Researchers from TU Wien, US, and Swiss institutions have used AI to solve a decades-old problem in particle physics: how to optimally formulate quantum field theories on a lattice for computer simulation. They developed a specialized neural network that parameterizes quantum field theories while respecting physical laws, enabling accurate simulations even on coarse computational grids. The approach uses fixed-point equations to ensure properties remain consistent across different lattice resolutions, dramatically reducing computational requirements for complex quantum field theory simulations.
Table of contents
In the computer, the whole world is a gridDifferent scale, same resultThe success of AI48 Impressions