SimAuthor: Harnessing Foundation Models for Persistent Scientific Simulator Authoring

2026-10-05

AI Summary

This study addresses the challenge of iteratively constructing scientific simulators using foundation models under weak empirical feedback. To this end, it proposes a persistent code-writing framework that translates model prior knowledge into executable simulators. The core strategies include decoupling scoring from structured feedback to precisely guide revisions, introducing a distribution-matching evaluation mechanism, and designing an experience accumulation and reuse strategy to enable efficient iterative optimization. Evaluated across six biomedical tasks, the proposed approach significantly outperforms existing baselines, effectively enhancing generalization to unseen data and improving downstream classification performance. This work establishes a novel paradigm for scientific simulation in data-limited scenarios.

Abstract

Foundation models can generate scientific code, but authoring a scientific simulator (an executable program encoding hypotheses about how mechanisms generate observable signals) requires iterative refinement. Scientific adequacy rarely admits a unique implementation or exact test, so simulators must instead be judged against limited real observations. We study this setting as scientific simulator authoring under weak empirical feedback, where distributional comparisons between simulated and real signals guide revision, and the target is the simulator itself rather than only its generated samples. We introduce SimAuthor, a persistent authoring harness that retains and revises executable simulators, separates scalar search scores from structured discrepancy feedback, and accumulates reusable implementation mechanisms. We evaluate SimAuthor on six biomedical tasks spanning cardiac and respiratory audio, photoplethysmography (PPG), and electrocardiography (ECG). Under a fixed 100-attempt budget, SimAuthor outperforms PUCT score search on all six tasks, generally outperforms textual-strategy optimization, and achieves the highest endpoint score on five of six. The authored simulators also improve on unseen recordings, transfer to independent pretrained representations, and yield substantial out-of-distribution gains in downstream ECG classification. Finally, 111 of 138 audited revisions alter program structure and account for 86.1% of the signed score improvement. These results suggest that persistent revision can progressively convert foundation-model knowledge into better executable scientific simulators from limited empirical evidence.

Problem

• Scientific Simulator Authoring
• Foundation Models
• Weak Empirical Feedback
• Iterative Refinement

Innovation

• Foundation Models
• Scientific Simulator Authoring
• Persistent Revision
• Weak Empirical Feedback
• Out-of-Distribution Generalization