Simulate, Reason, Decide: Scientific Reasoning with LLMs for Simulation-Driven Decision Making
Yuhan Yang, Ruipu Li, Alexander Rodr\'iguez
Why It Matters
What makes this one worth your time
This work is relevant for AI researchers and engineers interested in improving the transparency and reliability of decision-making systems that rely on scientific simulations.
MechSim enhances LLM-driven systems by enabling mechanism-level reasoning for scientific simulators.
Summary
The paper introduces MechSim, a neuro-symbolic reasoning framework that allows large language models (LLMs) to reason about the mechanisms, assumptions, and execution behavior of scientific simulators, aiming to improve transparency and decision-making reliability in high-stakes domains.
Key contributions
- Introduction of MechSim, a mechanism-grounded neuro-symbolic reasoning framework.
- Representation of simulators through a structured schema for improved reasoning.
- Demonstration of improved explanation quality and decision-making reliability in high-stakes domains.
Notable insights
- The framework uses a shared structured schema to represent simulators, capturing assumptions, variables, and execution traces.
- LLM agents function as constrained reasoning engines to generate evidence-grounded explanations.
Possible limitations
- Not stated in the abstract
Abstract
arXiv:2606.04505v1 Announce Type: new Abstract: Scientific simulators are increasingly being integrated into LLM-driven systems for high-stakes simulation-driven decision-making. However, existing frameworks primarily use LLMs to generate, calibrate, or execute simulators, treating them as black-box interfaces rather than as structured mechanistic systems that can be reasoned about. As a result, current approaches lack the ability to identify, represent, and reason about the assumptions and mechanisms underlying simulator behavior, limiting transparency, auditability, and decision justification. We introduce MechSim, a mechanism-grounded neuro-symbolic reasoning framework for executable scientific simulators. Unlike prior neuro-symbolic approaches that primarily reason over static symbolic structures, MechSim enables LLM agents to reason about the mechanisms, assumptions, and execution behavior of scientific simulators. Our framework represents simulators through a shared structured schema capturing assumptions, variables, mechanism dependencies, and execution traces. On top of this representation, LLM agents operate as constrained reasoning engines that generate structured, evidence-grounded explanations linking simulator outcomes to their underlying mechanisms. We evaluate our approach across multiple high-stakes domains and show that it improves mechanism-level explanation quality, simulator analysis, and downstream decision-making reliability.