cs.AIMay 17, 2025

SOCIA-∇\nabla: Textual Gradient Meets Multi-Agent Orchestration for Automated Simulator Generation

Authors: Yuncheng Hua, Sion Weatherhead, Mehdi Jafari, Hao Xue, Flora D. Salim

Organizations: School of Computer Science Engineering, University of New South Wales, Australia

Abstract

In this paper, we present SOCIA-∇\nabla, a diagnosis-driven multi-agent framework for automated simulator-code construction. SOCIA-∇\nabla iteratively refines executable simulator code within a textual computation graph, where LLM-driven agents synthesize, execute, evaluate, diagnose, and repair simulators. A grounded simulator blueprint attributes behavioral residuals to candidate mechanisms and their spatial or interaction contexts, which are then converted into localized textual repair signals with history-aware memory and constraint projection. We evaluate SOCIA-∇\nabla on the real-world LLMob personal-mobility benchmark and a controlled semi-synthetic geo-social mask-adoption benchmark. Across in-distribution simulation, out-of-distribution what-if generation, and geo-social diffusion, SOCIA-∇\nabla achieves the strongest overall simulator fidelity. SOCIA-∇\nabla's code and data are publicly available at https://github.com/cruiseresearchgroup/SOCIA-Nabla.

Figures & tables

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. SOCIA-EVO: Automated Simulator Construction via Dual-Anchored Bi-Level Optimization

    Apr 19, 2026Yuncheng Hua, Sion Weatherhead, Mehdi Jafari +2Evolutionary OptimizationLarge Language Model-Based Simulation

  2. EASE Configuration Facilitates A Reproducible Science of LLM Social Simulations

    May 28, 2026Sneheel Sarangi, Maximilian Puelma Touzel, Aurélien Bück-Kaeffer +3Multi-Agent LLM SystemsHuman Behavior Simulation

  3. AgoraSim: A Hybrid Agent-Based Modeling Framework

    Jul 7, 2026Chung-Chi ChenHuman Behavior SimulationLLM Agent Orchestration