With artificial intelligence to new physics
The particle accelerator of CERN, the European Organization for Nuclear Research, produces large amounts of complex data with high accuracy, as a result of which theoretical predictions also need to be more accurate and faster. With his research team, PhD candidate Ben Ruijl developed very innovative methods for predicting theoretical outcomes. Thursday November 2nd, Ruijl defends his thesis Advances in Computational Methods for Quantum Field Theory Calculations.
With the Large Hadron Collider in Geneva physicists try to find out more about elementary particles, the smallest building blocks of matter. The computation of physical theories is complicated and time consuming. PhD student Ben Ruijl, associated with the Leiden Institute of Advanced Computer Science (LIACS), has succeeded to significantly improve this process.
New software requires less computing power
The physical theories are translated into mathematical formulas, which predict high precision measurements of particle accelerators. Using techniques from artificial intelligence, such as the search algorithm Monte Carlo Tree Search, Ruijl developed methods to simplify the formulas. These methods apply to all sorts of theories. Ruijl has made his techniques available through the open source program FORM. With this software researchers worldwide can do complex calculations with less computing power, memory, electric current and financial resources. One computer now carries out a calculation in six days that previously required twenty computers for a year and a half.
New level of precision calculations
According to Jos Vermaseren, co-promoter, Ruijl's research is of great value, as it enables research into minimal deviations in the standard particle physics model. Precision calculations are important when discovering new physics, and Ruijl's research takes the current state of those calculations to the next level.
Ben Ruijl’s research is part of the HEPGAME program, in which informatics researchers at Leiden University collaborate with theoretical physicists of the National Institute for Subatomic Physics (Nikhef). The project deploys the latest artificial intelligence insights for solving complex physical issues. Ruijl’s PhD supervisors are Jaap van den Herik (LCDS), Aske Plaat (LIACS) and Jos Vermaseren (Nikhef).