Discovery of Interpretable Surrogates via Agentic AI: Application to Gravitational Waves
Abstract
Fast surrogate models for expensive simulations are now essential across the sciences, yet they typically operate as black boxes. We present \texttt{GWAgent}, a large language model (LLM)-based workflow that constructs interpretable analytic surrogates directly from simulation data. Surrogate modeling is well suited to agentic workflows because candidate models can be quantitatively validated against ground-truth simulations at each iteration. As a demonstration, we build a surrogate for gravitational waveforms from eccentric binary black hole mergers. We show that providing the agent with a physics-informed domain ansatz substantially improves output model accuracy. The resulting analytic surrogate attains a median Advanced LIGO mismatch of 6.9×10−4 together with an ∼ 8.4× speedup in waveform evaluation, surpassing both symbolic regression and conventional machine learning baselines. Beyond producing an accurate model, the workflow identifies compact physical structure from the learned representation. As an astrophysical application, we use \texttt{GWAgent} to analyze the eccentricity of GW200129 and infer e20Hz=0.099+0.063−0.044. These results show that validation-constrained agentic workflows can produce accurate, fast, and interpretable surrogates for scientific simulations and inference.
Related
- An agentic framework for gravitational-wave counterpart association in the multi-messenger era
- Agentic AI for Gravitational Wave Data Analysis: A Head-to-Head Comparison of Coding Agents Executing a Matched Filter Pipeline on Einstein Telescope Simulated Data
- Causal Discovery in Photospheric Magnetic Field Time Series for Interpretable Solar Flare Prediction
- Latin hypercube sampling to reduce the variance in Monte Carlo integration: a practical guide and an application to gravitational-wave astronomy
- Analytic Marginalization Over Calibration Uncertainties in Gravitational Wave Inference and Application to Black Hole Ringdown
BibTeX
@misc{islam2026discovery,
title = {{Discovery of Interpretable Surrogates via Agentic AI: Application to Gravitational Waves}},
author = {Tousif Islam and Digvijay Wadekar and Tejaswi Venumadhav and Matias Zaldarriaga and Ajit Kumar Mehta and Javier Roulet and Barak Zackay},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2605.11280},
eprint = {2605.11280},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.11280},
}
RIS
TY - UNPB
TI - Discovery of Interpretable Surrogates via Agentic AI: Application to Gravitational Waves
AU - Tousif Islam
AU - Digvijay Wadekar
AU - Tejaswi Venumadhav
AU - Matias Zaldarriaga
AU - Ajit Kumar Mehta
AU - Javier Roulet
AU - Barak Zackay
PY - 2026
DA - 2026/09/20
PB - arXiv
DO - 10.48550/arxiv.2605.11280
UR - https://arxiv.org/abs/2605.11280
AB - Fast surrogate models for expensive simulations are now essential across the sciences, yet they typically operate as black boxes. We present \texttt{GWAgent}, a large language model (LLM)-based workflow that constructs interpretable analytic surrogates directly from simulation data. Surrogate modeling is well suited to agentic workflows because candidate models can be quantitatively validated against ground-truth simulations at each iteration. As a demonstration, we build a surrogate for gravitational waveforms from eccentric binary black hole mergers. We show that providing the agent with a physics-informed domain ansatz substantially improves output model accuracy. The resulting analytic surrogate attains a median Advanced LIGO mismatch of 6.9×10^−4 together with an ∼ 8.4× speedup in waveform evaluation, surpassing both symbolic regression and conventional machine learning baselines. Beyond producing an accurate model, the workflow identifies compact physical structure from the learned representation. As an astrophysical application, we use \texttt{GWAgent} to analyze the eccentricity of GW200129 and infer e_20Hz=0.099^+0.063_−0.044. These results show that validation-constrained agentic workflows can produce accurate, fast, and interpretable surrogates for scientific simulations and inference.
ER -