advantage of undecimated wavelet transform to provide the highest level of information from radar images, as well as multilayer and multi-scale optimized architecture. In this work, radar images from 2009-2016 NASA Operation IceBridge Mission are used to train and test the network. Our network outperformed the state-of-the art accuracy.

@inproceedings{
 title = {Evaluating the scientific impact of XSEDE},
 type = {inproceedings},
 year = {2018},
 keywords = {Bibliometrics,H-index,Scientific impact,Technology Audit Service,XDMoD,XSEDE},
 month = {7},
 publisher = {Association for Computing Machinery},
 day = {22},
 id = {e3d55e69-2834-3602-8ced-b88348e2b954},
 created = {2019-09-03T20:21:08.139Z},
 accessed = {2019-09-03},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:21:08.247Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We use the bibliometrics approach to evaluate the scientific impact of XSEDE. By utilizing publication data from various sources, e.g., ISI Web of Science and Microsoft Academic Graph, we calculate the impact metrics of XSEDE publications and show how they compare with non-XSEDE publication from the same field of study, or non-XSEDE peers from the same journal issue. We explain the dataset and data soruces involved and how we retrieved, cleaned, and curated millions of related publication entries. We then introduce the metrics we used for evaluation and comparison, and the methods used to calculate them. Detailed analysis results of Field Weighted Citation Impact (FWCI) and the peers comparison will be presented and discussed. We also explain how the same approaches could be used to evaluate publications from a similar organization or institute, to demonstrate the general applicability of the present evaluation approach providing impact even beyond XSEDE.},
 bibtype = {inproceedings},
 author = {Wang, Fugang and Fox, Geoffrey C. and Von Laszewski, Gregor and Furlani, Thomas R. and Gallo, Steven M. and Whitson, Timothy and DeLeon, Robert L.},
 doi = {10.1145/3219104.3219124},
 booktitle = {Proceedings of the Practice and Experience on Advanced Research Computing (PEARC '18)}
}

We use the bibliometrics approach to evaluate the scientific impact of XSEDE. By utilizing publication data from various sources, e.g., ISI Web of Science and Microsoft Academic Graph, we calculate the impact metrics of XSEDE publications and show how they compare with non-XSEDE publication from the same field of study, or non-XSEDE peers from the same journal issue. We explain the dataset and data soruces involved and how we retrieved, cleaned, and curated millions of related publication entries. We then introduce the metrics we used for evaluation and comparison, and the methods used to calculate them. Detailed analysis results of Field Weighted Citation Impact (FWCI) and the peers comparison will be presented and discussed. We also explain how the same approaches could be used to evaluate publications from a similar organization or institute, to demonstrate the general applicability of the present evaluation approach providing impact even beyond XSEDE.

@inproceedings{
 title = {Task-parallel analysis of molecular dynamics trajectories},
 type = {inproceedings},
 year = {2018},
 keywords = {Data analytics,MD analysis,MD simulations analysis,Task-parallel},
 month = {8},
 publisher = {Association for Computing Machinery},
 day = {13},
 id = {63391b42-2571-3487-8f93-4879fc5ddd87},
 created = {2019-09-03T20:29:25.490Z},
 accessed = {2019-09-03},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:29:25.590Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Different parallel frameworks for implementing data analysis applications have been proposed by the HPC and Big Data communities. In this paper, we investigate three task-parallel frameworks: Spark, Dask and RADICAL-Pilot with respect to their ability to support data analytics on HPC resources and compare them with MPI. We investigate the data analysis requirements of Molecular Dynamics (MD) simulations which are significant consumers of supercomputing cycles, producing immense amounts of data. A typical large-scale MD simulation of a physical system of O(100k) atoms over \musecs can produce from O(10) GB to O(1000) GBs of data. We propose and evaluate different approaches for parallelization of a representative set of MD trajectory analysis algorithms, in particular the computation of path similarity and leaflet identification. We evaluate Spark, Dask and RADICAL-Pilot with respect to their abstractions and runtime engine capabilities to support these algorithms. We provide a conceptual basis for comparing and understanding different frameworks that enable users to select the optimal system for each application. We also provide a quantitative performance analysis of the different algorithms across the three frameworks.},
 bibtype = {inproceedings},
 author = {Paraskevakos, Ioannis and Chantzialexiou, George and Luckow, Andre and Cheatham, Thomas E. and Khoshlessan, Mahzad and Beckstein, Oliver and Fox, Geoffrey C. and Jha, Shantenu},
 doi = {10.1145/3225058.3225128},
 booktitle = {Proceedings of the 47th International Conference on Parallel Processing (ICPP 2018)}
}

Different parallel frameworks for implementing data analysis applications have been proposed by the HPC and Big Data communities. In this paper, we investigate three task-parallel frameworks: Spark, Dask and RADICAL-Pilot with respect to their ability to support data analytics on HPC resources and compare them with MPI. We investigate the data analysis requirements of Molecular Dynamics (MD) simulations which are significant consumers of supercomputing cycles, producing immense amounts of data. A typical large-scale MD simulation of a physical system of O(100k) atoms over \musecs can produce from O(10) GB to O(1000) GBs of data. We propose and evaluate different approaches for parallelization of a representative set of MD trajectory analysis algorithms, in particular the computation of path similarity and leaflet identification. We evaluate Spark, Dask and RADICAL-Pilot with respect to their abstractions and runtime engine capabilities to support these algorithms. We provide a conceptual basis for comparing and understanding different frameworks that enable users to select the optimal system for each application. We also provide a quantitative performance analysis of the different algorithms across the three frameworks.

@article{
 title = {Anatomy of machine learning algorithm implementations in MPI, Spark, and Flink},
 type = {article},
 year = {2018},
 keywords = {Artificial intelligence; Big data; Data flow analy,Data flow modeling; Flink; High performance compu,Learning algorithms},
 pages = {61-73},
 volume = {32},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85039854240&doi=10.1177%2F1094342017712976&partnerID=40&md5=0a1048e69609d95f438e0b2f01466624},
 publisher = {SAGE Publications Inc.},
 id = {b9bbb94d-cc17-323e-be60-b7586112a4f9},
 created = {2019-09-03T20:32:35.497Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:32:35.557Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Kamburugamuve201861},
 source_type = {article},
 notes = {cited By 1},
 private_publication = {false},
 abstract = {With the ever-increasing need to analyze large amounts of data to get useful insights, it is essential to develop complex parallel machine learning algorithms that can scale with data and number of parallel processes. These algorithms need to run on large data sets as well as they need to be executed with minimal time in order to extract useful information in a time-constrained environment. Message passing interface (MPI) is a widely used model for developing such algorithms in high-performance computing paradigm, while Apache Spark and Apache Flink are emerging as big data platforms for large-scale parallel machine learning. Even though these big data frameworks are designed differently, they follow the data flow model for execution and user APIs. Data flow model offers fundamentally different capabilities than the MPI execution model, but the same type of parallelism can be used in applications developed in both models. This article presents three distinct machine learning algorithms implemented in MPI, Spark, and Flink and compares their performance and identifies strengths and weaknesses in each platform. © 2017, © The Author(s) 2017.},
 bibtype = {article},
 author = {Kamburugamuve, S and Wickramasinghe, P and Ekanayake, S and Fox, Geoffrey Charles},
 doi = {10.1177/1094342017712976},
 journal = {International Journal of High Performance Computing Applications},
 number = {1}
}

With the ever-increasing need to analyze large amounts of data to get useful insights, it is essential to develop complex parallel machine learning algorithms that can scale with data and number of parallel processes. These algorithms need to run on large data sets as well as they need to be executed with minimal time in order to extract useful information in a time-constrained environment. Message passing interface (MPI) is a widely used model for developing such algorithms in high-performance computing paradigm, while Apache Spark and Apache Flink are emerging as big data platforms for large-scale parallel machine learning. Even though these big data frameworks are designed differently, they follow the data flow model for execution and user APIs. Data flow model offers fundamentally different capabilities than the MPI execution model, but the same type of parallelism can be used in applications developed in both models. This article presents three distinct machine learning algorithms implemented in MPI, Spark, and Flink and compares their performance and identifies strengths and weaknesses in each platform. © 2017, © The Author(s) 2017.

@inproceedings{
 title = {Twister: Net - Communication Library for Big Data Processing in HPC and Cloud Environments},
 type = {inproceedings},
 year = {2018},
 keywords = {Big-data,Collectives,HPC,MPI,Streaming},
 pages = {383-391},
 volume = {2018-July},
 month = {9},
 publisher = {IEEE Computer Society},
 day = {7},
 id = {94a46ef9-f5f0-3cb1-af3f-9bc0ff28d6d9},
 created = {2019-09-04T18:58:29.225Z},
 accessed = {2019-09-04},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T19:02:44.104Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Streaming processing and batch data processing are the dominant forms of big data analytics today, with numerous systems such as Hadoop, Spark, and Heron designed to process the ever-increasing explosion of data. Generally, these systems are developed as single projects with aspects such as communication, task management, and data management integrated together. By contrast, we take a component-based approach to big data by developing the essential features of a big data system as independent components with polymorphic implementations to support different requirements. Consequently, we recognize the requirements of both dataflow used in popular Apache Systems and the Bulk Synchronous Processing communication style common in High-Performance Computing (HPC) for different applications. Message Passing Interface (MPI) implementations are dominant in HPC but there are no such standard libraries available for big data. Twister:Net is a stand-alone, highly optimized dataflow style parallel communication library which can be used by big data systems or advanced users. Twister:Net can work both in cloud environments using TCP or HPC environments using MPI implementations. This paper introduces Twister:Net and compares it with existing systems to highlight its design and performance. © 2018 IEEE.},
 bibtype = {inproceedings},
 author = {Kamburugamuve, Supun and Wickramasinghe, Pulasthi and Govindarajan, Kannan and Uyar, Ahmet and Gunduz, Gurhan and Abeykoon, Vibhatha and Fox, Geoffrey},
 doi = {10.1109/CLOUD.2018.00055},
 booktitle = {IEEE International Conference on Cloud Computing, CLOUD}
}

Streaming processing and batch data processing are the dominant forms of big data analytics today, with numerous systems such as Hadoop, Spark, and Heron designed to process the ever-increasing explosion of data. Generally, these systems are developed as single projects with aspects such as communication, task management, and data management integrated together. By contrast, we take a component-based approach to big data by developing the essential features of a big data system as independent components with polymorphic implementations to support different requirements. Consequently, we recognize the requirements of both dataflow used in popular Apache Systems and the Bulk Synchronous Processing communication style common in High-Performance Computing (HPC) for different applications. Message Passing Interface (MPI) implementations are dominant in HPC but there are no such standard libraries available for big data. Twister:Net is a stand-alone, highly optimized dataflow style parallel communication library which can be used by big data systems or advanced users. Twister:Net can work both in cloud environments using TCP or HPC environments using MPI implementations. This paper introduces Twister:Net and compares it with existing systems to highlight its design and performance. © 2018 IEEE.

@inproceedings{
 title = {Evaluation of Production Serverless Computing Environments},
 type = {inproceedings},
 year = {2018},
 keywords = {Amazon Lambda,Event-driven Computing,FaaS,Google Functions,IBM OpenWhisk,Microsoft Azure Functions,Serverless},
 pages = {442-450},
 volume = {2018-July},
 month = {9},
 publisher = {IEEE Computer Society},
 day = {7},
 id = {61d699cf-14a8-3263-8df7-816812912f8c},
 created = {2019-09-04T19:02:30.749Z},
 accessed = {2019-09-04},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T19:03:24.401Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Serverless computing provides a small runtime container to execute lines of codes without infrastructure management which is similar to Platform as a Service (PaaS) but a functional level. Amazon started the event-driven compute named Lambda functions in 2014 with a 25 concurrent limitation, but it now supports at least a thousand of concurrent invocation to process event messages generated by resources like databases, storage and system logs. Other providers, i.e., Google, Microsoft, and IBM offer a dynamic scaling manager to handle parallel requests of stateless functions in which additional containers are provisioning on new compute nodes for distribution. However, while functions are often developed for microservices and lightweight workload, they are associated with distributed data processing using the concurrent invocations. We claim that the current serverless computing environments can support dynamic applications in parallel when a partitioned task is executable on a small function instance. We present results of throughput, network bandwidth, a file I/O and compute performance regarding the concurrent invocations. We deployed a series of functions for distributed data processing to address the elasticity and then demonstrated the differences between serverless computing and virtual machines for cost efficiency and resource utilization.},
 bibtype = {inproceedings},
 author = {Lee, Hyungro and Satyam, Kumar and Fox, Geoffrey},
 doi = {10.1109/CLOUD.2018.00062},
 booktitle = {IEEE International Conference on Cloud Computing, CLOUD}
}

Serverless computing provides a small runtime container to execute lines of codes without infrastructure management which is similar to Platform as a Service (PaaS) but a functional level. Amazon started the event-driven compute named Lambda functions in 2014 with a 25 concurrent limitation, but it now supports at least a thousand of concurrent invocation to process event messages generated by resources like databases, storage and system logs. Other providers, i.e., Google, Microsoft, and IBM offer a dynamic scaling manager to handle parallel requests of stateless functions in which additional containers are provisioning on new compute nodes for distribution. However, while functions are often developed for microservices and lightweight workload, they are associated with distributed data processing using the concurrent invocations. We claim that the current serverless computing environments can support dynamic applications in parallel when a partitioned task is executable on a small function instance. We present results of throughput, network bandwidth, a file I/O and compute performance regarding the concurrent invocations. We deployed a series of functions for distributed data processing to address the elasticity and then demonstrated the differences between serverless computing and virtual machines for cost efficiency and resource utilization.

@inproceedings{
 title = {Multi-task Spatiotemporal Neural Networks for Structured Surface Reconstruction},
 type = {inproceedings},
 year = {2018},
 pages = {1273-1282},
 volume = {2018-Janua},
 month = {5},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {3},
 id = {9cb576b7-9f0b-3f41-be49-fbbebe85ceb7},
 created = {2019-09-04T19:30:15.722Z},
 accessed = {2019-09-04},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T19:39:15.307Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Deep learning methods have surpassed the performance of traditional techniques on a wide range of problems in computer vision, but nearly all of this work has studied consumer photos, where precisely correct output is often not critical. It is less clear how well these techniques may apply on structured prediction problems where fine-grained output with high precision is required, such as in scientific imaging domains. Here we consider the problem of segmenting echogram radar data collected from the polar ice sheets, which is challenging because segmentation boundaries are often very weak and there is a high degree of noise. We propose a multi-task spatiotemporal neural network that combines 3D ConvNets and Recurrent Neural Networks (RNNs) to estimate ice surface boundaries from sequences of tomographic radar images. We show that our model outperforms the state-of-the-art on this problem by (1) avoiding the need for hand-tuned parameters, (2) extracting multiple surfaces (ice-air and ice-bed) simultaneously, (3) requiring less non-visual metadata, and (4) being about 6 times faster.},
 bibtype = {inproceedings},
 author = {Xu, Mingze and Fan, Chenyou and Paden, John D. and Fox, Geoffrey C. and Crandall, David J.},
 doi = {10.1109/WACV.2018.00144},
 booktitle = {Proceedings - 2018 IEEE Winter Conference on Applications of Computer Vision, WACV 2018}
}

Deep learning methods have surpassed the performance of traditional techniques on a wide range of problems in computer vision, but nearly all of this work has studied consumer photos, where precisely correct output is often not critical. It is less clear how well these techniques may apply on structured prediction problems where fine-grained output with high precision is required, such as in scientific imaging domains. Here we consider the problem of segmenting echogram radar data collected from the polar ice sheets, which is challenging because segmentation boundaries are often very weak and there is a high degree of noise. We propose a multi-task spatiotemporal neural network that combines 3D ConvNets and Recurrent Neural Networks (RNNs) to estimate ice surface boundaries from sequences of tomographic radar images. We show that our model outperforms the state-of-the-art on this problem by (1) avoiding the need for hand-tuned parameters, (2) extracting multiple surfaces (ice-air and ice-bed) simultaneously, (3) requiring less non-visual metadata, and (4) being about 6 times faster.

@inproceedings{
 title = {Automated tracking of 2D and 3D ice radar imagery using Viterbi and TRW-S},
 type = {inproceedings},
 year = {2018},
 keywords = {Glaciology,Ice thickness,Ice-bottom tracking,Image classification,Radar tomography},
 pages = {4162-4165},
 volume = {2018-July},
 month = {10},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {31},
 id = {5c0d1936-b708-3116-a9e6-39ad3dbc9ab2},
 created = {2019-09-04T19:47:14.814Z},
 accessed = {2019-09-04},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T20:01:27.922Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {We present improvements to existing implementations of the Viterbi and TRW-S algorithms applied to ice-bottom layer tracking on 2D and 3D radar imagery, respectively. Along with an explanation of our modifications and the reasoning behind them, we present a comparison between our results, the results obtained with the original implementations, and those obtained with other proposed methods of performing ice-bottom layer tracking.},
 bibtype = {inproceedings},
 author = {Berger, Victor and Xu, Mingze and Chu, Shane and Crandall, David and Paden, John and Fox, Geoffrey C.},
 doi = {10.1109/IGARSS.2018.8519411},
 booktitle = {International Geoscience and Remote Sensing Symposium (IGARSS)}
}

We present improvements to existing implementations of the Viterbi and TRW-S algorithms applied to ice-bottom layer tracking on 2D and 3D radar imagery, respectively. Along with an explanation of our modifications and the reasoning behind them, we present a comparison between our results, the results obtained with the original implementations, and those obtained with other proposed methods of performing ice-bottom layer tracking.

@article{
 title = {Features of πΔ photoproduction at high energies},
 type = {article},
 year = {2018},
 pages = {77-81},
 volume = {779},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044396630&doi=10.1016%2Fj.physletb.2018.01.075&partnerID=40&md5=12e6f3f9ea386dbf28749cd0713aa855},
 publisher = {Elsevier B.V.},
 id = {bd6a0c54-b47d-34a9-9952-cb2fc8e8f7f4},
 created = {2019-09-04T20:01:04.505Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T20:01:04.577Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Nys201877},
 source_type = {article},
 notes = {cited By 0},
 private_publication = {false},
 abstract = {Hybrid/exotic meson spectroscopy searches at Jefferson Lab require the accurate theoretical description of the production mechanism in peripheral photoproduction. We develop a model for πΔ photoproduction at high energies (5≤Elab≤16 GeV) that incorporates both the absorbed pion and natural-parity cut contributions. We fit the available observables, providing a good description of the energy and angular dependencies of the experimental data. We also provide predictions for the photon beam asymmetry of charged pions at Elab=9 GeV which is expected to be measured by GlueX and CLAS12 experiments in the near future. © 2018 The Author},
 bibtype = {article},
 author = {Nys, J and Mathieu, V and Fernández-Ramírez, C and Jackura, A and Mikhasenko, M and Pilloni, A and Sherrill, N and Ryckebusch, J and Szczepaniak, A P and Fox, Geoffrey Charles and Center, Joint Physics Analysis},
 doi = {10.1016/j.physletb.2018.01.075},
 journal = {Physics Letters, Section B: Nuclear, Elementary Particle and High-Energy Physics}
}

Hybrid/exotic meson spectroscopy searches at Jefferson Lab require the accurate theoretical description of the production mechanism in peripheral photoproduction. We develop a model for πΔ photoproduction at high energies (5≤Elab≤16 GeV) that incorporates both the absorbed pion and natural-parity cut contributions. We fit the available observables, providing a good description of the energy and angular dependencies of the experimental data. We also provide predictions for the photon beam asymmetry of charged pions at Elab=9 GeV which is expected to be measured by GlueX and CLAS12 experiments in the near future. © 2018 The Author

@inproceedings{
 title = {Task Scheduling in Big Data - Review, Research Challenges, and Prospects},
 type = {inproceedings},
 year = {2018},
 keywords = {Big Data,Dataflow,MapReduce,Static and Dynamic Task Scheduling,Task Scheduling Model,Twister2},
 pages = {165-173},
 month = {8},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {20},
 id = {5e707db7-0220-3b41-9d54-27a55ec76915},
 created = {2019-09-04T20:06:00.531Z},
 accessed = {2019-09-04},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-04T20:39:06.516Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {—In a Big data computing, the processing of data requires a large amount of CPU cycles and network bandwidth and disk I/O. Dataflow is a programming model for processing Big data which consists of tasks organized in a graph structure. Scheduling these tasks is one of the key active research areas which mainly aims to place the tasks on available resources. It is essential to effectively schedule the tasks, in a manner that minimizes task completion time and increases utilization of resources. In recent years, various researchers have discussed and presented different task scheduling algorithms. In this research study, we have investigated the state-of-art of various types of task scheduling algorithms, scheduling considerations for batch and streaming processing, and task scheduling algorithms in the well-known open-source big data platforms. Furthermore, this study proposes a new task scheduling system to alleviate the problems persists in the existing task scheduling for big data.},
 bibtype = {inproceedings},
 author = {Govindarajan, Kannan and Kamburugamuve, Supun and Wickramasinghe, Pulasthi and Abeykoon, Vibhatha and Fox, Geoffrey},
 doi = {10.1109/ICoAC.2017.8441494},
 booktitle = {2017 9th International Conference on Advanced Computing, ICoAC 2017}
}

—In a Big data computing, the processing of data requires a large amount of CPU cycles and network bandwidth and disk I/O. Dataflow is a programming model for processing Big data which consists of tasks organized in a graph structure. Scheduling these tasks is one of the key active research areas which mainly aims to place the tasks on available resources. It is essential to effectively schedule the tasks, in a manner that minimizes task completion time and increases utilization of resources. In recent years, various researchers have discussed and presented different task scheduling algorithms. In this research study, we have investigated the state-of-art of various types of task scheduling algorithms, scheduling considerations for batch and streaming processing, and task scheduling algorithms in the well-known open-source big data platforms. Furthermore, this study proposes a new task scheduling system to alleviate the problems persists in the existing task scheduling for big data.

@inproceedings{
 title = {Comet - Tales from the Long Tail - Two years in and 10,000 users later},
 type = {inproceedings},
 year = {2017},
 volume = {Part F1287},
 id = {784afb90-68f5-3a70-989a-ca60831cb9c1},
 created = {2018-02-27T18:07:44.552Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T20:41:12.075Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© 2017 ACM. The Comet petascale supercomputer was put into production as an XSEDE resource in early 2015 with the goal of serving a much larger user community than HPC systems of similar size. The Comet project set an audacious goal of reaching over 10,000 users in its four years of planned operation. That goal was achieved in less than two years, due in large part to the adoption of policies that favor smaller allocations and science gateways. Here we describe our experiences in operating and supporting Comet, highlight some of the important science that it has enabled, and provide some practical lessons that we have learned by operating a system designed for the long tail of science.},
 bibtype = {inproceedings},
 author = {Strandea, S.M. and Caia, H. and Cooper, T. and Flammer, K. and Irving, C. and Von Laszewski, G. and Majumdar, A. and Mishin, D. and Papadopoulos, P. and Pfeiffer, W. and Sinkovits, R.S. and Tatineni, M. and Wagner, R. and Wang, F. and Wilkins-Diehr, N. and Wolter, N. and Norman, M.L.},
 doi = {10.1145/3093338.3093383},
 booktitle = {Proceedings of the Practice and Experience in Advanced Research Computing 2017 on Sustainability, Success and Impact (PEARC17)}
}

© 2017 ACM. The Comet petascale supercomputer was put into production as an XSEDE resource in early 2015 with the goal of serving a much larger user community than HPC systems of similar size. The Comet project set an audacious goal of reaching over 10,000 users in its four years of planned operation. That goal was achieved in less than two years, due in large part to the adoption of policies that favor smaller allocations and science gateways. Here we describe our experiences in operating and supporting Comet, highlight some of the important science that it has enabled, and provide some practical lessons that we have learned by operating a system designed for the long tail of science.

@inproceedings{
 title = {High-Performance Massive Subgraph Counting Using Pipelined Adaptive-Group Communication.},
 type = {inproceedings},
 year = {2017},
 pages = {173-197},
 websites = {https://doi.org/10.3233/978-1-61499-882-2-173},
 id = {cede4ef4-5ce9-3d2b-95e1-05ddae4bceed},
 created = {2019-08-21T14:40:57.335Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-08-21T14:56:08.848Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CPAPER},
 private_publication = {false},
 abstract = {Subgraph counting aims to count the number of occurrences of a subgraph T (aka as a template) in a given graph G. The basic problem has found applications in diverse domains. The problem is known to be computationally challenging – the complexity grows both as a function of T and G. Recent applications have motivated solving such problems on massive networks with billions of vertices. In this chapter, we study the subgraph counting problem from a parallel computing perspective. We discuss efficient parallel algorithms for approximately resolving subgraph counting problems by using the color-coding technique. We then present several system-level strategies to substantially improve the overall performance of the algorithm in massive subgraph counting problems. We propose: 1) a novel pipelined Adaptive-Group communication pattern to improve inter-node scalability, 2) a fine-grained pipeline design to effectively reduce the memory space of intermediate results, 3) partitioning neighbor lists of subgraph vertices to achieve better thread concurrency and workload balance. Experimentation on an Intel Xeon E5 cluster shows that our implementation achieves 5x speedup of performance compared to the state-of-the-art work while reduces the peak memory utilization by a factor of 2 on large templates of 12 to 15 vertices and input graphs of 2 to 5 billions of edges.},
 bibtype = {inproceedings},
 author = {Chen, Langshi and Peng, Bo and Ossen, Sabra and Vullikanti, Anil and Marathe, Madhav and Jiang, Lei and Qiu, Judy},
 doi = {10.3233/978-1-61499-882-2-173},
 booktitle = {Big Data and HPC: Ecosystem and Convergence, TopHPC 2017}
}

Subgraph counting aims to count the number of occurrences of a subgraph T (aka as a template) in a given graph G. The basic problem has found applications in diverse domains. The problem is known to be computationally challenging – the complexity grows both as a function of T and G. Recent applications have motivated solving such problems on massive networks with billions of vertices. In this chapter, we study the subgraph counting problem from a parallel computing perspective. We discuss efficient parallel algorithms for approximately resolving subgraph counting problems by using the color-coding technique. We then present several system-level strategies to substantially improve the overall performance of the algorithm in massive subgraph counting problems. We propose: 1) a novel pipelined Adaptive-Group communication pattern to improve inter-node scalability, 2) a fine-grained pipeline design to effectively reduce the memory space of intermediate results, 3) partitioning neighbor lists of subgraph vertices to achieve better thread concurrency and workload balance. Experimentation on an Intel Xeon E5 cluster shows that our implementation achieves 5x speedup of performance compared to the state-of-the-art work while reduces the peak memory utilization by a factor of 2 on large templates of 12 to 15 vertices and input graphs of 2 to 5 billions of edges.

@techreport{
 title = {Introduction to Harp: when Big Data Meets HPC},
 type = {techreport},
 year = {2017},
 pages = {10},
 websites = {https://pdfs.semanticscholar.org/f69e/9f2852c881da4df0142360c745441075a28f.pdf?_ga=2.28162492.1830027813.1567539825-1063534713.1566236187},
 id = {20db5fdd-9973-379a-a4d9-d45ba9e7ef30},
 created = {2019-09-03T19:46:38.941Z},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2019-09-03T19:46:39.017Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Harp Data analytics is undergoing a revolution in many scientific domains, demanding cost-effective parallel data analysis techniques. We consider the challenges of creating a high performance data analysis software framework in the context of the current HPC-ABDS software stack (High Performance Computing enhanced Apache Big Data Stack) [1]. We have summarized a list of current data processing software from either HPC or commercial sources [2]. Many critical components of the commodity stack (such as Hadoop) come from Apache open source projects for community usage, while HPC (such as collective communication) is needed to bring performance and other parallel computing capabilities. Many machine learning algorithms are built on iterative computation, which can be formulated as í µí°´"µí°´" = í µí°¹(í µí°·, í µí°´"µí°´"()) (1) where D is the observed dataset, A is model parameters to learn, and F is the model update function. The algorithm keeps updating model A until convergence, either by reaching a threshold criterion or fixed number of iterations. There are several advantages of this iterative procedure as apparently simple functions can iterate and produce complex behavior for interesting problems. The power of iteration and its extensions lies in the approximation or accuracy that can be obtained at each step even if the computation stops abruptly before converges to the final answer. To effectively support large-scale data processing, Twister [3] introduced iterative MapReduce using long-running processes or threads with in-memory caching of invariant data. Harp [4] introduces full collective communication in Table 1 (broadcast, reduce, allgather, allreduce, rotation, regroup or push & pull), adding a separate communication abstraction where the Harp prototype implements the MapCollective concept as a plug-in to Hadoop Ecosystem (see Figure 1 and Figure 2). Instead of using the shuffling phase, Harp uses optimized collective communication operations for data movement since fine-grained data alignment for multiple models is critical for improving performance. It further provides high-level interfaces with various synchronization patterns for parallelizing iterative computation. These enhancements make it possible to exploit HPC capabilities for big data software systems. Figure 1 Map-Collective Model Figure 2 Harp Architecture Shuffle M M M M Collective Communication M M M M R R MapCollective Model MapReduce Model YARN MapReduce V2 Harp MapReduce Applications MapCollective Applications Application Framework Resource Manager},
 bibtype = {techreport},
 author = {Zhang, Bingjing and Peng, Bo and Chen, Langshi and Li, Ethan and Zhou, Yiming and Qiu, Judy}
}

Harp Data analytics is undergoing a revolution in many scientific domains, demanding cost-effective parallel data analysis techniques. We consider the challenges of creating a high performance data analysis software framework in the context of the current HPC-ABDS software stack (High Performance Computing enhanced Apache Big Data Stack) [1]. We have summarized a list of current data processing software from either HPC or commercial sources [2]. Many critical components of the commodity stack (such as Hadoop) come from Apache open source projects for community usage, while HPC (such as collective communication) is needed to bring performance and other parallel computing capabilities. Many machine learning algorithms are built on iterative computation, which can be formulated as í µí°´"µí°´" = í µí°¹(í µí°·, í µí°´"µí°´"()) (1) where D is the observed dataset, A is model parameters to learn, and F is the model update function. The algorithm keeps updating model A until convergence, either by reaching a threshold criterion or fixed number of iterations. There are several advantages of this iterative procedure as apparently simple functions can iterate and produce complex behavior for interesting problems. The power of iteration and its extensions lies in the approximation or accuracy that can be obtained at each step even if the computation stops abruptly before converges to the final answer. To effectively support large-scale data processing, Twister [3] introduced iterative MapReduce using long-running processes or threads with in-memory caching of invariant data. Harp [4] introduces full collective communication in Table 1 (broadcast, reduce, allgather, allreduce, rotation, regroup or push & pull), adding a separate communication abstraction where the Harp prototype implements the MapCollective concept as a plug-in to Hadoop Ecosystem (see Figure 1 and Figure 2). Instead of using the shuffling phase, Harp uses optimized collective communication operations for data movement since fine-grained data alignment for multiple models is critical for improving performance. It further provides high-level interfaces with various synchronization patterns for parallelizing iterative computation. These enhancements make it possible to exploit HPC capabilities for big data software systems. Figure 1 Map-Collective Model Figure 2 Harp Architecture Shuffle M M M M Collective Communication M M M M R R MapCollective Model MapReduce Model YARN MapReduce V2 Harp MapReduce Applications MapCollective Applications Application Framework Resource Manager

@techreport{
 title = {Pervasive Technology Institute Annual Report: Research Innovations and Advanced Cyberinfrastructure Services in Support of IU Strategic Goals During FY 2017},
 type = {techreport},
 year = {2017},
 websites = {http://hdl.handle.net/2022/21809},
 id = {43d53477-a758-369b-b696-1091b9496913},
 created = {2020-09-10T00:01:45.285Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-09-10T00:01:45.285Z},
 read = {true},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Stewart2017},
 source_type = {RPRT},
 private_publication = {false},
 bibtype = {techreport},
 author = {Stewart, Craig; and Plale, Beth; and Welch, Von; and Pierce, Marlon; and Fox, Geoffrey C.; and Doak, Thomas G.; Hancock, David Y.; Henschel, Robert; and Link, Matthew R.; and Miller, Therese; and Wernert, Eric; and Boyles, Michael J.; and Fulton, Ben; and Weakley, Le Mai; and Ping, Robert; and Gniady, Tassie; and Snapp-Childs, Winona;}
}

@inproceedings{
 title = {Providing Statistical Reliability Guarantee for Cloud Clusters},
 type = {inproceedings},
 year = {2016},
 city = {Washington, DC.},
 id = {48047e2a-0dfc-3f3e-9d10-22999f0ace9c},
 created = {2018-02-27T18:07:43.249Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.249Z},
 read = {false},
 starred = {true},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Yang, Zhouhan and Du, Anna Ye and Das, Sanjukta and Ramesh, Ram and Furlani, Thomas and von Laszewski, Gregor and Qiao, Chunming},
 booktitle = {Providing Statistical Reliability Guarantee for Cloud Clusters. Submitted to Global Communications Conference}
}

@inproceedings{
 title = {User managed virtual clusters in comet},
 type = {inproceedings},
 year = {2016},
 volume = {17-21-July},
 id = {a7195775-747e-3e3e-80da-9f18cfb74122},
 created = {2018-02-27T18:07:45.201Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.201Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© 2016 ACM. Hardware virtualization has been gaining a significant share of computing time in the last years. Using virtual machines (VMs) for parallel computing is an attractive option for many users. A VM gives users a freedom of choosing an operating system, software stack and security policies, leaving the physical hardware, OS management, and billing to physical cluster administrators. The well-known solutions for cloud computing, both commercial (Amazon Cloud, Google Cloud, Yahoo Cloud, etc.) and open-source (OpenStack, Eucalyptus) provide platforms for running a single VM or a group of VMs. With all the benefits, there are also some drawbacks, which include reduced performance when running code inside of a VM, increased complexity of cluster management, as well as the need to learn new tools and protocols to manage the clusters. At SDSC, we have created a novel framework and infrastructure by providing virtual HPC clusters to projects using the NSF sponsored Comet supercomputer. Managing virtual clusters on Comet is similar to managing a baremetal cluster in terms of processes and tools that are employed. This is beneficial because such processes and tools are familiar to cluster administrators. Unlike platforms like AWS, Comet's virtualization capability supports installing VMs from ISOs (i.e., a CD-ROM or DVD image) or via an isolated management VLAN (PXE). At the same time, we're helping projects take advantage of VMs by providing an enhanced client tool for interaction with our management system called Cloudmesh client. Cloudmesh client can also be used to manage virtual machines on OpenStack, AWS, and Azure. The article describes our design and approach to running virtual clusters, the tools we developed, and initial user experience.},
 bibtype = {inproceedings},
 author = {Wagner, R. and Papadopoulos, P. and Mishin, D. and Cooper, T. and Tatineti, M. and Von Laszewski, G. and Wang, F. and Fox, G.C.},
 doi = {10.1145/2949550.2949555},
 booktitle = {ACM International Conference Proceeding Series}
}

© 2016 ACM. Hardware virtualization has been gaining a significant share of computing time in the last years. Using virtual machines (VMs) for parallel computing is an attractive option for many users. A VM gives users a freedom of choosing an operating system, software stack and security policies, leaving the physical hardware, OS management, and billing to physical cluster administrators. The well-known solutions for cloud computing, both commercial (Amazon Cloud, Google Cloud, Yahoo Cloud, etc.) and open-source (OpenStack, Eucalyptus) provide platforms for running a single VM or a group of VMs. With all the benefits, there are also some drawbacks, which include reduced performance when running code inside of a VM, increased complexity of cluster management, as well as the need to learn new tools and protocols to manage the clusters. At SDSC, we have created a novel framework and infrastructure by providing virtual HPC clusters to projects using the NSF sponsored Comet supercomputer. Managing virtual clusters on Comet is similar to managing a baremetal cluster in terms of processes and tools that are employed. This is beneficial because such processes and tools are familiar to cluster administrators. Unlike platforms like AWS, Comet's virtualization capability supports installing VMs from ISOs (i.e., a CD-ROM or DVD image) or via an isolated management VLAN (PXE). At the same time, we're helping projects take advantage of VMs by providing an enhanced client tool for interaction with our management system called Cloudmesh client. Cloudmesh client can also be used to manage virtual machines on OpenStack, AWS, and Azure. The article describes our design and approach to running virtual clusters, the tools we developed, and initial user experience.

@article{
 title = {NIST Big Data Interoperability Framework: Volume 3, Use Cases and General Requirements (NIST Special Publication 1500-3)},
 type = {article},
 year = {2016},
 keywords = {Big Data,Big Data Application Provider,Big Data Framework Provider,Big Data characteristics,Big Data taxonomy,Data Consumer,Data Provider,Management Fabric,Security and Privacy Fabric,System Orchestrator,data science,reference architecture,use cases.},
 volume = {3},
 websites = {http://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.1500-3.pdf},
 id = {75fe5440-d1c2-346b-8d97-69a42dcf9a45},
 created = {2018-08-09T16:38:14.350Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-08-09T16:38:14.350Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Big Data is a term used to describe the large amount of data in the networked, digitized, sensor-laden, information-driven world. While opportunities exist with Big Data, the data can overwhelm traditional technical approaches and the growth of data is outpacing scientific and technological advances in data analytics. To advance progress in Big Data, the NIST Big Data Public Working Group (NBD-PWG) is working to develop consensus on important fundamental concepts related to Big Data. The results are reported in the NIST Big Data Interoperability Framework series of volumes. This volume, Volume 3, contains the original 51 Version 1 use cases gathered by the NBD-PWG Use Cases and Requirements Subgroup and the requirements generated from those use cases. The use cases are presented in their original and summarized form. Requirements, or challenges, were extracted from each use case, and then summarized over all the use cases. These generalized requirements were used in the development of the NIST Big Data Reference Architecture (NBDRA), which is presented in Volume 6. Currently, the subgroup is accepting additional use case submissions using the more detailed Use Case Template 2. The Use Case Template 2 and the two Version 2 use cases collected to date are presented and summarized in this volume.},
 bibtype = {article},
 author = {Subgroup, NIST Big Data Public Working Group: Use Cases and Requirements},
 doi = {http://dx.doi.org/10.6028/NIST.SP.1500-3},
 number = {June}
}

Big Data is a term used to describe the large amount of data in the networked, digitized, sensor-laden, information-driven world. While opportunities exist with Big Data, the data can overwhelm traditional technical approaches and the growth of data is outpacing scientific and technological advances in data analytics. To advance progress in Big Data, the NIST Big Data Public Working Group (NBD-PWG) is working to develop consensus on important fundamental concepts related to Big Data. The results are reported in the NIST Big Data Interoperability Framework series of volumes. This volume, Volume 3, contains the original 51 Version 1 use cases gathered by the NBD-PWG Use Cases and Requirements Subgroup and the requirements generated from those use cases. The use cases are presented in their original and summarized form. Requirements, or challenges, were extracted from each use case, and then summarized over all the use cases. These generalized requirements were used in the development of the NIST Big Data Reference Architecture (NBDRA), which is presented in Volume 6. Currently, the subgroup is accepting additional use case submissions using the more detailed Use Case Template 2. The Use Case Template 2 and the two Version 2 use cases collected to date are presented and summarized in this volume.

@inproceedings{
 title = {Peer comparison of XSEDE and NCAR publication data},
 type = {inproceedings},
 year = {2015},
 volume = {2015-Octob},
 id = {f456a391-914e-392f-a047-073a1c3ea3a7},
 created = {2018-02-27T18:07:43.102Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.102Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© 2015 IEEE. We present a framework that compares the publication impact based on a comprehensive peer analysis of papers produced by scientists using XSEDE and NCAR resources. The analysis is introducing a percentile ranking based approach of citations of the XSEDE and NCAR papers compared to peer publications in the same journal that do not use these resources. This analysis is unique in that it evaluates the impact of the two facilities by comparing the reported publications from them to their peers from within the same journal issue. From this analysis, we can see that papers that utilize XSEDE and NCAR resources are cited statistically significantly more often. Hence we find that reported publications indicate that XSEDE and NCAR resources exert a strong positive impact on scientific research.},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Wang, F. and Fox, G.C. and Hart, D.L. and Furlani, T.R. and Deleon, R.L. and Gallo, S.M.},
 doi = {10.1109/CLUSTER.2015.98},
 booktitle = {Proceedings - IEEE International Conference on Cluster Computing, ICCC}
}

© 2015 IEEE. We present a framework that compares the publication impact based on a comprehensive peer analysis of papers produced by scientists using XSEDE and NCAR resources. The analysis is introducing a percentile ranking based approach of citations of the XSEDE and NCAR papers compared to peer publications in the same journal that do not use these resources. This analysis is unique in that it evaluates the impact of the two facilities by comparing the reported publications from them to their peers from within the same journal issue. From this analysis, we can see that papers that utilize XSEDE and NCAR resources are cited statistically significantly more often. Hence we find that reported publications indicate that XSEDE and NCAR resources exert a strong positive impact on scientific research.

@inproceedings{
 title = {TAS view of XSEDE users and usage},
 type = {inproceedings},
 year = {2015},
 volume = {2015-July},
 id = {95c675ea-0819-3b42-96bf-2c6bdb2e2eef},
 created = {2018-02-27T18:07:43.108Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.108Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {© 2015 ACM. The Technology Audit Service has developed, XDMoD, a resource management tool. This paper utilizes XDMoD and the XDMoD data warehouse that it draws from to provide a broad overview of several aspects of XSEDE users and their usage. Some important trends include: 1) in spite of a large yearly turnover, there is a core of users persisting over many years, 2) user job submission has changed from primarily faculty members to students and postdocs, 3) increases in usage in Molecular Biosciences and Materials Research has outstripped that of other fields of science, 4) the distribution of user external funding is bimodal with one group having a large ratio of external funding to internal XSEDE funding (ie, CPU cycles) and a second group having a small ratio of external to internal (CPU cycle) funding, 5) user job efficiency is also bimodal with a group of presumably new users running mainly small inefficient jobs and another group of users running larger more efficient jobs, 6) finally, based on an analysis of citations of published papers, the scientific impact of XSEDE coupled with the service providers is demonstrated in the statistically significant advantage it provides to the research of its users.},
 bibtype = {inproceedings},
 author = {DeLeon, R.L. and Furlani, T.R. and Gallo, S.M. and White, J.P. and Jones, M.D. and Patra, A. and Innus, M. and Yearke, T. and Palmer, J.T. and Sperhac, J.M. and Rathsam, R. and Simakov, N. and Von Laszewski, G. and Wang, F.},
 doi = {10.1145/2792745.2792766},
 booktitle = {ACM International Conference Proceeding Series}
}

© 2015 ACM. The Technology Audit Service has developed, XDMoD, a resource management tool. This paper utilizes XDMoD and the XDMoD data warehouse that it draws from to provide a broad overview of several aspects of XSEDE users and their usage. Some important trends include: 1) in spite of a large yearly turnover, there is a core of users persisting over many years, 2) user job submission has changed from primarily faculty members to students and postdocs, 3) increases in usage in Molecular Biosciences and Materials Research has outstripped that of other fields of science, 4) the distribution of user external funding is bimodal with one group having a large ratio of external funding to internal XSEDE funding (ie, CPU cycles) and a second group having a small ratio of external to internal (CPU cycle) funding, 5) user job efficiency is also bimodal with a group of presumably new users running mainly small inefficient jobs and another group of users running larger more efficient jobs, 6) finally, based on an analysis of citations of published papers, the scientific impact of XSEDE coupled with the service providers is demonstrated in the statistically significant advantage it provides to the research of its users.

@techreport{
 title = {Scalability Analysis of the Multi-Look Time Domain Processor on XSEDE Compute Resources},
 type = {techreport},
 year = {2015},
 id = {30b921c4-3b1c-38fe-a1fd-4bbffe099d26},
 created = {2018-02-27T18:07:45.563Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.563Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {techreport},
 author = {von Laszewski, Gregor and Wang, Fugang and Fox, Geoffrey C and Li, Jilu and Paden, John}
}

@article{
 title = {NIST Special Publication 1500-6 : NIST Big Data Interoperability Framework - Reference Architecture},
 type = {article},
 year = {2015},
 keywords = {Adoption,barriers,implementation,interfaces,market maturity,organizational maturity,project maturity,system modernization.},
 pages = {1-62},
 volume = {6},
 id = {a3700b76-affd-3b91-b8f5-3f86be66c7bb},
 created = {2018-08-09T16:38:14.534Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-08-09T16:38:14.534Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {A341Big Data is a term used to describe the large amount of data in the networked, digitized, sensor-laden, information-driven world. While opportunities exist with Big Data, the data can overwhelm traditional technical approaches and the growth of data is outpacing scientific and technological advances in data analytics. To advance progress in Big Data, the NIST Big Data Public Working Group (NBD-PWG) is working to develop consensus on important, fundamental concepts related to Big Data. The results are reported in the NIST Big Data Interoperability Framework series of volumes. This volume, Volume 6, summarizes the work performed by the NBD-PWG to characterize Big Data from an architecture perspective, presents the NIST Big Data Reference Architecture (NBDRA) conceptual model, and discusses the components and fabrics of the NBDRA.},
 bibtype = {article},
 author = {(NBD-PWG), NIST Big Data Public Working Group},
 doi = {10.6028/NIST.SP.1500-6},
 number = {June}
}

A341Big Data is a term used to describe the large amount of data in the networked, digitized, sensor-laden, information-driven world. While opportunities exist with Big Data, the data can overwhelm traditional technical approaches and the growth of data is outpacing scientific and technological advances in data analytics. To advance progress in Big Data, the NIST Big Data Public Working Group (NBD-PWG) is working to develop consensus on important, fundamental concepts related to Big Data. The results are reported in the NIST Big Data Interoperability Framework series of volumes. This volume, Volume 6, summarizes the work performed by the NBD-PWG to characterize Big Data from an architecture perspective, presents the NIST Big Data Reference Architecture (NBDRA) conceptual model, and discusses the components and fabrics of the NBDRA.

@techreport{
 title = {Pervasive Technology Institute annual report: Research innovations and advanced cyberinfrastructure services in support of IU strategic goals during FY 2015},
 type = {techreport},
 year = {2015},
 keywords = {CACR,D2I,DSC,NCGAS,PTI,RT,Technical Report,advanced cyberinfrastructure,engagement,outreach,research,storage,students},
 websites = {http://hdl.handle.net/2022/20566},
 id = {6ed4d646-f4ef-3f3d-9d07-6bb1c7049729},
 created = {2020-09-10T14:25:38.828Z},
 accessed = {2020-09-10},
 file_attached = {true},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-09-10T16:51:14.671Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 bibtype = {techreport},
 author = {Stewart, Craig A.; Plale, Beth; Welch, Von; Link, Matthew R.; Miller, Therese; Wernert, Eric A.; Boyles, Michael J.; Fulton, Ben; Hancock, David Y.; Henschel, Robert; Michael, Scott A.; Pierce, Marlon; Ping, Robert J.; Gniady, Tassie; Fox, Geoffrey C.; Mi, Gary;}
}

@techreport{
 title = {Indiana University’s advanced cyberinfrastructure in service of IU strategic goals: Activities of the Research Technologies Division of UITS and National Center for Genome Analysis Support – two Pervasive Technology Institute cyberinfrastructure and servi},
 type = {techreport},
 year = {2015},
 keywords = {ABITC,Clinical Affairs Schools,IUSM,NCGAS,PTI,advanced cyberinfrastructure,digital collections,engagement,health sciences,research,storage,students},
 websites = {https://scholarworks.iu.edu/dspace/handle/2022/19805},
 publisher = {Indiana University},
 id = {bc79e6b3-1f18-3a6c-9190-eb3607ab350f},
 created = {2020-09-10T17:46:58.547Z},
 accessed = {2020-09-10},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-09-10T17:46:58.547Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 bibtype = {techreport},
 author = {Stewart, Craig A.; and Plale, Beth; and Welch, Von; and Fox, Geoffrey C.; and Link, Matthew R.; and Miller, Therese; and Wernert, Eric A.; and Boyles, Michael J.; and Fulton, Ben; and Hancock, David Y.; Henschel, Robert; and Michael, Scott A.; and Pierce, Marlon; and Ping, Robert J.; and Miksik, Gary; and Gniady, Tassie;}
}

@article{
 title = {Comprehensive, open-source resource usage measurement and analysis for HPC systems},
 type = {article},
 year = {2014},
 volume = {26},
 id = {7dddfe6c-2289-304d-af52-c2de1e59407d},
 created = {2018-02-27T18:07:43.115Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.115Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {The important role high-performance computing (HPC) resources play in science and engineering research, coupled with its high cost (capital, power and manpower), short life and oversubscription, requires us to optimize its usage - an outcome that is only possible if adequate analytical data are collected and used to drive systems management at different granularities - job, application, user and system. This paper presents a method for comprehensive job, application and system-level resource use measurement, and analysis and its implementation. The steps in the method are system-wide collection of comprehensive resource use and performance statistics at the job and node levels in a uniform format across all resources, mapping and storage of the resultant job-wise data to a relational database, which enables further implementation and transformation of the data to the formats required by specific statistical and analytical algorithms. Analyses can be carried out at different levels of granularity: job, user, application or syste m-wide. Measurements are based on a new lightweight job-centric measurement tool 'TACC-Stats', which gathers a comprehensive set of resource use metrics on all compute nodes and data logged by the system scheduler. The data mapping and analysis tools are an extension of the XDMoD project. The method is illustrated with analyses of resource use for the Texas Advanced Computing Center's Lonestar4, Ranger and Stampede supercomputers and the HPC cluster at the Center for Computational Research. The illustrations are focused on resource use at the system, job and application levels and reveal many interesting insights into system usage patterns and also anomalous behavior due to failure/misuse. The method can be applied to any system that runs the TACC-Stats measurement tool and a tool to extract job execution environment data from the system scheduler. Copyright © 2014 John Wiley  &  Sons, Ltd.},
 bibtype = {article},
 author = {Browne, J.C. and Deleon, R.L. and Patra, A.K. and Barth, W.L. and Hammond, J. and Jones, M.D. and Furlani, T.R. and Schneider, B.I. and Gallo, S.M. and Ghadersohi, A. and Gentner, R.J. and Palmer, J.T. and Simakov, N. and Innus, M. and Bruno, A.E. and White, J.P. and Cornelius, C.D. and Yearke, T. and Marcus, K. and Von Laszewski, G. and Wang, F.},
 doi = {10.1002/cpe.3245},
 journal = {Concurrency Computation Practice and Experience},
 number = {13}
}

The important role high-performance computing (HPC) resources play in science and engineering research, coupled with its high cost (capital, power and manpower), short life and oversubscription, requires us to optimize its usage - an outcome that is only possible if adequate analytical data are collected and used to drive systems management at different granularities - job, application, user and system. This paper presents a method for comprehensive job, application and system-level resource use measurement, and analysis and its implementation. The steps in the method are system-wide collection of comprehensive resource use and performance statistics at the job and node levels in a uniform format across all resources, mapping and storage of the resultant job-wise data to a relational database, which enables further implementation and transformation of the data to the formats required by specific statistical and analytical algorithms. Analyses can be carried out at different levels of granularity: job, user, application or syste m-wide. Measurements are based on a new lightweight job-centric measurement tool 'TACC-Stats', which gathers a comprehensive set of resource use metrics on all compute nodes and data logged by the system scheduler. The data mapping and analysis tools are an extension of the XDMoD project. The method is illustrated with analyses of resource use for the Texas Advanced Computing Center's Lonestar4, Ranger and Stampede supercomputers and the HPC cluster at the Center for Computational Research. The illustrations are focused on resource use at the system, job and application levels and reveal many interesting insights into system usage patterns and also anomalous behavior due to failure/misuse. The method can be applied to any system that runs the TACC-Stats measurement tool and a tool to extract job execution environment data from the system scheduler. Copyright © 2014 John Wiley & Sons, Ltd.

@techreport{
 title = {Towards understanding cloud usage through resource allocation analysis on xsede},
 type = {techreport},
 year = {2014},
 publisher = {Community Grids Lab Publications},
 id = {23758b18-3510-3bce-87a1-454406b77105},
 created = {2018-02-27T18:07:43.254Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.254Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {RPRT},
 private_publication = {false},
 bibtype = {techreport},
 author = {Lee, Hyungro and von Laszewski, Gregor and Wang, Fugang and Fox, Geoffrey C}
}

@inproceedings{
 title = {Towards a scientific impact measuring framework for large computing facilities - A case study on XSEDE},
 type = {inproceedings},
 year = {2014},
 id = {88e14f27-482c-3177-a78a-75d07ee784c3},
 created = {2018-02-27T18:07:43.621Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.621Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {We present a framework that (a) integrates publication and citation data retrieval, (b) allows scientific impact metrics generation at different aggregation levels, and (c) provides correlation analysis of impact metrics based on publication and citation data with resource allocation for a computing facility. Furthermore, we use this framework to conduct a scientific impact metrics evaluation of XSEDE.We carry out an extensive statistical analysis correlating XSEDE alloca-tion size to the impact metrics aggregated by project and field of science. This analysis not only helps to provide an indication of XSEDE's scientific impact, but also provides insight regarding maximizing the return on investment in terms of allocation by taking into account the field of sci-ence or project based impact metrics. The findings from this analysis can be utilized by the XSEDE resource allocation committee to help assess and identify projects with higher scientific impact. It can also help provide metrics regard-ing the return on investment for XSEDE resources, or other institutional or campus resources for which an analysis of impact based on publications is important. Copyright 2014 ACM.},
 bibtype = {inproceedings},
 author = {Wang, F. and Von Laszewski, G. and Fox, G.C. and Furlani, T.R. and DeLeon, R.L. and Gallo, S.M.},
 doi = {10.1145/2616498.2616507},
 booktitle = {ACM International Conference Proceeding Series}
}

We present a framework that (a) integrates publication and citation data retrieval, (b) allows scientific impact metrics generation at different aggregation levels, and (c) provides correlation analysis of impact metrics based on publication and citation data with resource allocation for a computing facility. Furthermore, we use this framework to conduct a scientific impact metrics evaluation of XSEDE.We carry out an extensive statistical analysis correlating XSEDE alloca-tion size to the impact metrics aggregated by project and field of science. This analysis not only helps to provide an indication of XSEDE's scientific impact, but also provides insight regarding maximizing the return on investment in terms of allocation by taking into account the field of sci-ence or project based impact metrics. The findings from this analysis can be utilized by the XSEDE resource allocation committee to help assess and identify projects with higher scientific impact. It can also help provide metrics regard-ing the return on investment for XSEDE resources, or other institutional or campus resources for which an analysis of impact based on publications is important. Copyright 2014 ACM.

@inproceedings{
 title = {Accessing multiple clouds with Cloudmesh},
 type = {inproceedings},
 year = {2014},
 id = {68eb4f51-3b4b-325f-9189-687be036f77b},
 created = {2018-02-27T18:07:43.708Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.708Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {We present the design of a toolkit that can be employed by users and administrators to manage virtual machines on multi-cloud environments. It can be run by individual users or offered as a service to a shared user community. We have practically demonstrated its use as part of a Future-Grid service, allowing users of FutureGrid to utilize such a service. Furthermore, we discuss implications and solutions for a unified metrics system assisting the users to find and utilize resources appropriate for their applications. Lastly, we discuss how to move such a multi-cloud environment forward by integrating clouds managed by the community or are offered as public clouds. This includes the introduction of a mutual trust agreement on a user and project basis. We have developed a number of components that support the creation of such a multi-cloud environment. This system is known as Cloudmesh and has been used in practice to achieve virtual machine management in multiple clouds. An important distinguishing factor of Cloudmesh is that it also allows the use of bare metal provisioning for supporting service providers and authorized users, offering services beyond those available by typical clouds. Copyright 2014 ACM.},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Wang, F. and Lee, H. and Chen, H. and Fox, G.C.},
 doi = {10.1145/2609441.2609638},
 booktitle = {BigSystem 2014 - Proceedings of the 2014 ACM International Workshop on Software-Defined Ecosystems, Co-located with HPDC 2014}
}

We present the design of a toolkit that can be employed by users and administrators to manage virtual machines on multi-cloud environments. It can be run by individual users or offered as a service to a shared user community. We have practically demonstrated its use as part of a Future-Grid service, allowing users of FutureGrid to utilize such a service. Furthermore, we discuss implications and solutions for a unified metrics system assisting the users to find and utilize resources appropriate for their applications. Lastly, we discuss how to move such a multi-cloud environment forward by integrating clouds managed by the community or are offered as public clouds. This includes the introduction of a mutual trust agreement on a user and project basis. We have developed a number of components that support the creation of such a multi-cloud environment. This system is known as Cloudmesh and has been used in practice to achieve virtual machine management in multiple clouds. An important distinguishing factor of Cloudmesh is that it also allows the use of bare metal provisioning for supporting service providers and authorized users, offering services beyond those available by typical clouds. Copyright 2014 ACM.

@article{
 title = {DDDAS-Parallel Simulation of Threat Management in Urban Water Distribution Systems for Cloud Computing},
 type = {article},
 year = {2014},
 volume = {16},
 id = {1af614dd-0b4b-31de-a79d-72c8e9db487c},
 created = {2018-02-27T18:07:44.440Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.440Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and von Laszewski, Gregor},
 doi = {10.1109/MCSE.2012.89},
 journal = {Computing in Science & Engineering},
 number = {1}
}

@inbook{
 type = {inbook},
 year = {2014},
 pages = {27-59},
 publisher = {Springer},
 id = {7468a1a9-1995-341e-b3a4-996591ee8fcd},
 created = {2018-02-27T18:07:45.001Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.001Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CHAP},
 private_publication = {false},
 bibtype = {inbook},
 author = {Von Laszewski, Gregor and Fox, Geoffrey C},
 chapter = {The FutureGrid Testbed for Big Data},
 title = {Cloud Computing for Data-Intensive Applications}
}

@article{
 title = {Performance metrics and auditing framework using application kernels for high-performance computer systems},
 type = {article},
 year = {2013},
 volume = {25},
 id = {eee0e9c2-eee7-327a-9de1-37ff671de8c6},
 created = {2018-02-27T18:07:43.490Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.490Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {This paper describes XSEDE Metrics on Demand, a comprehensive auditing framework for use by high-performance computing centers, which provides metrics regarding resource utilization, resource performance, and impact on scholarship and research. This role-based framework is designed to meet the following objectives: (1) provide the user community with a tool to manage their allocations and optimize their resource utilization; (2) provide operational staff with the ability to monitor and tune resource performance; (3) provide management with a tool to monitor utilization, user base, and performance of resources; and (4) provide metrics to help measure scientific impact. Although initially focused on the XSEDE program, XSEDE Metrics on Demand can be adapted to any high-performance computing environment. The framework includes a computationally lightweight application kernel auditing system that utilizes performance kernels to measure overall system performance. This allows continuous resource auditing to measure all aspects of system performance including filesystem performance, processor and memory performance, and network latency and bandwidth. Metrics that focus on scientific impact, such as publications, citations and external funding, will be included to help quantify the important role high-performance computing centers play in advancing research and scholarship. Copyright © 2012 John Wiley  &  Sons, Ltd.},
 bibtype = {article},
 author = {Furlani, T.R. and Jones, M.D. and Gallo, S.M. and Bruno, A.E. and Lu, C.-D. and Ghadersohi, A. and Gentner, R.J. and Patra, A. and Deleon, R.L. and Von Laszewski, G. and Wang, F. and Zimmerman, A.},
 doi = {10.1002/cpe.2871},
 journal = {Concurrency Computation Practice and Experience},
 number = {7}
}

This paper describes XSEDE Metrics on Demand, a comprehensive auditing framework for use by high-performance computing centers, which provides metrics regarding resource utilization, resource performance, and impact on scholarship and research. This role-based framework is designed to meet the following objectives: (1) provide the user community with a tool to manage their allocations and optimize their resource utilization; (2) provide operational staff with the ability to monitor and tune resource performance; (3) provide management with a tool to monitor utilization, user base, and performance of resources; and (4) provide metrics to help measure scientific impact. Although initially focused on the XSEDE program, XSEDE Metrics on Demand can be adapted to any high-performance computing environment. The framework includes a computationally lightweight application kernel auditing system that utilizes performance kernels to measure overall system performance. This allows continuous resource auditing to measure all aspects of system performance including filesystem performance, processor and memory performance, and network latency and bandwidth. Metrics that focus on scientific impact, such as publications, citations and external funding, will be included to help quantify the important role high-performance computing centers play in advancing research and scholarship. Copyright © 2012 John Wiley & Sons, Ltd.

@inproceedings{
 title = {Co-processing SPMD computation on CPUs and GPUs cluster},
 type = {inproceedings},
 year = {2013},
 id = {7f314363-abf0-3705-b9e8-387c89940c26},
 created = {2018-02-27T18:07:44.383Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.383Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Heterogeneous parallel systems with multi processors and accelerators are becoming ubiquitous due to better cost-performance and energy-efficiency. These heterogeneous processor architectures have different instruction sets and are optimized for either task-latency or throughput purposes. Challenges occur in regard to programmability and performance when running SPMD tasks on heterogeneous devices. In order to meet these challenges, we implemented a parallel runtime system that used to co-process SPMD computation on CPUs and GPUs clusters. Furthermore, we are proposing an analytic model to automatically schedule SPMD tasks on heterogeneous clusters. Our analytic model is derived from the roofline model, and therefore it can be applied to a wider range of SPMD applications and hardware devices. The experimental results of the C-means, GMM, and GEMV show good speedup in practical heterogeneous cluster environments. © 2013 IEEE.},
 bibtype = {inproceedings},
 author = {Li, H. and Fox, G. and Von Laszewski, G. and Chauhan, A.},
 doi = {10.1109/CLUSTER.2013.6702632},
 booktitle = {Proceedings - IEEE International Conference on Cluster Computing, ICCC}
}

Heterogeneous parallel systems with multi processors and accelerators are becoming ubiquitous due to better cost-performance and energy-efficiency. These heterogeneous processor architectures have different instruction sets and are optimized for either task-latency or throughput purposes. Challenges occur in regard to programmability and performance when running SPMD tasks on heterogeneous devices. In order to meet these challenges, we implemented a parallel runtime system that used to co-process SPMD computation on CPUs and GPUs clusters. Furthermore, we are proposing an analytic model to automatically schedule SPMD tasks on heterogeneous clusters. Our analytic model is derived from the roofline model, and therefore it can be applied to a wider range of SPMD applications and hardware devices. The experimental results of the C-means, GMM, and GEMV show good speedup in practical heterogeneous cluster environments. © 2013 IEEE.

@inproceedings{
 title = {Using XDMoD to facilitate XSEDE operations, planning and analysis},
 type = {inproceedings},
 year = {2013},
 id = {99c0a7e4-8148-3e09-89b6-83db78d99c14},
 created = {2018-02-27T18:07:45.034Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.034Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {The XDMoD auditing tool provides, for the first time, a comprehensive tool to measure both utilization and performance of high-end cyberinfrastructure (CI), with initial focus on XSEDE. Here, we demonstrate, through several case studies, its utility for providing important metrics regarding resource utilization and performance of TeraGrid/XSEDE that can be used for detailed analysis and planning as well as improving operational efficiency and performance. Measuring the utilization of high-end cyberinfrastructure such as XSEDE helps provide a detailed understanding of how a given CI resource is being utilized and can lead to improved performance of the resource in terms of job throughput or any number of desired job characteristics. In the case studies considered here, a detailed historical analysis of XSEDE usage data using XDMoD clearly demonstrates the tremendous growth in the number of users, overall usage, and scale of the simulations routinely carried out. Not surprisingly, physics, chemistry, and the engineering disciplines are shown to be heavy users of the resources. However, as the data clearly show, molecular biosciences are now a significant and growing user of XSEDE resources, accounting for more than 20 percent of all SUs consumed in 2012. XDMoD shows that the resources required by the various scientific disciplines are very different. Physics, Astronomical sciences, and Atmospheric sciences tend to solve large problems requiring many cores. Molecular biosciences applications on the other hand, require many cycles but do not employ core counts that are as large. Such distinctions are important in guiding future cyberinfrastructure design decisions. XDMoD's implementation of a novel application kernel-based auditing system to measure overall CI system performance and quality of service is shown, through several examples, to provide a useful means to automatically detect under performing hardware and software. This capability is especially critical given the complex composition of today's advanced CI. Examples include an application kernel based on a widely used quantum chemistry program that uncovered a software bug in the I/O stack of a commercial parallel file system, which was subsequently fixed by the vendor in the form of a software patch that is now part of their standard release. This error, which resulted in dramatically increased execution times as well as outright job failure, would likely have gone unnoticed for sometime and was only uncovered as a result of implementation of XDMoD's suite of application kernels. © 2013 by the Association for Computing Machinery, Inc.},
 bibtype = {inproceedings},
 author = {Furlani, T.R. and Schneider, B.L. and Jones, M.D. and Towns, J. and Hart, D.L. and Gallo, S.M. and Deleon, R.L. and Lu, C.-D. and Ghadersohi, A. and Gentner, R.J. and Patra, A.K. and Laszewski, G.V. and Wang, F. and Palmer, J.T. and Simakov, N.},
 doi = {10.1145/2484762.2484763},
 booktitle = {ACM International Conference Proceeding Series}
}

The XDMoD auditing tool provides, for the first time, a comprehensive tool to measure both utilization and performance of high-end cyberinfrastructure (CI), with initial focus on XSEDE. Here, we demonstrate, through several case studies, its utility for providing important metrics regarding resource utilization and performance of TeraGrid/XSEDE that can be used for detailed analysis and planning as well as improving operational efficiency and performance. Measuring the utilization of high-end cyberinfrastructure such as XSEDE helps provide a detailed understanding of how a given CI resource is being utilized and can lead to improved performance of the resource in terms of job throughput or any number of desired job characteristics. In the case studies considered here, a detailed historical analysis of XSEDE usage data using XDMoD clearly demonstrates the tremendous growth in the number of users, overall usage, and scale of the simulations routinely carried out. Not surprisingly, physics, chemistry, and the engineering disciplines are shown to be heavy users of the resources. However, as the data clearly show, molecular biosciences are now a significant and growing user of XSEDE resources, accounting for more than 20 percent of all SUs consumed in 2012. XDMoD shows that the resources required by the various scientific disciplines are very different. Physics, Astronomical sciences, and Atmospheric sciences tend to solve large problems requiring many cores. Molecular biosciences applications on the other hand, require many cycles but do not employ core counts that are as large. Such distinctions are important in guiding future cyberinfrastructure design decisions. XDMoD's implementation of a novel application kernel-based auditing system to measure overall CI system performance and quality of service is shown, through several examples, to provide a useful means to automatically detect under performing hardware and software. This capability is especially critical given the complex composition of today's advanced CI. Examples include an application kernel based on a widely used quantum chemistry program that uncovered a software bug in the I/O stack of a commercial parallel file system, which was subsequently fixed by the vendor in the form of a software patch that is now part of their standard release. This error, which resulted in dramatically increased execution times as well as outright job failure, would likely have gone unnoticed for sometime and was only uncovered as a result of implementation of XDMoD's suite of application kernels. © 2013 by the Association for Computing Machinery, Inc.

@article{
 title = {On-demand service hosting on production grid infrastructures},
 type = {article},
 year = {2013},
 volume = {66},
 id = {1c6d614f-c60c-3728-b61e-88359b730034},
 created = {2018-02-27T18:07:45.484Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.484Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {The Software as a Service (SaaS) methodology is a key paradigm of Cloud computing. In this paper, we focus on an interesting topic - to dynamically host services on existing production Grid infrastructures. In general, production Grids normally employ a Job-Submission-Execution (JSE) model with rigid access interfaces. In this paper, we implement the Cyberaide onServe, a lightweight middleware with a virtual appliance. The Cyberaide onServe implements the SaaS model on production Grids by translating the SaaS model to the JSE model. The Cyberaide onServe can be deployed on demand in a virtual appliance, host users' software as a Web service, accept Web service invocations; finally, the Cyberaide onServe can execute them on production Grids. We have deployed the Cyberaide onServe on the TeraGrid and the test results show that the Cyberaide onServe can provide SaaS functionalities with a good performance. © 2011 Springer Science+Business Media, LLC.},
 bibtype = {article},
 author = {Wang, L. and Kurze, T. and Tao, J. and Kunze, M. and Von Laszewski, G.},
 doi = {10.1007/s11227-011-0666-5},
 journal = {Journal of Supercomputing},
 number = {3}
}

The Software as a Service (SaaS) methodology is a key paradigm of Cloud computing. In this paper, we focus on an interesting topic - to dynamically host services on existing production Grid infrastructures. In general, production Grids normally employ a Job-Submission-Execution (JSE) model with rigid access interfaces. In this paper, we implement the Cyberaide onServe, a lightweight middleware with a virtual appliance. The Cyberaide onServe implements the SaaS model on production Grids by translating the SaaS model to the JSE model. The Cyberaide onServe can be deployed on demand in a virtual appliance, host users' software as a Web service, accept Web service invocations; finally, the Cyberaide onServe can execute them on production Grids. We have deployed the Cyberaide onServe on the TeraGrid and the test results show that the Cyberaide onServe can provide SaaS functionalities with a good performance. © 2011 Springer Science+Business Media, LLC.

@inproceedings{
 title = {Abstract image management and universal image registration for cloud and HPC infrastructures},
 type = {inproceedings},
 year = {2012},
 id = {9a0c14a5-f96e-3e5b-bbbd-3263ff38e350},
 created = {2018-02-27T18:07:43.693Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.693Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {Cloud computing has become an important driver for delivering infrastructure as a service (IaaS) to users with on-demand requests for customized environments and sophisticated software stacks. Within the FutureGrid (FG) project, we offer different IaaS frameworks as well as high performance computing infrastructures by allowing users to explore them as part of the FG testbed. To ease the use of these infrastructures, as part of performance experiments, we have designed an image management framework, which allows us to create user defined software stacks based on abstract image management and uniform image registration. Consequently, users can create their own customized environments very easily. The complex processes of the underlying infrastructures are managed by our sophisticated software tools and services. Besides being able to manage images for IaaS frameworks, we also allow the registration and deployment of images onto bare-metal by the user. This level of functionality is typically not offered in a HPC (high performance computing) infrastructure. However, our approach provides users with the ability to create their own environments changing the paradigm of administrator-controlled dynamic provisioning to user-controlled dynamic provisioning, which we also call raining. Thus, users obtain access to a testbed with the ability to manage state-of-the-art software stacks that would otherwise not be supported in typical compute centers. Security is also considered by vetting images before they are registered in a infrastructure. In this paper, we present the design of our image management framework and evaluate two of its major components. This includes the image creation and image registration. Our design and implementation can support the current FG user community interested in such capabilities. © 2012 IEEE.},
 bibtype = {inproceedings},
 author = {Diaz, J. and Von Laszewski, G. and Wang, F. and Fox, G.},
 doi = {10.1109/CLOUD.2012.94},
 booktitle = {Proceedings - 2012 IEEE 5th International Conference on Cloud Computing, CLOUD 2012}
}

Cloud computing has become an important driver for delivering infrastructure as a service (IaaS) to users with on-demand requests for customized environments and sophisticated software stacks. Within the FutureGrid (FG) project, we offer different IaaS frameworks as well as high performance computing infrastructures by allowing users to explore them as part of the FG testbed. To ease the use of these infrastructures, as part of performance experiments, we have designed an image management framework, which allows us to create user defined software stacks based on abstract image management and uniform image registration. Consequently, users can create their own customized environments very easily. The complex processes of the underlying infrastructures are managed by our sophisticated software tools and services. Besides being able to manage images for IaaS frameworks, we also allow the registration and deployment of images onto bare-metal by the user. This level of functionality is typically not offered in a HPC (high performance computing) infrastructure. However, our approach provides users with the ability to create their own environments changing the paradigm of administrator-controlled dynamic provisioning to user-controlled dynamic provisioning, which we also call raining. Thus, users obtain access to a testbed with the ability to manage state-of-the-art software stacks that would otherwise not be supported in typical compute centers. Security is also considered by vetting images before they are registered in a infrastructure. In this paper, we present the design of our image management framework and evaluate two of its major components. This includes the image creation and image registration. Our design and implementation can support the current FG user community interested in such capabilities. © 2012 IEEE.

@inproceedings{
 title = {Data analytics driven cyberinfrastructure operations, planning and analysis using XDMoD},
 type = {inproceedings},
 year = {2012},
 id = {7595db9c-e419-361a-825f-8ac0f9b97c94},
 created = {2018-02-27T18:07:43.795Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.795Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Furlani, Thomas R and Schneider, Barry I and Jones, Matthew D and Towns, John and Hart, David L and Patra, Abani K and DeLeon, Robert L and Gallo, Steven M and Lu, Charng-Da and Ghadersohi, Amin},
 booktitle = {submitted SC12 Conference, Salt Lake City, Utah}
}

@inproceedings{
 title = {Qualitative Comparison of Multiple Cloud Frameworks},
 type = {inproceedings},
 year = {2012},
 pages = {734-741},
 id = {165fe13d-774d-3c4f-8011-6b3a9e8c37c6},
 created = {2018-02-27T18:07:43.800Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.800Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 abstract = {Today, many cloud Infrastructure as a Service(IaaS) frameworks exist. Users, developers, and administrators have to make a decision about which environment is best suited for them. Unfortunately, the comparison of such frameworks is difficult because either users do not have access to all of them or they are comparing the performance of such systems on different resources, which make it difficult to obtain objective comparisons. Hence, the community benefits from the availability of a testbed on which comparisons between the IaaS frameworks can be conducted. FutureGrid aims to offer a number of IaaS including Nimbus, Eucalyptus, OpenStack, and OpenNebula. One of the important features that FutureGrid provides is not only the ability to compare between IaaS frameworks, but also to compare them in regards to bare-metal and traditional high performance computing services. In this paper, we outline some of our initial findings by providing such a testbed. As one of our conclusions, we also present our work on making access to the various infrastructures on FutureGrid easier. © 2012 IEEE.},
 bibtype = {inproceedings},
 author = {Von Laszewski, Gregor and Diaz, Javier and Wang, Fugang and Fox, G.C. Geoffrey C},
 doi = {10.1109/CLOUD.2012.104},
 booktitle = {Proceedings: 2012 IEEE 5th international conference on cloud computing}
}

Today, many cloud Infrastructure as a Service(IaaS) frameworks exist. Users, developers, and administrators have to make a decision about which environment is best suited for them. Unfortunately, the comparison of such frameworks is difficult because either users do not have access to all of them or they are comparing the performance of such systems on different resources, which make it difficult to obtain objective comparisons. Hence, the community benefits from the availability of a testbed on which comparisons between the IaaS frameworks can be conducted. FutureGrid aims to offer a number of IaaS including Nimbus, Eucalyptus, OpenStack, and OpenNebula. One of the important features that FutureGrid provides is not only the ability to compare between IaaS frameworks, but also to compare them in regards to bare-metal and traditional high performance computing services. In this paper, we outline some of our initial findings by providing such a testbed. As one of our conclusions, we also present our work on making access to the various infrastructures on FutureGrid easier. © 2012 IEEE.

@inproceedings{
 title = {Design of an accounting and metric-based cloud-shifting and cloud-seeding framework for federated clouds and bare-metal environments},
 type = {inproceedings},
 year = {2012},
 id = {287621b5-bc58-3423-9c2d-2576805b78e3},
 created = {2018-02-27T18:07:43.867Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.867Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {We present the design of a dynamic provisioning system that is able to manage the resources of a federated cloud environment by focusing on their utilization. With our framework, it is not only possible to allocate resources at a particular time to a specific Infrastructure as a Service framework, but also to utilize them as part of a typical HPC environment controlled by batch queuing systems. Through this interplay between virtualized and non-virtualized resources, we provide a flexible resource management framework that can be adapted based on users' demands. The need for such a framework is motivated by real user data gathered during our operation of FutureGrid (FG). We observed that the usage of the different infrastructures vary over time changing from being over-utilized to underutilize and vice versa. Therefore, the proposed framework will be beneficial for users of environments such a FutureGrid where several infrastructures are supported with limited physical resources. Copyright 2012 ACM.},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Lee, H. and Diaz, J. and Wang, F. and Tanaka, K. and Karavinkoppa, S. and Fox, G.C. and Furlani, T.},
 doi = {10.1145/2378975.2378982},
 booktitle = {FederatedClouds'12 - Proceedings of the 2012 Workshop on Cloud Services, Federation, and the 8th Open Cirrus Summit, Co-located with ICAC'12}
}

We present the design of a dynamic provisioning system that is able to manage the resources of a federated cloud environment by focusing on their utilization. With our framework, it is not only possible to allocate resources at a particular time to a specific Infrastructure as a Service framework, but also to utilize them as part of a typical HPC environment controlled by batch queuing systems. Through this interplay between virtualized and non-virtualized resources, we provide a flexible resource management framework that can be adapted based on users' demands. The need for such a framework is motivated by real user data gathered during our operation of FutureGrid (FG). We observed that the usage of the different infrastructures vary over time changing from being over-utilized to underutilize and vice versa. Therefore, the proposed framework will be beneficial for users of environments such a FutureGrid where several infrastructures are supported with limited physical resources. Copyright 2012 ACM.

@inproceedings{
 title = {Message from the program co-chairs},
 type = {inproceedings},
 year = {2012},
 id = {596e5904-a0be-31a0-9cfb-f69faa296217},
 created = {2018-02-27T18:07:44.077Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.077Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Grossman, R. and Milojicic, D. and Kozuch, M. and McGeer, R.},
 booktitle = {FederatedClouds'12 - Proceedings of the 2012 Workshop on Cloud Services, Federation, and the 8th Open Cirrus Summit, Co-located with ICAC'12}
}

@inproceedings{
 title = {Design of a Dynamic Provisioning System for a Federated Cloud and Bare-metal Environment},
 type = {inproceedings},
 year = {2012},
 id = {e99ec46c-feec-32c7-8803-5b81947cafc2},
 created = {2018-02-27T18:07:45.184Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.184Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {von Laszewski, Gregor and Lee, Hyungro and Diaz, Javier and Wang, Fugang and Tanaka, Koji and Karavinkoppa, Shubhada and Fox, Geoffrey C and Furlani, Tom},
 booktitle = {Proc. Workshop on Cloud Services, Federation, and the 8th Open Cirrus Summit}
}

@book{
 title = {Supporting experimental computer science},
 type = {book},
 year = {2012},
 publisher = {INRIA},
 id = {ef76f06f-5d07-3d20-a2a0-04f9a4899694},
 created = {2018-02-27T18:07:45.295Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.295Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {DISS},
 private_publication = {false},
 bibtype = {book},
 author = {Desprez, Frédéric and Fox, Geoffrey and Jeannot, Emmanuel and Keahey, Kate and Kozuch, Michael and Margery, David and Neyron, Pierre and Nussbaum, Lucas and Perez, Christian and Richard, Olivier}
}

@article{
 title = {FutureGrid User Support},
 type = {article},
 year = {2011},
 id = {8f7d053b-b438-3e61-b155-215dac504e9f},
 created = {2018-02-27T18:07:43.373Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.373Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Pike, Gregory and Younge, Andrew and von Laszewski, Gregor and Wang, Fugang and Diaz, Javier and Kulshrestha, Archit and Fox, Geoffrey},
 journal = {FutureGrid NSF Review, Indiana University}
}

@inproceedings{
 title = {FutureGrid image repository: A generic catalog and storage system for heterogeneous virtual machine images},
 type = {inproceedings},
 year = {2011},
 id = {7f252262-e6a4-33f7-976f-349f1b5251ac},
 created = {2018-02-27T18:07:43.495Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.495Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {FutureGrid (FG) is an experimental, high-performance testbed that supports HPC, cloud and grid computing experiments for both application and computer scientist. FutureGrid includes the use of virtualization technology to allow the support of a wide range of operating systems in order to include a testbed for various cloud computing infrastructure as a service frameworks. Therefore, efficient management of a variety of virtual machine images becomes a key issue. Current cloud frameworks do not provide a way to manage images for different IaaS frameworks. They typically provide their own image repositories, but in general they do not allow us to store the needed metadata to handle other IaaS images. We present a generic catalog and image repository to store images of any type. Our image repository has a convenient interface that distinguishes image types. Therefore, it is not only useful for FutureGrid, but also for any application that needs to manage images. © 2011 IEEE.},
 bibtype = {inproceedings},
 author = {Diaz, J. and Von Laszewski, G. and Wang, F. and Younge, A.J. and Fox, G.},
 doi = {10.1109/CloudCom.2011.85},
 booktitle = {Proceedings - 2011 3rd IEEE International Conference on Cloud Computing Technology and Science, CloudCom 2011}
}

FutureGrid (FG) is an experimental, high-performance testbed that supports HPC, cloud and grid computing experiments for both application and computer scientist. FutureGrid includes the use of virtualization technology to allow the support of a wide range of operating systems in order to include a testbed for various cloud computing infrastructure as a service frameworks. Therefore, efficient management of a variety of virtual machine images becomes a key issue. Current cloud frameworks do not provide a way to manage images for different IaaS frameworks. They typically provide their own image repositories, but in general they do not allow us to store the needed metadata to handle other IaaS images. We present a generic catalog and image repository to store images of any type. Our image repository has a convenient interface that distinguishes image types. Therefore, it is not only useful for FutureGrid, but also for any application that needs to manage images. © 2011 IEEE.

@article{
 title = {eMOLST: a documentation flow for distributed health informatics},
 type = {article},
 year = {2011},
 pages = {1857-1867},
 volume = {23},
 publisher = {Wiley Online Library},
 id = {5a261f18-cd98-38ed-9573-8bfd2d755c0b},
 created = {2018-02-27T18:07:43.508Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.508Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {von Laszewski, Gregor and Dayal, Jai and Wang, Lizhe},
 journal = {Concurrency and Computation: Practice and Experience},
 number = {16}
}

@inproceedings{
 title = {Towards generic FutureGrid image management},
 type = {inproceedings},
 year = {2011},
 id = {313a7508-1029-3a1a-8fe9-b1051369fd73},
 created = {2018-02-27T18:07:44.173Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.173Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {In this paper, we briefly outline the current design of a generic image management service for FutureGrid. The service is intended to generate, store, and verify images while interfacing with different localized cloud IaaS image. Additionally, we will also use the service to generate images for traditional bare-metal deployments. © 2011 Authors.},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Diaz, J. and Wang, F. and Younge, A.J. and Kulshrestha, A. and Fox, G.},
 doi = {10.1145/2016741.2016758},
 booktitle = {Proceedings of the TeraGrid 2011 Conference: Extreme Digital Discovery, TG'11}
}

In this paper, we briefly outline the current design of a generic image management service for FutureGrid. The service is intended to generate, store, and verify images while interfacing with different localized cloud IaaS image. Additionally, we will also use the service to generate images for traditional bare-metal deployments. © 2011 Authors.

@article{
 title = {Towards cloud deployments using FutureGrid},
 type = {article},
 year = {2011},
 volume = {47408},
 id = {690e014b-90bc-34f1-a919-c8b76cf3604a},
 created = {2018-02-27T18:07:44.337Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.337Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {von Laszewski, Gregor and Diaz, Javier and Wang, Fugang and Fox, Geoffrey C and von Laszewski, Gregor and Wang, Fugang and Fox, Geoffrey C},
 journal = {Indiana University, Bloomington, IN}
}

@article{
 title = {Task scheduling with ANN-based temperature prediction in a data center: a simulation-based study},
 type = {article},
 year = {2011},
 pages = {381-391},
 volume = {27},
 publisher = {Springer},
 id = {04174a6b-fcfe-351f-bb0b-af3af5ed7907},
 created = {2018-02-27T18:07:44.367Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.367Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and von Laszewski, Gregor and Huang, Fang and Dayal, Jai and Frulani, Tom and Fox, Geoffrey},
 journal = {Engineering with Computers},
 number = {4}
}

@article{
 title = {Towards on demand it service deployment},
 type = {article},
 year = {2011},
 pages = {249-262},
 volume = {7},
 id = {bea82399-8829-3152-8b1a-a088c986f6d9},
 created = {2018-02-27T18:07:44.695Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.695Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Dayal, Jai and Rathbone, Casey and Wang, Lizhe and von Laszewski, Gregor},
 journal = {Internet Policies and Issues}
}

@article{
 title = {Threat detection in urban water distribution systems with simulations conducted in grids and clouds},
 type = {article},
 year = {2011},
 volume = {95},
 id = {093ee057-682f-3864-b428-f755ac862c19},
 created = {2018-02-27T18:07:44.728Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.728Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {We present a workflow-based algorithm for identifying threads to an urban water management system. Through Grid computing we provide the necessary high-performance computing resources to deliver quickly solutions to the problem. We prototyped a new middleware called cyberaide, that enables easy access to Grid resources through portals or the command line. A workflow system is used to manage resources in fault tolerant fashion. In addition, we contrast the architecture with a Hadoop implementation. Resources from TeraGrid and FutureGrid are used to test the feasibility of using the toolkit for a scientific application. © Civil-Comp Press, 2011.},
 bibtype = {article},
 author = {Von Laszewski, G. and Wang, L. and Wang, F. and Fox, G.C. and Mahinthakumar, G.K.},
 journal = {Civil-Comp Proceedings}
}

We present a workflow-based algorithm for identifying threads to an urban water management system. Through Grid computing we provide the necessary high-performance computing resources to deliver quickly solutions to the problem. We prototyped a new middleware called cyberaide, that enables easy access to Grid resources through portals or the command line. A workflow system is used to manage resources in fault tolerant fashion. In addition, we contrast the architecture with a Hadoop implementation. Resources from TeraGrid and FutureGrid are used to test the feasibility of using the toolkit for a scientific application. © Civil-Comp Press, 2011.

@article{
 title = {WANG, Lizhe},
 type = {article},
 year = {2011},
 pages = {714-722},
 volume = {42},
 id = {5920bc96-e3f1-3ad9-b33e-e9ad90d057cf},
 created = {2018-02-27T18:07:44.945Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.945Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, L and Kunze, M and Tao, J and Von Laszewski, G and Chen, D and Deng, Z and Huang, F and Khan, S U and Dayal, J and Chen, J},
 journal = {Advances in Engineering Software},
 number = {9}
}

@inproceedings{
 title = {Analysis of virtualization technologies for high performance computing environments},
 type = {inproceedings},
 year = {2011},
 keywords = {Cloud computing,Computer software selection and evaluation; Compu,Distributed systems; High performance computing; H},
 pages = {9-16},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-80053156886&doi=10.1109%2FCLOUD.2011.29&partnerID=40&md5=9b37566e08b56cb54c83c9be6b8f615d},
 city = {Washington, DC},
 id = {4240f768-ac59-3a09-b77f-39bce7c459c1},
 created = {2018-02-27T18:07:45.217Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.217Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Younge20119},
 source_type = {conference},
 notes = {cited By 99; Conference of 2011 IEEE 4th International Conference on Cloud Computing, CLOUD 2011 ; Conference Date: 4 July 2011 Through 9 July 2011; Conference Code:86655},
 private_publication = {false},
 abstract = {As Cloud computing emerges as a dominant paradigm in distributed systems, it is important to fully understand the underlying technologies that make Clouds possible. One technology, and perhaps the most important, is virtualization. Recently virtualization, through the use of hypervisors, has become widely used and well understood by many. However, there are a large spread of different hypervisors, each with their own advantages and disadvantages. This paper provides an in-depth analysis of some of today's commonly accepted virtualization technologies from feature comparison to performance analysis, focusing on the applicability to High Performance Computing environments using FutureGrid resources. The results indicate virtualization sometimes introduces slight performance impacts depending on the hypervisor type, however the benefits of such technologies are profound and not all virtualization technologies are equal. From our experience, the KVM hypervisor is the optimal choice for supporting HPC applications within a Cloud infrastructure. © 2011 IEEE.},
 bibtype = {inproceedings},
 author = {Younge, A J and Henschel, R and Brown, J T and Von Laszewski, G and Qiu, J and Fox, G C},
 doi = {10.1109/CLOUD.2011.29},
 booktitle = {Proceedings - 2011 IEEE 4th International Conference on Cloud Computing, CLOUD 2011}
}

As Cloud computing emerges as a dominant paradigm in distributed systems, it is important to fully understand the underlying technologies that make Clouds possible. One technology, and perhaps the most important, is virtualization. Recently virtualization, through the use of hypervisors, has become widely used and well understood by many. However, there are a large spread of different hypervisors, each with their own advantages and disadvantages. This paper provides an in-depth analysis of some of today's commonly accepted virtualization technologies from feature comparison to performance analysis, focusing on the applicability to High Performance Computing environments using FutureGrid resources. The results indicate virtualization sometimes introduces slight performance impacts depending on the hypervisor type, however the benefits of such technologies are profound and not all virtualization technologies are equal. From our experience, the KVM hypervisor is the optimal choice for supporting HPC applications within a Cloud infrastructure. © 2011 IEEE.

@article{
 title = {Towards building a cloud for scientific applications},
 type = {article},
 year = {2011},
 pages = {714-722},
 volume = {42},
 publisher = {Elsevier},
 id = {5983aa05-a583-3c3f-9120-9714af5450ad},
 created = {2018-02-27T18:07:45.451Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.451Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Kunze, Marcel and Tao, Jie and von Laszewski, Gregor},
 journal = {Advances in Engineering software},
 number = {9}
}

@inproceedings{
 title = {Design of the Futuregrid experiment management framework},
 type = {inproceedings},
 year = {2010},
 id = {93ed57a3-3688-31a8-8632-f4d5edf8b7ff},
 created = {2018-02-27T18:07:43.317Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.317Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {FutureGrid provides novel computing capabilities that enable reproducible experiments while simultaneously supporting dynamic provisioning. This paper describes the FutureGrid experiment management framework to create and execute large scale scientic experiments for researchers around the globe. The experiments executed are performed by the various users of FutureGrid ranging from administrators to software developers and end users. The Experiment management framework will consist of software tools that record user and system actions to generate a reproducible set of tasks and resource congurations. Additionally, the experiment management framework can be used to share not only the experiment setup, but also performance information for the specic instantiation of the experiment. This makes it possible to compare a variety of experiment setups and analyze the impact Grid and Cloud software stacks have.},
 bibtype = {inproceedings},
 author = {Von Laszewski, G. and Fox, G.C. and Wang, F. and Younge, A.J. and Kulshrestha, A. and Pike, G.G. and Smithy, W. and Vöcklerz, J. and Figueiredox, R.J. and Fortesx, J. and Keahey, K.},
 doi = {10.1109/GCE.2010.5676126},
 booktitle = {2010 Gateway Computing Environments Workshop, GCE 2010}
}

FutureGrid provides novel computing capabilities that enable reproducible experiments while simultaneously supporting dynamic provisioning. This paper describes the FutureGrid experiment management framework to create and execute large scale scientic experiments for researchers around the globe. The experiments executed are performed by the various users of FutureGrid ranging from administrators to software developers and end users. The Experiment management framework will consist of software tools that record user and system actions to generate a reproducible set of tasks and resource congurations. Additionally, the experiment management framework can be used to share not only the experiment setup, but also performance information for the specic instantiation of the experiment. This makes it possible to compare a variety of experiment setups and analyze the impact Grid and Cloud software stacks have.

@inproceedings{
 title = {Cyberaide onServe: Software as a Service on Production Grids},
 type = {inproceedings},
 year = {2010},
 pages = {395-403},
 publisher = {IEEE},
 id = {d875a5a5-5c19-3f74-97a6-a2f0ea9e2b14},
 created = {2018-02-27T18:07:43.597Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.597Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Kurze, Tobias and Wang, Lizhe and von Laszewski, Gregor and Tao, Jie and Kunze, Marcel and Kramer, David and Karl, Wolfgang},
 booktitle = {Parallel Processing (ICPP), 2010 39th International Conference on}
}

@inproceedings{
 title = {Framing the issues of cloud computing & sustainability: A design perspective},
 type = {inproceedings},
 year = {2010},
 keywords = {Cloud computing,Computer systems; Human computer interaction; Sus,Environmental effects; Interactivity; Social issue},
 pages = {603-608},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-79952364799&doi=10.1109%2FCloudCom.2010.77&partnerID=40&md5=4f0fd589048fbf4048df7fd1b5ea859d},
 city = {Indianapolis, IN},
 id = {989645d3-1fcf-37d6-86df-6e138399d577},
 created = {2018-02-27T18:07:43.886Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.886Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Pan2010603},
 source_type = {conference},
 notes = {cited By 4; Conference of 2nd IEEE International Conference on Cloud Computing Technology and Science, CloudCom 2010 ; Conference Date: 30 November 2010 Through 3 December 2010; Conference Code:84061},
 private_publication = {false},
 abstract = {In this paper, we describe the present lack of understanding about if the potential environmental effects of transitions to cloud computing are positive or negative. We describe that research about the human interactivity implications of and for cloud computing has yet to enter the arena of Human Computer Interaction (HCI) in a significant way. We describe a short inventory of what is presently in the HCI literature apropos of cloud computing and interactivity. In addition, we offer a description of how we think the issues of cloud computing in the perspective of HCI may be framed, as well as an inventory of social issues implicated in cloud computing. Finally, we suggest some projects and problems that may be appropriate for advancing cloud computing in the perspective of HCI with sustainability as a key goal. © 2010 IEEE.},
 bibtype = {inproceedings},
 author = {Pan, Y and Maini, S and Blevis, E},
 doi = {10.1109/CloudCom.2010.77},
 booktitle = {Proceedings - 2nd IEEE International Conference on Cloud Computing Technology and Science, CloudCom 2010}
}

In this paper, we describe the present lack of understanding about if the potential environmental effects of transitions to cloud computing are positive or negative. We describe that research about the human interactivity implications of and for cloud computing has yet to enter the arena of Human Computer Interaction (HCI) in a significant way. We describe a short inventory of what is presently in the HCI literature apropos of cloud computing and interactivity. In addition, we offer a description of how we think the issues of cloud computing in the perspective of HCI may be framed, as well as an inventory of social issues implicated in cloud computing. Finally, we suggest some projects and problems that may be appropriate for advancing cloud computing in the perspective of HCI with sustainability as a key goal. © 2010 IEEE.

@inproceedings{
 title = {Schedule distributed virtual machines in a service oriented environment},
 type = {inproceedings},
 year = {2010},
 pages = {230-236},
 publisher = {IEEE},
 id = {8dbde77e-d252-32a0-a2d4-eed189715e41},
 created = {2018-02-27T18:07:44.074Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.074Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Wang, Lizhe and Von Laszewski, Gregor and Kunze, Marcel and Tao, Jie},
 booktitle = {Advanced Information Networking and Applications (AINA), 2010 24th IEEE International Conference on}
}

@inproceedings{
 title = {Power aware scheduling for parallel tasks via task clustering},
 type = {inproceedings},
 year = {2010},
 pages = {629-634},
 publisher = {IEEE},
 id = {a1c3dee8-9268-3c10-ac9f-f0011eecf2df},
 created = {2018-02-27T18:07:44.156Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.156Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Wang, Lizhe and Tao, Jie and von Laszewski, Gregor and Chen, Dan},
 booktitle = {Parallel and Distributed Systems (ICPADS), 2010 IEEE 16th International Conference on}
}

@article{
 title = {Provide virtual distributed environments for grid computing on demand},
 type = {article},
 year = {2010},
 pages = {213-219},
 volume = {41},
 publisher = {Elsevier},
 id = {e3464e63-b01b-357d-a584-e876a15cdc16},
 created = {2018-02-27T18:07:44.158Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.158Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Von Laszewski, Gregor and Kunze, Marcel and Tao, Jie and Dayal, Jai},
 journal = {Advances in Engineering Software},
 number = {2}
}

@inbook{
 type = {inbook},
 year = {2010},
 pages = {77-88},
 publisher = {Springer},
 id = {9bb32d70-c082-3508-8f06-3c8499b42918},
 created = {2018-02-27T18:07:44.354Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.354Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CHAP},
 private_publication = {false},
 bibtype = {inbook},
 author = {von Laszewski, Gregor and Wang, Lizhe},
 chapter = {GreenIT service level agreements},
 title = {Grids and Service-Oriented Architectures for Service Level Agreements}
}

@inproceedings{
 title = {Enabling energy-efficient analysis of massive neural signals using GPGPU},
 type = {inproceedings},
 year = {2010},
 pages = {147-154},
 publisher = {IEEE},
 id = {e7e7daa8-6b14-311a-9fb6-5a005638b735},
 created = {2018-02-27T18:07:44.446Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.446Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Chen, Dan and Wang, Lizhe and Wang, Shuaiting and Xiong, Muzhou and von Laszewski, Gregor and Li, Xiaoli},
 booktitle = {Green Computing and Communications (GreenCom), 2010 IEEE/ACM Int'l Conference on & Int'l Conference on Cyber, Physical and Social Computing (CPSCom)}
}

@book{
 title = {Cyberaide virtual applicance: On-demand deploying middleware for cyberinfrastructure},
 type = {book},
 year = {2010},
 source = {Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering},
 volume = {34 LNICST},
 id = {5ae11bfc-f932-392b-bf7e-e0057ae3d7ba},
 created = {2018-02-27T18:07:44.665Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.665Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Cyberinfrastructure offers a vision of advanced knowledge infrastructure for research and education. It integrates diverse resources across geographically distributed resources and human communities. Cyberaide is a service oriented architecture and abstraction framework that integrates a large number of available commodity libraries and allows users to access cyberinfrastructure through Web 2.0 technologies. This paper describes the Cyberaide virtual appliance, a solution of on-demand deployment of cyberinfrastructure middleware, i.e. Cyberaide. The proposed solution is based on an open and free technology and software - Cyberaide JavaScript, a service oriented architecture (SOA) and grid abstraction framework that allows users to access the grid infrastructures through JavaScript. The Cyberaide virtual appliance is built by installing and configuring Cyberaide JavaScript in a virtual machine. Established Cyberaide virtual appliances can then be used via a Web browser, allowing users to create, distribute and maintain cyberinfrastructure related software more easily even without the need to do the "tricky" installation process on their own. We argue that our solution of providing Cyberaide virtual appliance can make users easy to access cyberinfrastructure, manage their work and build user organizations. © Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering 2010.},
 bibtype = {book},
 author = {Kurze, T. and Wang, L. and Von Laszewski, G. and Tao, J. and Kunze, M. and Wang, F. and Kramer, D. and Karl, W. and Ekanayake, J.},
 doi = {10.1007/978-3-642-12636-9_10}
}

Cyberinfrastructure offers a vision of advanced knowledge infrastructure for research and education. It integrates diverse resources across geographically distributed resources and human communities. Cyberaide is a service oriented architecture and abstraction framework that integrates a large number of available commodity libraries and allows users to access cyberinfrastructure through Web 2.0 technologies. This paper describes the Cyberaide virtual appliance, a solution of on-demand deployment of cyberinfrastructure middleware, i.e. Cyberaide. The proposed solution is based on an open and free technology and software - Cyberaide JavaScript, a service oriented architecture (SOA) and grid abstraction framework that allows users to access the grid infrastructures through JavaScript. The Cyberaide virtual appliance is built by installing and configuring Cyberaide JavaScript in a virtual machine. Established Cyberaide virtual appliances can then be used via a Web browser, allowing users to create, distribute and maintain cyberinfrastructure related software more easily even without the need to do the "tricky" installation process on their own. We argue that our solution of providing Cyberaide virtual appliance can make users easy to access cyberinfrastructure, manage their work and build user organizations. © Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering 2010.

@article{
 title = {Provide virtual machine information for grid computing},
 type = {article},
 year = {2010},
 pages = {1362-1374},
 volume = {40},
 publisher = {IEEE},
 id = {bc9ca47b-b15a-316e-8f3f-ed29205184e4},
 created = {2018-02-27T18:07:44.682Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.682Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Von Laszewski, Gregor and Chen, Dan and Tao, Jie and Kunze, Marcel},
 journal = {IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans},
 number = {6}
}

@inproceedings{
 title = {Towards energy aware scheduling for precedence constrained parallel tasks in a cluster with DVFS},
 type = {inproceedings},
 year = {2010},
 id = {8ce57fd6-50a7-3186-88ba-caa1760724a7},
 created = {2018-02-27T18:07:44.898Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.898Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {Reducing energy consumption for high end computing can bring various benefits such as, reduce operating costs, increase system reliability, and environment respect. This paper aims to develop scheduling heuristics and to present application experience for reducing power consumption of parallel tasks in a cluster with the Dynamic Voltage Frequency Scaling (DVFS) technique. In this paper, formal models are presented for precedence-constrained parallel tasks, DVFS enabled clusters, and energy consumption. This paper studies the slack time for non-critical jobs, extends their execution time and reduces the energy consumption without increasing the task's execution time as a whole. Additionally, Green Service Level Agreement is also considered in this paper. By increasing task execution time within an affordable limit, this paper develops scheduling heuristics to reduce energy consumption of a tasks execution and discusses the relationship between energy consumption and task execution time. Models and scheduling heuristics are examined with a simulation study. Test results justify the design and implementation of proposed energy aware scheduling heuristics in the paper. © 2010 IEEE.},
 bibtype = {inproceedings},
 author = {Wang, L. and Von Laszewski, G. and Dayal, J. and Wang, F.},
 doi = {10.1109/CCGRID.2010.19},
 booktitle = {CCGrid 2010 - 10th IEEE/ACM International Conference on Cluster, Cloud, and Grid Computing}
}

Reducing energy consumption for high end computing can bring various benefits such as, reduce operating costs, increase system reliability, and environment respect. This paper aims to develop scheduling heuristics and to present application experience for reducing power consumption of parallel tasks in a cluster with the Dynamic Voltage Frequency Scaling (DVFS) technique. In this paper, formal models are presented for precedence-constrained parallel tasks, DVFS enabled clusters, and energy consumption. This paper studies the slack time for non-critical jobs, extends their execution time and reduces the energy consumption without increasing the task's execution time as a whole. Additionally, Green Service Level Agreement is also considered in this paper. By increasing task execution time within an affordable limit, this paper develops scheduling heuristics to reduce energy consumption of a tasks execution and discusses the relationship between energy consumption and task execution time. Models and scheduling heuristics are examined with a simulation study. Test results justify the design and implementation of proposed energy aware scheduling heuristics in the paper. © 2010 IEEE.

@article{
 title = {Multicores in Cloud Computing: Research Challenges for Applications.},
 type = {article},
 year = {2010},
 pages = {958-964},
 volume = {5},
 id = {9c54bc21-b581-3043-b32c-f8f22e0e85fd},
 created = {2018-02-27T18:07:45.232Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.232Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Tao, Jie and von Laszewski, Gregor and Marten, Holger},
 journal = {JCP},
 number = {6}
}

@inproceedings{
 title = {Schedule Virtual Machines in a Distributed Service Oriented Environment},
 type = {inproceedings},
 year = {2010},
 pages = {230-236},
 publisher = {Eggenstein-Leopoldshafen, Germany},
 id = {ae57c289-5472-34cb-8cc8-7fbb3e07244b},
 created = {2018-02-27T18:07:45.400Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.400Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 abstract = {Virtual machines offer unique advantages to the scientific computing community, such as Quality of Service(QoS) guarantee, performance isolation, easy resource management, and the on-demand deployment of computing environments. Using virtual machines as a computing resource within a dis- tributed environment, such as Service Oriented Architecture (SOA), creates a variety of new issues and challenges that must be overcome. Traditionally, parallel task scheduling algorithms only focus on handling CPU resources. Using of a virtual machine, however, requires the monitoring and management of additional resource properties. Additionally, CPU, memory, storage, and software licenses must also be considered within the scheduling algorithm. The objective of this paper is to address these challenges of a multi-dimensional scheduling algorithm for virtual machines within a SOA. To do this, we deploy a testbed SOA environment composed of virtual machines which are capable of being registered, indexed, allocated, accessed, and controlled by our new parallel task scheduling algorithm.},
 bibtype = {inproceedings},
 author = {Wang, Lizhe and von Laszewski, Gregor and Kunze, Marcel and Tao, Jie and Dayal, Jai and Rathbone, Casey},
 booktitle = {24th IEEE International Conference on Advanced Information Networking and Applications}
}

Virtual machines offer unique advantages to the scientific computing community, such as Quality of Service(QoS) guarantee, performance isolation, easy resource management, and the on-demand deployment of computing environments. Using virtual machines as a computing resource within a dis- tributed environment, such as Service Oriented Architecture (SOA), creates a variety of new issues and challenges that must be overcome. Traditionally, parallel task scheduling algorithms only focus on handling CPU resources. Using of a virtual machine, however, requires the monitoring and management of additional resource properties. Additionally, CPU, memory, storage, and software licenses must also be considered within the scheduling algorithm. The objective of this paper is to address these challenges of a multi-dimensional scheduling algorithm for virtual machines within a SOA. To do this, we deploy a testbed SOA environment composed of virtual machines which are capable of being registered, indexed, allocated, accessed, and controlled by our new parallel task scheduling algorithm.

@article{
 title = {Virtual data system on distributed virtual machines in computational grids},
 type = {article},
 year = {2010},
 pages = {194-204},
 volume = {6},
 publisher = {Inderscience Publishers},
 id = {c17f453d-8e99-3c3c-b45d-6e8d1193ca2d},
 created = {2018-02-27T18:07:45.508Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.508Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Laszewski, Gregor Von and Tao, Jie and Kunze, Marcel},
 journal = {International Journal of Ad Hoc and Ubiquitous Computing},
 number = {4}
}

@inproceedings{
 title = {Efficient resource management for cloud computing environments},
 type = {inproceedings},
 year = {2010},
 pages = {357-364},
 publisher = {IEEE},
 id = {0a0b5c3e-2915-3253-9c03-0199378eddc7},
 created = {2018-02-27T18:07:45.575Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.575Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Younge, Andrew J and Von Laszewski, Gregor and Wang, Lizhe and Lopez-Alarcon, Sonia and Carithers, Warren},
 booktitle = {Green Computing Conference, 2010 International}
}

@inproceedings{
 title = {What is cyberinfrastructure},
 type = {inproceedings},
 year = {2010},
 pages = {37},
 websites = {http://dx.doi.org/10.1145/1878335.1878347,http://portal.acm.org/citation.cfm?doid=1878335.1878347},
 publisher = {ACM Press},
 city = {New York, New York, USA},
 id = {01def705-6282-3a8c-be4f-5411ddb65538},
 created = {2019-09-06T19:06:37.503Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2020-09-09T18:07:05.158Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Stewart2010},
 source_type = {inproceedings},
 private_publication = {false},
 abstract = {Cyberinfrastructure is a word commonly used but lacking a single, precise definition. One recognizes intuitively the analogy with infrastructure, and the use of cyber to refer to thinking or computing - but what exactly is cyberinfrastructure as opposed to information technology infrastructure? Indiana University has developed one of the more widely cited definitions of cyberinfrastructure: Cyberinfrastructure consists of computing systems, data storage systems, advanced instruments and data repositories, visualization environments, and people, all linked together by software and high performance networks to improve research productivity and enable breakthroughs not otherwise possible. A second definition, more inclusive of scholarship generally and educational activities, has also been published and is useful in describing cyberinfrastructure: Cyberinfrastructure consists of computational systems, data and information management, advanced instruments, visualization environments, and people, all linked together by software and advanced networks to improve scholarly productivity and enable knowledge breakthroughs and discoveries not otherwise possible. In this paper, we describe the origin of the term cyberinfrastructure based on the history of the root word infrastructure, discuss several terms related to cyberinfrastructure, and provide several examples of cyberinfrastructure. © 2010 ACM.},
 bibtype = {inproceedings},
 author = {Stewart, Craig A and Simms, Stephen and Plale, Beth and Link, Matthew and Hancock, David Y. and Fox, Geoffrey Charles},
 doi = {10.1145/1878335.1878347},
 booktitle = {Proceedings of the 38th annual fall conference on SIGUCCS - SIGUCCS '10}
}

Cyberinfrastructure is a word commonly used but lacking a single, precise definition. One recognizes intuitively the analogy with infrastructure, and the use of cyber to refer to thinking or computing - but what exactly is cyberinfrastructure as opposed to information technology infrastructure? Indiana University has developed one of the more widely cited definitions of cyberinfrastructure: Cyberinfrastructure consists of computing systems, data storage systems, advanced instruments and data repositories, visualization environments, and people, all linked together by software and high performance networks to improve research productivity and enable breakthroughs not otherwise possible. A second definition, more inclusive of scholarship generally and educational activities, has also been published and is useful in describing cyberinfrastructure: Cyberinfrastructure consists of computational systems, data and information management, advanced instruments, visualization environments, and people, all linked together by software and advanced networks to improve scholarly productivity and enable knowledge breakthroughs and discoveries not otherwise possible. In this paper, we describe the origin of the term cyberinfrastructure based on the history of the root word infrastructure, discuss several terms related to cyberinfrastructure, and provide several examples of cyberinfrastructure. © 2010 ACM.

@article{
 title = {Grid virtualization engine: design, implementation, and evaluation},
 type = {article},
 year = {2009},
 pages = {477-488},
 volume = {3},
 publisher = {IEEE},
 id = {36a838a0-c66c-305d-a657-e657d6a0fefe},
 created = {2018-02-27T18:07:43.200Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.200Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Wang, Lizhe and Von Laszewski, Gregor and Tao, Jie and Kunze, Marcel},
 journal = {IEEE Systems Journal},
 number = {4}
}

@inproceedings{
 title = {Thermal aware workload scheduling with backfilling for green data centers},
 type = {inproceedings},
 year = {2009},
 pages = {289-296},
 publisher = {IEEE},
 id = {ffc14f18-2367-36d0-a8cb-71b6b617ddeb},
 created = {2018-02-27T18:07:43.396Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.396Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Wang, Lizhe and von Laszewski, Gregor and Dayal, Jai and Furlani, Thomas R},
 booktitle = {Performance Computing and Communications Conference (IPCCC), 2009 IEEE 28th International}
}

@inproceedings{
 title = {Open grid computing environments},
 type = {inproceedings},
 year = {2009},
 id = {c0e84d43-89de-3695-8095-653b21f10258},
 created = {2018-02-27T18:07:43.873Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.873Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Pierce, Marlon and Marru, Suresh and Wu, Wenjun and Kandaswami, Gopi and von Laszewski, Gregor and Dooley, Rion and Dahan, Maytal and Wilkins-Diehr, Nancy and Thomas, Mary and Center, T},
 booktitle = {Proceedings of the Fourth Annual TeraGrid Conference}
}

@inproceedings{
 title = {Cyberaide Virtual Applicance: On-Demand Deploying Middleware for Cyberinfrastructure},
 type = {inproceedings},
 year = {2009},
 pages = {132-144},
 publisher = {Springer},
 id = {923385a4-b2e5-36e1-91c0-9600ce8fd212},
 created = {2018-02-27T18:07:43.950Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:43.950Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Kurze, Tobias and Wang, Lizhe and Von Laszewski, Gregor and Tao, Jie and Kunze, Marcel and Wang, Fugang and Kramer, David and Karl, Wolfgang and Ekanayake, Jaliya},
 booktitle = {International Conference on Cloud Computing}
}

@techreport{
 title = {Water Threat Management Report},
 type = {techreport},
 year = {2009},
 id = {d09f0e03-255c-3e50-94de-0c82ccc0ac14},
 created = {2018-02-27T18:07:44.770Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.770Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {techreport},
 author = {von Laszewski, Gregor}
}

@inproceedings{
 title = {Cyberaide creative: On-demand cyberinfrastructure provision in clouds},
 type = {inproceedings},
 year = {2009},
 id = {b7125108-7315-3447-9458-9ab9dea0d958},
 created = {2018-02-27T18:07:44.972Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.972Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 private_publication = {false},
 abstract = {As demand for grid and cloud computing solutions increases, the need for user oriented software to provide access to theses resources also increases. Until recently the use of computing resources was limited to those with exceptional knowledge of the system design and configuration. With the advent of grid middleware projects this started to change allowing for new users not familiar with the complex grid infrastructure and client software to use the systems for their own research. The Cyberaide Gridshell demonstrated this by developing a user friendly interface to submit jobs to a grid. Following this theme it is our objective to create a tool that will take another step further by abstracting the creation and configuration of the infrastructure and system software away from the end-user. This will be achieved through the use of cloud resources provided by VMware virtualization and deployment via a web interface. We will show the benefits of deploying cyberinfrastructures, like clusters and grids, on a cloud design by demonstrating the ease of cyberinfrastructure deployment and the versatility of the systems that can be spawned on demand. © 2009 IEEE.},
 bibtype = {inproceedings},
 author = {Rathbone, C. and Wang, L. and Von Laszewski, G. and Wang, F.},
 doi = {10.1109/I-SPAN.2009.23},
 booktitle = {I-SPAN 2009 - The 10th International Symposium on Pervasive Systems, Algorithms, and Networks}
}

As demand for grid and cloud computing solutions increases, the need for user oriented software to provide access to theses resources also increases. Until recently the use of computing resources was limited to those with exceptional knowledge of the system design and configuration. With the advent of grid middleware projects this started to change allowing for new users not familiar with the complex grid infrastructure and client software to use the systems for their own research. The Cyberaide Gridshell demonstrated this by developing a user friendly interface to submit jobs to a grid. Following this theme it is our objective to create a tool that will take another step further by abstracting the creation and configuration of the infrastructure and system software away from the end-user. This will be achieved through the use of cloud resources provided by VMware virtualization and deployment via a web interface. We will show the benefits of deploying cyberinfrastructures, like clusters and grids, on a cloud design by demonstrating the ease of cyberinfrastructure deployment and the versatility of the systems that can be spawned on demand. © 2009 IEEE.

@inproceedings{
 title = {Flexible framework for commodity FPGA cluster computing},
 type = {inproceedings},
 year = {2009},
 pages = {465-471},
 publisher = {IEEE},
 id = {cb98f8c4-9ba0-332a-9755-2212644e945a},
 created = {2018-02-27T18:07:45.012Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.012Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {CONF},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Espenshade, Jeremy and Lukowiak, Marcin and Shaaban, Muhammad and von Laszewski, Gregor},
 booktitle = {Field-Programmable Technology, 2009. FPT 2009. International Conference on}
}

@inproceedings{
 title = {Towards Thermal Aware Workload Scheduling in a Data Center},
 type = {inproceedings},
 year = {2009},
 id = {000cf64c-be1e-3289-bf6c-8b925a84edbb},
 created = {2018-02-27T18:07:45.264Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:45.264Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {inproceedings},
 author = {Wang, Lizhe and Dayal, Jai and von Laszewski, Gregor},
 booktitle = {Pervasive Systems, Algorithms, and Networks (ISPAN), 2009 10th International Symposium on}
}

@article{
 title = {FutureGrid Image Management Framework to Support Cloud and HPC Dynamic Provisioning},
 type = {article},
 id = {78633120-7094-319b-b1f4-779149877fd2},
 created = {2018-02-27T18:07:44.650Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 group_id = {f0704cbc-a3d0-3264-b41e-15f4ed0c92ee},
 last_modified = {2018-02-27T18:07:44.650Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 source_type = {JOUR},
 private_publication = {false},
 bibtype = {article},
 author = {Diaz, Javier and von Laszewski, Gregor and Wang, Fugang and Fox, Geoffrey}
}