Assistant Professor Pavel Panchekha and Professor Ganesh Gopalakrishnan have received a multi-institutional Community Infrastructure for Research in Computer and Information Science and Engineering (CIRC) grant from the National Science Foundation (NSF) for a research project expected to improve real-world numerical software packages by allowing them to operate faster and more reliably across platforms.
The collaborative research project, “Workbench for Reliable and Efficient Numerics”, has received a continuing grant valued at $1,998,953 ($1,158,953 for Utah; the collaborating efforts are lead by Professors Cindy Rubio-González at the University of California Davis and Zachary Tatlock at the University of Washington). This will allow Panchekha and Gopalakrishnan as Principal and Co-Principal Investigators to create a workbench for scientists and engineers to better address numerical issues in their day-to-day work.
“Numerical issues are issues caused by the gap between mathematical (real) numbers and the number representations used on computers, like floating point. Ultimately, this gap makes it difficult for scientists and engineers to develop software that does numerical computation accurately and runs reliably and efficiently on a variety of hardware and software platforms,” states the project abstract. “Over the years, the research community has studied these issues and developed a number of tools that make developing numerical software easier, but these tools have become difficult to use together.”
The project builds on existing FPBench standardization and interoperability efforts to address the complexities of real-world numerical workflows. The outcomes of this research project are expected to simplify this process for national laboratories, industry members, and academics alike, as well as build community between these sectors through initiatives such as community meetings, workshops, and Research Experiences for Undergraduates (REUs).
Numerical computing is behind scientific simulations, engineering calculations, and financial models. It’s also at the core of high-tech advances like artificial intelligence. “We’re always demanding that our programs are faster, more capable, and more reliable. There are really exciting advances in the last few years in how we do that for numerical programs, and this grant is about taking those advances and putting them, via newly designed tooling, in the hands of users,” said Prof. Panchekha. He plans to build on his experience working with scientists at national labs and researchers in industry to make numerical work easier. “It’s all about bringing software, and the researchers behind it, together to make something people can use,” he added.
To view the full award abstract from NSF.gov, click here.