Speaker
Costis Papageorgakis
(Queen Mary University of London)
Description
I will present a new bootstrap method for conformal field theories at finite temperature that uses neural networks to capture infinite operator contributions without positivity constraints or truncations. The approach combines the KMS condition, thermal dispersion relations, and neural networks that learn the high-spin behaviour of the thermal block expansion. I will demonstrate the method on generalised free fields and apply it to thermal corrections in holographic CFTs. This neural network approach bypasses traditional computational bottlenecks in finite temperature bootstrap calculations.
Author
Costis Papageorgakis
(Queen Mary University of London)
