跳到正文
返回论文列表
cs.CE提交于 已译

架构保障可靠性:从自然语言到经验证多物理仿真的智能体工作流

Reliability by Architecture: Agentic Workflows for Validated Multi-Physics Simulation from Natural Language

S{\o}ren Taverniers · Eric Darve · Hamdi A. Tchelepi

中文摘要

准备、执行并验证强耦合多物理仿真需要专家投入大量时间。智能体工作流(agentic workflow)利用大语言模型(LLM)从自然语言描述协调这些流程,但可靠性要求对物理一致性和问题描述保真度进行检查。该智能体框架(harness)实现显式任务分配:LLM 解读用户请求与仿真器诊断信息,生成预定义构建规则未覆盖的输入组件,并提出修复方案;确定性例程则构建受支持的组件,并检查输入、编辑和结果。使用多物理仿真器 GEOS 评估 CO2 封存问题,涵盖从组分流动到含摩擦断层接触的流动-孔隙力学。在两个流动问题的 80 次尝试中,接地(grounding)将结构有效输入的比例从 5/20 提升至 20/20,但只有配合完整的基于规则的输入准备,接受率才达到 20/20。在相同输入下,局部修复纠正了三个预置缺陷,同时保留有效输入。辅助测试识别出通过物理检查但与问题描述或参考结果存在差异的仿真。

关键要点

  1. 01LLM 负责解读请求、生成规则外输入并提议修复,确定性例程负责构建与校验,实现显式任务分工。
  2. 02在 GEOS 的 CO2 封存问题上,接地将结构有效输入从 5/20 提升至 20/20。
  3. 03仅靠接地不足以达到 20/20 接受率,需完整基于规则的输入准备。
  4. 04局部修复可纠正三个预置缺陷且不破坏有效输入。
  5. 05辅助测试能发现通过物理检查但偏离问题描述或参考结果的仿真。

解读

尚无解读。

原始英文摘要

arXiv:2611.23473v1 Announce Type: cross Abstract: Preparing, executing, and validating strongly coupled multi-physics simulations require expert time. Agentic workflows use large language models (LLMs) to coordinate them from natural-language descriptions, but reliability requires checks of physical consistency and fidelity to the problem description. Our agentic harness implements explicit task allocation: LLMs interpret user requests and simulator diagnostics, generate input components not covered by predefined construction rules, and propose repairs, while deterministic routines construct supported components and check inputs, edits, and results. We evaluate CO2-storage problems using the multi-physics simulator GEOS, from compositional flow to flow-poromechanics with frictional fault contact. In 80 attempts on two flow problems, grounding raises the proportion of structurally valid inputs from 5/20 to 20/20, but acceptance reaches 20/20 only with complete rule-based input preparation. With identical inputs, local repairs correct three seeded defects while preserving valid inputs. Supporting tests identify simulations passing physics checks despite differing from the problem description or reference results.

同方向论文 · cs.CE

查看全部 →