A Tool-Grounded AI Agent for Gauge Coupling Unification and Proton Decay

Abstract

We present proton_agent, a tool-grounded AI agent that enables natural-language interaction with computational tools for gauge coupling unification and proton decay in grand unified theories. The agent translates user requests into computational tasks, selects and executes the appropriate tools, and interprets their outputs. All numerical calculations, model evaluations, experimental comparisons, parameter scans, and visualizations are handled by a deterministic scientific backend to ensure numerical accuracy and reproducibility. A central objective of this framework is to enable fully local deployment using free and open-source LLMs, while retaining compatibility with commercial LLM APIs. The framework computes one- and two-loop gauge beta coefficients for user-defined scalar and fermion multiplets and performs gauge coupling evolution through the Standard Model and various intermediate symmetry-breaking stages of $SO(10)$, including $SU(3)_C \times SU(2)_L \times SU(2)_R \times U(1)_{B-L}$, $SU(3)_C \times SU(2)_L \times U(1)_R \times U(1)_{B-L}$, $SU(4)_C \times SU(2)_L \times U(1)_R$, and $SU(4)_C \times SU(2)_L \times SU(2)_R$. For proton decay, the framework evaluates gauge-mediated dimension-six contributions in non-supersymmetric $SU(5)$ and $SO(10)$, as well as dressed dimension-five contributions in their supersymmetric counterparts. It supports benchmark calculations for user-specified parameter points, random and grid-based parameter scans, Markov-chain Monte Carlo exploration, comparisons with experimental limits, and publication-ready two- and three-dimensional visualizations. We describe the agent architecture, demonstrate its capabilities through representative benchmark calculations and regression tests, and discuss its current physics scope.