Pacific Northwest National Laboratory (PNNL) Computational Scientist Kasia Świrydowicz received a Best Paper Award from the 29th International Conference on Information, Communication and Automation Technologies (ICAT). The conference was held on June 11-14th in Sarajevo, Bosnia and Herzegovina.
In the paper, Świrydowicz, along with Shrirang Abhyankar—a power engineering senior scientist at PNNL—and collaborators from Oak Ridge National Laboratory, describe a method of optimizing power flow analysis on graphical processing units (GPUs).
“This problem, like many others, essentially boils down to a math problem,” said Świrydowicz, lead author on the paper.
The math Świrydowicz refers to is a branch of linear algebra called linear solvers—her area of expertise. This type of math has a wide variety of applications, from computational fluid dynamics to optimization, but most linear solvers are computationally expensive. Świrydowicz and her co-authors noticed that linear solvers dominated the computational cost for power grid optimizations and developed a technique to speed up the computation on the regular data structures of GPUs.
“Our technique eliminated the main bottleneck for efficient optimizations on GPUs,” said Świrydowicz.
This research was supported by the Department of Energy's Advanced Scientific Computing Research (ASCR) program. Prior to joining PNNL, Świrydowicz was a postdoctoral researcher at the National Renewable Energy Laboratory and Virginia Tech University. She received her PhD in mathematics from Virginia Tech University in 2017.