Agentic Discovery of Exchange-Correlation Density Functionals

2026-05-06

AI Summary

This work addresses the limitations of conventional density functional theory, where exchange-correlation functionals are manually designed and struggle to efficiently explore high-dimensional functional spaces for improved accuracy. The authors propose the first automated functional discovery framework powered by a large language model agent, which integrates evolutionary heuristic search with explicit physical constraints. Within a structured “plan–execute–reflect” iterative loop, the framework systematically modifies functional forms and optimizes parameters while effectively avoiding unphysical solutions. The newly discovered functional, SAFS26-a, demonstrates approximately 9% improved performance over the current gold-standard ωB97M-V on standard thermochemical benchmark sets, validating the approach’s effectiveness in simultaneously achieving physical consistency and predictive accuracy.

Abstract

The development of accurate exchange-correlation (XC) functionals remains a longstanding challenge in density functional theory (DFT). The vast majority of XC functionals have been hand designed by human researchers combining physical insight, exact constraints, and empirical fitting. Recent advances in large language models enable a systematic, automated alternative to this human-driven design loop. This report presents an agentic search system in which an LLM proposes structured functional-form changes guided by evolutionary history. The system attempts to improve functional performance through an iterative plan-execute-summarize loop, where improvements are measurable by optimizing functional parameters against a standard thermochemistry dataset, then evaluating performance on a held-out subset. The strongest discovered functional, SAFS26-a (Seed Agentic Functional Search 2026), improves upon the gold-standard ωB97M-V baseline by ~9%. These results also surface a cautionary lesson for AI-assisted science: models powerful enough to discover genuine improvements are equally capable of exploiting unphysical shortcuts to game the benchmark; domain expertise translated into explicitly enforced constraints remains essential to keeping results scientifically grounded.

Problem

• exchange-correlation functional
• density functional theory
• automated discovery
• functional design
• thermochemistry benchmark

Innovation

• agentic search
• exchange-correlation functional
• density functional theory
• large language model
• automated discovery