Engineering topics

Fox, Geoffrey C.

Publications and source records attributed to Fox, Geoffrey C..

4 records

DOE OSTI2023

Opportunities for enhancing MLCommons efforts while leveraging insights from educational MLCommons earthquake benchmarks efforts

MLCommons is an effort to develop and improve the artificial intelligence (AI) ecosystem through benchmarks, public data sets, and research. It consists of members from start-ups, leading companies, academics, and non-profits from around the world. The goal is to make machine learning better for everyone. In order to increase participation by others, educational institutions provide valuable opportunities for engagement. In this article, we identify numerous insights obtained from different viewpoints as part of efforts to utilize high-performance computing (HPC) big data systems in existing education while developing and conducting science benchmarks for earthquake prediction. As this activity was conducted across multiple educational efforts, we project if and how it is possible to make such efforts available on a wider scale. This includes the integration of sophisticated benchmarks into courses and research activities at universities, exposing the students and researchers to topics that are otherwise typically not sufficiently covered in current course curricula as we witnessed from our practical experience across multiple organizations. As such, we have outlined the many lessons we learned throughout these efforts, culminating in the need for benchmark carpentry for scientists using advanced computational resources. The article also presents the analysis of an earthquake prediction code benchmark while focusing on the accuracy of the results and not only on the runtime; notedly, this benchmark was created as a result of our lessons learned. Energy traces were produced throughout these benchmarks, which are vital to analyzing the power expenditure within HPC environments. Additionally, one of the insights is that in the short time of the project with limited student availability, the activity was only possible by utilizing a benchmark runtime pipeline while developing and using software to generate jobs from the permutation of hyperparameters automatically. It integrates a templated job management framework for executing tasks and experiments based on hyperparameters while leveraging hybrid compute resources available at different institutions. The software is part of a collection called cloudmesh with its newly developed components, cloudmesh-ee (experiment executor) and cloudmesh-cc (compute coordinator).

58 GEOSCIENCES↗

DOE OSTI2022

Solving Newton’s equations of motion with large timesteps using recurrent neural networks based operators

Classical molecular dynamics simulations are based on solving Newton’s equations of motion. Using a small timestep, numerical integrators such as Verlet generate trajectories of particles as solutions to Newton’s equations. We introduce operators derived using recurrent neural networks that accurately solve Newton’s equations utilizing sequences of past trajectory data, and produce energy-conserving dynamics of particles using timesteps up to 4000 times larger compared to the Verlet timestep. We demonstrate significant speedup in many example problems including 3D systems of up to 16 particles.

Newton’s equations↗

NASA NTRS2012

QuakeSim 2.0

QuakeSim 2.0 improves understanding of earthquake processes by providing modeling tools and integrating model applications and various heterogeneous data sources within a Web services environment. QuakeSim is a multisource, synergistic, data-intensive environment for modeling the behavior of earthquake faults individually, and as part of complex interacting systems. Remotely sensed geodetic data products may be explored, compared with faults and landscape features, mined by pattern analysis applications, and integrated with models and pattern analysis applications in a rich Web-based and visualization environment. Integration of heterogeneous data products with pattern informatics tools enables efficient development of models. Federated database components and visualization tools allow rapid exploration of large datasets, while pattern informatics enables identification of subtle, but important, features in large data sets. QuakeSim is valuable for earthquake investigations and modeling in its current state, and also serves as a prototype and nucleus for broader systems under development. The framework provides access to physics-based simulation tools that model the earthquake cycle and related crustal deformation. Spaceborne GPS and Inter ferometric Synthetic Aperture (InSAR) data provide information on near-term crustal deformation, while paleoseismic geologic data provide longerterm information on earthquake fault processes. These data sources are integrated into QuakeSim's QuakeTables database system, and are accessible by users or various model applications. UAVSAR repeat pass interferometry data products are added to the QuakeTables database, and are available through a browseable map interface or Representational State Transfer (REST) interfaces. Model applications can retrieve data from Quake Tables, or from third-party GPS velocity data services; alternatively, users can manually input parameters into the models. Pattern analysis of GPS and seismicity data has proved useful for mid-term forecasting of earthquakes, and for detecting subtle changes in crustal deformation. The GPS time series analysis has also proved useful as a data-quality tool, enabling the discovery of station anomalies and data processing and distribution errors. Improved visualization tools enable more efficient data exploration and understanding. Tools provide flexibility to science users for exploring data in new ways through download links, but also facilitate standard, intuitive, and routine uses for science users and end users such as emergency responders.

Donnellan, Andrea↗

NASA NTRS1997

High Performance Input/Output Systems for High Performance Computing and Four-Dimensional Data Assimilation

The approach of this task was to apply leading parallel computing research to a number of existing techniques for assimilation, and extract parameters indicating where and how input/output limits computational performance. The following was used for detailed knowledge of the application problems: 1. Developing a parallel input/output system specifically for this application 2. Extracting the important input/output characteristics of data assimilation problems; and 3. Building these characteristics s parameters into our runtime library (Fortran D/High Performance Fortran) for parallel input/output support.

Fox, Geoffrey C.↗