Mathis Gerdes

I am a theoretical physicist at MIT, working at the intersection of physics and machine learning. Much of my work aims to answer non-perturbative questions in physics that lie out of reach for traditional methods, by developing ML-based computational approaches faithful to the underlying theory. To that end I build known physical structure (symmetries, geometry, and constraints) into machine-learning methods, with applications ranging from gauge theories on the lattice to the geometry of string compactifications.

Before my current position at MIT, I did my PhD at the University of Amsterdam with Miranda Cheng and Christoph Weniger.