Benchmark RadarAI BENCHMARK PROFILE
StatMechBench-v0
A benchmark of six Ising-type problems for evaluating LLM agents' ability to discover statistical mechanical mappings from raw partition functions.
- Released
- 2026-07-29
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
This probes AI capabilities for structural discovery in theoretical physics, highlighting limitations in current LLM reasoning and the need for verification beyond numerical agreement.
Motivation
An important skill in theoretical physics is to recognize when a new problem can be transformed into a known model.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.