The center will unite mathematicians, engineers and computer scientists at Brown, NYU and Georgia Tech to tackle longstanding problems in how simulations handle extreme physical events.
PROVIDENCE, R.I. [Brown University] - With a $5 million grant from the U.S. Department of Energy's National Nuclear Security Administration (NNSA), a Brown University team will lead a new research center aimed at improving computer simulations of fluid flows, blast dynamics and design optimization processes.
The new Center for Information Geometric Mechanics and Optimization (CIGMO) - a collaboration among researchers from Brown, New York University and Georgia Tech- is supported by the NNSA's Predictive Science Academic Alliance Program, which engages academic researchers in projects that align with NNSA's mission to maintain a safe, secure and reliable nuclear stockpile.
"Computer simulations play a key role in planning, technology development and decision-making," said Brendan Keith, an assistant professor of applied mathematics at Brown and CIGMO director. "We're working to make them run better - to be more accurate, more robust, more stable and more efficient."
The research team will explore new approaches to solving problems that arise in computer simulations of extreme physical phenomena, like high-speed collisions, hypersonic flows and explosions. These simulations often violate key model parameters known as inequality constraints, according to Keith.
"Problems involving inequality constraints have existed since the founding days of scientific computing," Keith said. "They arise when simulations of physical systems produce physically impossible quantities like negative air pressures or two solids that pass through each other. These problems can cause code to blow up and algorithms to break, so dealing with them is critical to running efficient and useful simulations."
Traditionally, Keith said, researchers have dealt with these problems on a case-by-case basis - designing custom fixes specific to a given simulation or type of simulation. But CIGMO will explore a more systematic approach, applying information geometry - a way of approaching probabilities and constraints through the lens of shapes and curves.
"We're working to develop a unifying framework that can be applied broadly across fluid and solid mechanics simulations and in other areas," Keith said.
CIGMO will bring together researchers from mathematics, engineering, computer science and other disciplines. Other researchers from Brown who will participate include Yuri Bazilevs and Jerome Darbon.
"Probabilities, like that of winning in the lottery, are necessarily between 0% and 100% - it doesn't make sense to talk about probability that is -10% or 120%," said Florian Schäfer, CIGMO co-director and an assistant professor at NYU. "What I find fascinating is that many inequalities in physical problems arise for the same reason. Thus, a lot can be gained from viewing computer simulation as statistical reasoning about an uncertain physical reality."
The CIGMO research team will concentrate its work in three key simulation areas: High-speed fluid flows like those involved in hypersonic travel, blast and impact dynamics, and design optimization processes like topology optimization that are used in additive design and manufacturing.
Keith says he's hopeful that progress made in these key areas will be applicable to computer simulations across extreme phenomena, advancing both national security and industry applications.
