Auto-Configuring Scientific Simulators with Lightweight Coding-Agent Adapters

A lightweight adapter called SIGA can cut the time domain scientists spend configuring advanced simulators from hours to roughly five minutes, according to a preprint posted to arXiv. The adapter wraps an off-the-shelf coding agent, supplying the simulator’s executable contract without rebuilding the agent’s core loop. Configuring an advanced scientific simulator—translating a modeling goal into a valid, runnable input deck—is a persistent bottleneck that costs domain scientists hours to days [1][2]. Input decks function as executable interfaces: simulator-specific vocabulary, cross-file references, schema constraints, and validation rules must align before a simulation can run [2]. The preprint, authored by Matthew Ho and submitted to arXiv on June 8, 2026, introduces SIGA, a coding-agent adapter that supplies this contract through retrieval, procedural memory, agent-callable validation, and validation-gated termination while leaving the underlying model and loop frozen [1][2]. Coding agents already navigate files, edit code, run commands, and repair outputs; what they lack is the simulator’s executable contract, and rebuilding the agent loop risks discarding harness-calibrated tool-use and self-correction behavior [2]. Because SIGA’s contract is small and external, the adapter supports self-evolution: prior trajectories can rewrite the adapter contents without modifying the underlying agent [2]. On GEOS, a multiphysics subsurface simulator, SIGA’s main gain is reliability. On harder held-out tasks it improves TreeSim from 0.720 to 0.789 and reduces across-run standard deviation by about 16x by preventing empty or invalid decks [1][2]. In a human calibration, SIGA reaches in about five minutes the deck quality a domain expert reached in about three hours [1][2]. Transfers to OpenFOAM and LAMMPS show the recipe is portable but interface-dependent: completion gates help when structural completeness is the bottleneck, while memory and retrieval help when value correctness is [2]. The paper appeared on arXiv, an open-access repository of electronic preprints that is not peer-reviewed but is moderated before posting [6]. As of November 2024, the repository receives about 24,000 articles per month and has surpassed two million total articles [6]. The SIGA preprint was last revised on June 25, 2026 [1].