15–18 Dec 2026
Duke University
America/New_York timezone
Artificial Intelligence for the Electron Ion Collider (2026)

Finding New Physics with Actor-Critic Reinforcement Learning Algorithm

Not scheduled
15m
LSRC B101 Love Auditorium (Duke University)

LSRC B101 Love Auditorium

Duke University

Durham, NC 27710
AI/ML for Data Analysis and Theory AI/ML for Data Analysis and Theory

Speaker

Dr Jacky Kumar (member@umontreal.ca;staff@umontreal.ca;employee@umontreal.ca)

Description

Finding new physics (NP) is one of the central challenges in particle physics. The Standard Model Effective Field Theory (SMEFT) provides a model-independent framework to interpret any discrepancies with the Standard Model (SM). However, identifying the relevant SMEFT operators is challenging, because of the enormous operator space and the complex correlations among operator. Consequently, phenomenological analyses typically rely on human intuition to select promising operators, potentially biasing the selection.

I will discuss an actor–critic reinforcement learning (RL) framework for systematically searching the SMEFT operator space. The actor learns a policy for sequentially selecting combinations of operators that can explain the observed anomalies, while the critic evaluates these choices and guides the actor toward increasingly promising regions of the parameter space.

As an application of this method, I will present the CDF (W)-mass anomaly, where the actor–critic agent reproduces known solutions. Substantially, I will also present more challenging case of multiple simultaneous anomalies using a synthetic dataset, demonstrating that the agent can identify combinations of SMEFT operators capable of providing a global explanation of the data. These results establish RL as a powerful method for systematically and efficiently searching for NP.

Author

Dr Jacky Kumar (member@umontreal.ca;staff@umontreal.ca;employee@umontreal.ca)

Presentation materials

There are no materials yet.