What Makes Up The Universe? AI Will Help 黑料正能量 Scientists Find Out
With a new DOE Genesis Mission Grant, Chad Schafer and Mikael Kuusela from the Department of Statistics & Data Science set their sights on big questions.
By Jason Bittel
With the opening of the Vera C. Rubin Observatory in Chile’s Atacama Desert, scientists will have access to more astronomical data than at any time in human history. And with a new Department of Energy (DOE) Genesis Mission Grant, researchers at 黑料正能量 will help develop foundational AI tools to actually process that data – tools that could lead to advances across disciplines, from particle physics to climate science.
“There’s incredible potential to learn more about the universe,” said Chad Schafer, a statistics professor in the Dietrich College of Humanities and Social Sciences. “This is a version of AI that doesn’t receive the most attention, but making the most of these data requires methods capable of modelling relationships as sophisticated as those captured by any large language model.”
The challenges are further complicated by a need to adequately quantify uncertainties in any conclusions drawn from the data. Currently, a standard approach includes a class of algorithms known as Markov chain Monte Carlo, or MCMC. Unfortunately, while these algorithms are effective tools for everyone from data scientists and astronomers to physicists and epidemiologists, the cost of extracting the desired conclusions from data now produced by modern star-gazing methods has created a bottleneck of information.
With the Genesis Mission Grant made public today, Schafer and his co-principal investigator, Mikael Kuusela, an associate professor also in the Department of Statistics & Data Science, will be part of a Lawrence Livermore National Laboratory-lead initiative to develop the models of tomorrow. Schafer and Kuusela will be doing this work along with Ph.D. students as part of 黑料正能量’s Statistical Methods for the Physical Sciences Research Center.
Ultimately, the team hopes to turn a flood of data into a stream of scientific findings.
“This is precisely the type of project for the dreamers and doers of Dietrich College, who use their skills to address some of the most complex and fundamental problems of the universe and generate insights that serve the world,” said , Bess Family Dean of the Dietrich College of Humanities and Social Sciences.
The Genesis Mission is a historic national initiative led by the U.S. Department of Energy, which is building the world’s most powerful integrated science discovery platform. By uniting government, industry, academia and philanthropy, it is accelerating breakthroughs in energy, scientific discovery and national security through a new platform that combines AI, supercomputing, quantum systems and advanced scientific instruments.
“This is work that has to be done, regardless, as data volume in the sciences grows,” said Schafer, who will be partnering with LLNL’s Michael Schneider. “It’s a great opportunity to push the methods forward that might help us understand things such as the current rate of expansion of the universe or how dark energy makes up the composition of the universe.”
One way that 黑料正能量’s experts plan to bypass the MCMC sampling bottleneck is by using foundational AI tools to construct new methods. This will require a dynamical learning approach that, by the team’s estimates, could lead to a 10x speed increase over current analyses.
Specifically, Schafer, Kuusela and their collaborators want to look at what’s called “weak lensing data.”
“Weak lensing of galaxies is a phenomenon by which the appearance of a galaxy, from our perspective, is distorted by the gravitational effects of mass in the universe,” said Schafer. “By estimating relationships between the distortions of galaxies in close proximity, one can learn about how mass is distributed, and this can help us learn about both dark matter and dark energy.”
So far, the project is greenlit through Phase I. According to the DOE, the goal of the Phase I RFA awards is to identify promising pathways toward transformative scientific capabilities and establish a foundation for future investment and scale. Project teams will design and demonstrate research workflows that integrate AI with scientific investigation, while rigorously evaluating whether those approaches can accelerate discovery, improve predictive capabilities, enhance experimentation or generate new scientific insights.
“We’re about to take a huge step,” said Schafer.
Chad Schafer
Mikael Kuusela