Reproduce “When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mourchid Adegbindin, Brenda Dietrich, Jamol Pender, Andrea Lodi
The authors measure run-to-run variability in mixed-integer solver performance and predict which instances may be more variable. Their Weighted Variability Score combines spread in runtime, final optimality gap, LP iterations, and branch-and-bound nodes across repeated runs. Among 1,009 SCIP instances, median WVS was 0.085; a predictor achieved MAE 0.058 and rank correlation 0.65 on filtered consensus labels, not solver-independent prediction.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.