IMG 20171202 110703472

Northeastern builds bridges between string theory and data science

To those of us who are not theoretical or particle physicists, the idea that something can predict the existence of extra dimensions in space may seem mind-bending. For theoretical high energy physicist and Northeastern Assistant Professor of Physics James Halverson, it’s part of his every day: string theory.  

String theory works to create a fundamental, unified theory of nature that can combine quantum mechanics (the interactions of very small particles in nature) and general relativity (Einstein’s theory of gravity) while explaining other aspects of known physics such as the standard model of particle physics, dark matter, and early universe cosmology. But most importantly, string theory is an ambitious candidate of being the unified theory of everything. One aspect of research in this field, and part of the focus of Halverson’s work, is to study how the possible geometries and shapes of extra dimensions allowed by string theory would affect physics in our world.  

With this research, comes a lot of data. There are rules for the construction of extra-dimensional geometries and electromagnetic fluxes, of which there can be many. To make something of these extremely large numbers, a new subfield in string theory is arising thanks to the work of Halverson and Associate Dean and Associate Professor of Physics Brent Nelson, which combines theoretical physics and string theory, with data science.  

“Understanding the implications of these rules for real world physics demands getting control of the large number of solutions for string theory, as often represented by a large number of geometries,” Halverson said. “That’s where the data science story comes in, since techniques from computer science are moving into fields with large data sets, revolutionizing how we are able to think about those data sets and what we can do with them.” 

With this in mind, Halverson, Nelson, and postdoctoral researcher Cody Long began organizing the first ever meeting where physicists and data scientists could come together to discuss the applications of data science to string theory. And from November 30 to December 2, the 2017 Workshop on Data Science and String Theory was held on Northeastern’s campus in Boston.  

With over 50 participants including professors and scientists from around the world and data scientists from industry, this brand-new meeting brought interdisciplinary collaboration and innovation to string theory and its large datasets.  

“To my knowledge, this is the first time ever that people outside of academia have come to string theory meetings as participants rather than observers,” Halverson said. “String theory is widely recognized as a promising direction for unifying physics. Many data scientists are aware of this, in part due to popular science books and television shows, and some are interested in playing a role in the research. The techniques that they can offer could be crucial to new discoveries, in which case they will lead to publications authored by both communities.” 

Given all the anticipation about this direction, the meeting was well-attended and a huge success. With plenary sessions and breakout groups, both string theorists and data scientists had the opportunity to present, discuss, and brainstorm around their expertise and future plans for the subfield. Breakout groups brought exciting discussions around topics such as network science, supervised machine learning, and deep-reinforcement learning.  

Together with the team from Northeastern, participants from numerous excellent universities and data science firms such Harvard, Oxford, MIT, and CharlesRiver Analytics, pioneered this first meeting that marked the beginning of more to come. This March, Ludwig Maximilian University in Munich is planning a similar meeting, and a third will follow in China this May. Connections were established for projects that pair physicists with data scientists to continue their research.  

“By interfacing with the data scientists directly, many string theorists came up with new projects that are likely to turn into publications in the next six months,” said Halverson. “For Northeastern, we are now established as a leader in bringing the people of data science and string theory together. We are some of the first movers of this very interdisciplinary front, and are looking forward to ways in which data science could influence fundamental theoretical physics.” 

Workshop session. Photo courtesy of James Halverson.