Topology-Preserving Data Sketching for Scientific Visualization
FY 2020 DOE Office of Science Early Career Research Program

Award Number and Duration

DOE DE-SC0021015

September 1, 2020 to August 31, 2025 (Estimated)

PI and Point of Contact

Bei Wang Phillips (Publish Under Bei Wang)
Assistant Professor
School of Computing and Scientific Computing and Imaging Institute
University of Utah
beiwang AT sci.utah.edu
http://www.sci.utah.edu/~beiwang

Overview

We are experiencing an information overload from streams of data that arise from scientific instruments and simulations. For example, material scientists use molecular dynamics (MD) simulations to study how fluids (such as gas, oil, and water) interact with heterogeneous porous solids (such as ceramics, cement, and rock) to improve transport phenomena within porous materials, which play critical roles in our energy sector. Such simulations generate large, time-varying, and complex forms of data under different physical and chemical conditions. Keeping track of interesting phenomena and applying appropriate actions (such as storage, analysis, and visualization) while the simulation is running is necessary but challenging. To address this challenge, the goal is no longer to capture and store observations or simulation in detail, but rather to process data efficiently and approximately in order to create a summary - a sketch - which allows queries over large volumes of data to be answered quickly.

The objective of this research is to conduct a systematic study of topology-preserving data sketching techniques to improve visual exploration and understanding of large scientific data. The project will employ topological sketches, that is, compressed representations of the full data that preserve their important structural properties, to support analysis and visualization as the data are generated. Our proposed solution transforms data sketching ideas from statistics, geometry, and linear algebra to develop new topological sketches of complex data. Such sketches will exploit the high spatial resolution and temporal fidelity of in situ data in an intelligent and scalable way. They will reduce data in situ while preserving its structural properties, and subsequently support interactive data exploration. In addition, topological triggers will be integrated into an adaptive workflow to support anomaly detection, computational steering, and decision optimization. The multidisciplinary nature of the proposed work will be broadly applicable in many scientific areas, including applications in computational fluid dynamics and materials science.

Publications and Manuscripts

Manuscripts

Software Downloads

Presentations, Educational Development and Broader Impacts

1. Bei Wang Invited Talk (virtual) at CAM Colloquium, Committee on Computational and Applied Mathematics (CCAM), University of Chicago, March 11, 2021. Title: Sketching Merge Trees.
   

2. Workshop: Application Spotlights: Challenges in the Visualization of Bioelectric Fields for Cardiac and Neural Research at IEEE VIS, October 25-30, 2020.
   Organizers: Bei Wang, Rob MacLeod, Wilson Good.

Students

Collaborators

Dr. Pania Newell, Assistant Professor, Department of Mechanical Engineering; University of Utah.

Dr. Anastasia Ilgen, Sandia National Lab.

Dr. Roxana Bujack, Los Alamos National Laboratory.

Dr. Emilie Purvine, Pacific Northwest National Laboratory.

Dr. Gunther Weber, Lawrence Berkeley National Laboratory.

Acknowledgement

This material is based upon work supported or partially supported by the United States Department of Energy (DOE) under Grant No. DE-SC0021015.

Any opinions, findings, and conclusions or recommendations expressed in this project are those of author(s) and do not necessarily reflect the views of the DOE.

Web page last update: March 13, 2021.