$2 Million Gift from the DRW Foundation Enables Caltech to Advance Mathematics and AI Research
The DRW Foundation's generous support empowers Caltech's Division of Physics, Mathematics and Astronomy (PMA) to advance machine learning and artificial intelligence while applying these tools to catalyze breakthroughs in mathematical theory that were once thought to be restricted to or even beyond the scope of humans.
The project will be undertaken by a team of postdoctoral fellows, staff research scientists, software engineers, and students who will be assembled and led by Sergei Gukov, John D. MacArthur Professor of Theoretical Physics and Mathematics and director of the Richard N. Merkin Center for Pure and Applied Mathematics. While the work holds the potential to solve math problems that have puzzled generations of mathematicians, the prospective benefits could extend beyond the abstract and theoretical. Several of these unsolved problems are directly relevant to other fields including environmental science and developmental biology.
If successful, Gukov's team will develop artificial general intelligence, an advanced form of AI that exhibits characteristics of human intuition and innovation with the ability to autonomously learn, adapt, and imagine novel solutions to complex problems. In so doing, the project may help refine and accelerate the art of problem solving for countless practical applications.
"The DRW Foundation's generous gift emphasizes the possibilities for AI and machine learning to decipher fundamental constructs in mathematics," says Caltech President Thomas F. Rosenbaum, the Sonja and William Davidow Presidential Chair and professor of physics. "This exciting initiative brings together scientists and engineers with different skill sets under Sergei's leadership in a way that accentuates Caltech's strengths."
Established in 2004, the DRW Foundation is the philanthropic arm of DRW, a global trading firm that applies modeling, technology, and quantitative research to identify opportunities in global financial markets. DRW championed the use of robust modeling when it was founded in 1992 by CEO Donald R. Wilson, Jr., and it remains engaged in the evolution of technology to advance trading and other spheres.
"I am incredibly excited about DRW's partnership with Caltech, particularly about collaborating with Professor Gukov," Wilson says. "His work rethinks conventional wisdom with respect to machine learning model creation and paves the way for the future across mathematics, machine learning, and markets. I look forward to the learning partnership between Professor Gukov's group and the AI Research team at DRW."
In addition to its philanthropy, DRW is helping raise awareness on campus about careers in quantitative trading, for which the rigors of a Caltech education provide excellent preparation. Last academic year, DRW representatives began connecting with the Institute's students in career advising and recruitment events. DRW is a partner in the CMS-EE Partners Program through which companies engage directly with students and researchers, while also supporting educational and outreach efforts ranging from hackathons and programming competitions to K-12 events and diversity initiatives.
"We recognize the talent and innovation that Caltech students bring to the table," Wachi Bandara, DRW's Head of Artificial Intelligence, says. "Our relationships with Caltech's faculty and students have only reinforced our belief that these individuals possess the extraordinary curiosity and research skills required to push the boundary of modern machine learning."
Current generative AI tools like ChatGPT are not yet able to adapt to new situations and information in the way people can. Without more human-like problem solving abilities, AI's applications and benefits are limited.
To develop AI architecture that can learn, adjust, and reason, Gukov's team has planned an iterative process, beginning by training it to solve average research-level problems—like those that can be found in a reputable math journal—before moving on to a class of math problems that are considered to be extremely difficult.
Finally, the team aims to address Millennium-Prize problems, the most complex unsolved problems in mathematics. To solve problems like these, AI that can reason better and faster than any previous or current human mathematician will need to be developed.
The intersection of pure mathematics and machine learning is a quickly developing field in which Gukov already has made an impact, having used machine learning to untangle problems in knot theory.
"Machine learning tools have the ability to quickly search through many potential solutions," Gukov says. "By enhancing these tools to adapt during the learning process and dynamically select the most suitable algorithms and architectures for specific problems, we can expect scientific knowledge and technological advances to grow faster than ever before. And this, we hope, will lead to solutions to some of the greatest challenges facing humanity."
DRW's gift enables Gukov to leverage the research environment at the Institute, which includes porous boundaries between disciplines and leading expertise in relevant fields, a new home for the American Institute of Mathematics, and highly talented and motivated students, postdocs, and faculty members to pursue the most challenging questions quickly and nimbly. The gift also will create a graduate degree specialization in mathematics and machine learning, preparing new leaders in the field.
"DRW's support helps build upon the strengths of the Division of Physics, Mathematics and Astronomy," says Fiona A. Harrison, the Harold A. Rosen Professor of Physics and holder of the Kent and Joyce Kresa Leadership Chair in PMA. "It will help advance the fields of mathematics and machine learning while providing our postdocs and students cutting-edge learning and research opportunities."
More about the DRW Foundation can be found on its website.