Abstract
In this paper, we present SOCIA-Nabla, an end-to-end, agentic framework that treats simulator construction asinstance optimization over code within a textual computation graph. Specialized LLM-driven agents are embedded as graph nodes, and a workflow manager executes a loss-driven loop: code synthesis -> execution -> evaluation -> code repair. The optimizer performs Textual-Gradient Descent (TGD), while human-in-the-loop interaction is reserved for task-spec confirmation, minimizing expert effort and keeping the code itself as the trainable object. Across three CPS tasks, i.e., User Modeling, Mask Adoption, and Personal Mobility, SOCIA-Nabla attains state-of-the-art overall accuracy. By unifying multi-agent orchestration with a loss-aligned optimization view, SOCIA-Nabla converts brittle prompt pipelines into reproducible, constraint-aware simulator code generation that scales across domains and simulation granularities. This work is under review, and we will release the code soon.
Abstract (translated)
在这篇论文中，我们介绍了SOCIA-Nabla，这是一种端到端、具备代理能力的框架，它将模拟器构造视为文本计算图中的代码实例优化问题。该框架中嵌入了专门由大型语言模型驱动的代理作为图节点，并且一个工作流管理器执行了一个以损失为导向的循环：代码合成 -> 执行 -> 评估 -> 代码修复。优化器采用文本梯度下降（Textual-Gradient Descent, TGD），而人机交互则保留用于任务具体要求确认，从而将专家的努力最小化并将代码本身作为可训练的对象。在三个CPS（Cyber-Physical Systems）任务中，即用户建模、口罩采用和个人移动性方面，SOCIA-Nabla 达到了最先进的总体准确性。通过将多代理编排与损失对齐的优化视角统一起来，SOCIA-Nabla 将脆弱的提示管道转化为可重复且具有约束意识的模拟器代码生成过程，并且该方法可以跨领域和模拟粒度进行扩展。这项工作正在审稿中，我们很快会发布代码。
URL
https://arxiv.org/abs/2510.18551
