|
Résumé |
I will present a novel numerical treatment of the truncated conformal-bootstrap equations that employs Reinforcement Learning algorithms. A particularly interesting algorithm that can handle multi-dimensional searches in continuous spaces is the soft Actor-Critic algorithm. I will report results of preliminary applications to two-dimensional CFTs, where the Reinforcement-Learning agent was asked to identify well-known theories like the 2D Ising and tri-critical Ising models, or to track CFT data on the conformal manifold of the compactified scalar on a circle. The method can be used to study arbitrary (unitary or non-unitary) CFTs in any spacetime dimension. I will discuss the advantages and disadvantages of the approach, and its future prospects. |
