OpenTrain AI
No verified implementation yet

Think like a Scientist: Physics-guided LLM Agent for Equation Discovery

Jianke Yang, Ohm Venkatachalam, Mohammad Kianezhad, Sharvaree Vadgama, Rose Yu

February 12, 2026arXiv: 2602.12259
0 repos~a few days to reproduce
arXiv PDF

Abstract

Explaining observed phenomena through symbolic, interpretable formulas is a fundamental goal of science. Recently, large language models (LLMs) have emerged as promising tools for symbolic equation discovery, owing to their broad domain knowledge and strong reasoning capabilities. However, most existing LLM-based systems try to guess equations directly from data, without modeling the multi-step reasoning process that...

Results & Benchmarks

TaskDatasetMetricValue
Agentic tool usePySRNMSE2.82
Agentic tool useKeplerAgent @1NMSE1.50

Hardware Requirements

  • Expect multi-day setup/compute for meaningful reproduction based on current guidance.

Best Implementation

Maintained implementation evidence is not confirmed for this paper yet.

Use the Implementation Status and Reproduction Path sections below for the current action plan.

Reproduction Path

Follow this baseline workflow to decide if this paper is worth immediate prototyping.

  1. 1

    Use the paper and benchmark evidence to scope a baseline reproduction plan.

  2. 2

    Track assumptions and missing details in an experiment log before coding.

Time to first repro: a few daysEstimate is based on paper-only reproduction flow

Additional Implementations

No additional verified repositories beyond the primary recommendation.

Hugging Face Artifacts

No trustworthy direct or curated related Hugging Face artifacts were found yet.