OptiRepair: Closed-Loop Diagnosis and Repair of Supply Chain Optimization Models with LLM Agents
arXiv:2602.19439v1 Announce Type: new Abstract: Problem Definition. Supply chain optimization models frequently become infeasible because of modeling errors. Diagnosis and repair require scarce OR expertise: analysts must interpret solver diagnostics, trace root causes across echelons, and fix formulations without sacrificing operational soundness. Whether AI agents can perform this task remains untested. Methodology/Results. OptiRepair splits this task into a domain-agnostic feasibility phase (iterative IIS-guided repair of any LP) and a domain-specific validation phase (five rationality checks grounded in inventory theory). We test 22 API models from 7 families on 976 multi-echelon supply chain problems and train two 8B-parameter models using self-taught reasoning with solver-verified rewards. The trained models reach 81.7% Rational Recovery Rate (RRR) -- the fraction of problems resolved to both feasibility and operational rationality -- versus 42.2% for the best API model and 21
arXiv:2602.19439v1 Announce Type: new Abstract: Problem Definition. Supply chain optimization models frequently become infeasible because of modeling errors. Diagnosis and repair require scarce OR expertise: analysts must interpret solver diagnostics, trace root causes across echelons, and fix formulations without sacrificing operational soundness. Whether AI agents can perform this task remains untested. Methodology/Results. OptiRepair splits this task into a domain-agnostic feasibility phase (iterative IIS-guided repair of any LP) and a domain-specific validation phase (five rationality checks grounded in inventory theory). We test 22 API models from 7 families on 976 multi-echelon supply chain problems and train two 8B-parameter models using self-taught reasoning with solver-verified rewards. The trained models reach 81.7% Rational Recovery Rate (RRR) -- the fraction of problems resolved to both feasibility and operational rationality -- versus 42.2% for the best API model and 21.3% on average. The gap concentrates in Phase 1 repair: API models average 27.6% recovery rate versus 97.2% for trained models. Managerial Implications. Two gaps separate current AI from reliable model repair: solver interaction (API models restore only 27.6% of infeasible formulations) and operational rationale (roughly one in four feasible repairs violate supply chain theory). Each requires a different intervention: solver interaction responds to targeted training; operational rationale requires explicit specification as solver-verifiable checks. For organizations adopting AI in operational planning, formalizing what "rational" means in their context is the higher-return investment.
Executive Summary
This study introduces OptiRepair, a closed-loop diagnosis and repair system for supply chain optimization models using Large Language Model (LLM) agents. The system consists of a feasibility phase and a validation phase, achieving an 81.7% Rational Recovery Rate (RRR) in resolving infeasible formulations. The study highlights the gap between current AI capabilities and reliable model repair, particularly in solver interaction and operational rationale. The findings suggest that targeted training and explicit specification of operational rationale are necessary to improve AI performance in model repair. The study's results have significant implications for organizations adopting AI in operational planning, emphasizing the importance of formalizing the concept of 'rational' in their context.
Key Points
- ▸ OptiRepair is a closed-loop diagnosis and repair system for supply chain optimization models using LLM agents.
- ▸ The system achieves an 81.7% RRR in resolving infeasible formulations.
- ▸ The study highlights the gap between current AI capabilities and reliable model repair, particularly in solver interaction and operational rationale.
Merits
Improved AI Performance in Model Repair
OptiRepair demonstrates significant improvement in AI performance in resolving infeasible formulations, highlighting the potential of LLM agents in supply chain optimization.
Enhanced Understanding of AI Limitations
The study sheds light on the limitations of current AI capabilities in model repair, particularly in solver interaction and operational rationale, providing valuable insights for future research.
Demerits
Limited Generalizability
The study's results may not be generalizable to other domains or problem types, limiting the applicability of the findings.
Dependence on Solver Interaction
The performance of OptiRepair relies heavily on the interaction with solvers, which may not be feasible in all scenarios, highlighting the need for further research in this area.
Expert Commentary
The study's findings are significant, as they demonstrate the potential of LLM agents in supply chain optimization. However, the limitations of current AI capabilities in model repair, particularly in solver interaction and operational rationale, highlight the need for further research in this area. The study's results have significant implications for organizations adopting AI in operational planning, emphasizing the importance of formalizing the concept of 'rational' in their context. Moreover, the study's findings contribute to the ongoing discussion on model repair in supply chain optimization, providing new insights and approaches for resolving infeasible formulations.
Recommendations
- ✓ Future research should focus on developing more sophisticated AI models that can address the limitations of current AI capabilities in model repair, particularly in solver interaction and operational rationale.
- ✓ Organizations adopting AI in operational planning should prioritize formalizing the concept of 'rational' in their context to ensure reliable model repair, and policymakers should consider the implications of AI in operations research and develop strategies to address the limitations of current AI capabilities.