Brief

Researchers have developed MIPLIB-NL, a new benchmark designed to evaluate how well large language models can translate natural language into optimization formulations and executable code. This benchmark is derived from real-world mixed-integer linear programs from MIPLIB 2017, addressing the limitations of existing toy-sized or synthetic datasets. Experiments indicate that current LLMs perform significantly worse on MIPLIB-NL compared to existing benchmarks, revealing challenges with industrial-scale problems that were previously masked. AI

Researchers have developed a novel solver-free framework for tackling Integer Linear Programming (ILP) problems, which are common in combinatorial optimization. This new method directly explores feasible regions without relying on traditional solvers or machine learning training. It utilizes a Locally-Balanced Proposal for its transition kernel and incorporates Parallel Tempering, including a new penalty tempering technique that adjusts constraint barriers. The framework demonstrates superior performance compared to established solvers like SCIP and Gurobi on several benchmarks, showing greater robustness to distribution shifts than learning-based approaches. AI