用双锚点优化自动构建高保真仿真器，解决长期推理中的偏差与不稳问题。
SOCIA-EVO: Automated Simulator Construction via Dual-Anchored Bi-Level Optimization
- 分层优化：分离结构修正与参数校准，避免误差混淆
- 生成的仿真器在统计上与真实数据一致，验证了保真度
- 适合需要高可靠性仿真的科研与工程场景
自动化仿真器构建要求分布保真，不同于通用代码生成。我们发现长时程大模型智能体存在两类失效模式：上下文漂移和因结构与参数误差混杂导致的优化不稳定性。为此提出SOCIA-EVO，一种双锚点演化框架。该框架引入：(1) 静态蓝图以施加经验约束；(2) 双层优化机制，解耦结构精炼与参数校准；(3) 自我更新的策略手册，通过贝叶斯加权检索管理修正假设。通过执行反馈否定无效策略，SOCIA-EVO实现稳健收敛，生成在统计上与观测数据一致的仿真器。代码与数据已公开：https://github.com/cruiseresearchgroup/SOCIA/tree/evo。
原文摘要 · Abstract (English)
Automated simulator construction requires distributional fidelity, distinguishing it from generic code generation. We identify two failure modes in long-horizon LLM agents: contextual drift and optimization instability arising from conflating structural and parametric errors. We propose SOCIA-EVO, a dual-anchored evolutionary framework. SOCIA-EVO introduces: (1) a static blueprint to enforce empirical constraints; (2) a bi-level optimization to decouple structural refinement from parameter calibration; and (3) a self-curating Strategy Playbook that manages remedial hypotheses via Bayesian-weighted retrieval. By falsifying ineffective strategies through execution feedback, SOCIA-EVO achieves robust convergence, generating simulators that are statistically consistent with observational data. The code and data of SOCIA-EVO are available here: https://github.com/cruiseresearchgroup/SOCIA/tree/evo.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。