SOCIA-Nabla: Textual Gradient Meets Multi-Agent Orchestration for Automated Simulator Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hua, Yuncheng, Weatherhead, Sion, Jafari, Mehdi, Xue, Hao, Salim, Flora D.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908639883689984
author Hua, Yuncheng
Weatherhead, Sion
Jafari, Mehdi
Xue, Hao
Salim, Flora D.
author_facet Hua, Yuncheng
Weatherhead, Sion
Jafari, Mehdi
Xue, Hao
Salim, Flora D.
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18551
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SOCIA-Nabla: Textual Gradient Meets Multi-Agent Orchestration for Automated Simulator Generation
Hua, Yuncheng
Weatherhead, Sion
Jafari, Mehdi
Xue, Hao
Salim, Flora D.
Artificial Intelligence
I.2.7
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.
title SOCIA-Nabla: Textual Gradient Meets Multi-Agent Orchestration for Automated Simulator Generation
topic Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2510.18551