When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming
Abstract
Mixed-integer programming (MIP) solvers are widely used, yet their performance can vary substantially even under seemingly neutral perturbations. Such variability complicates benchmarking, and the interpretation of computational studies. We introduce the Weighted Variability Score (WVS), a robust multi-component metric that aggregates dispersion in normalized runtime, final optimality gap, LP iterations, and branch-and-bound nodes across repeated runs. On a large subset of MIPLIB 2017 instances, each solved in 100 randomized runs under a four-hour limit, WVS reveals a right-skewed distribution with a substantial upper tail: many instances are essentially stable, but some exhibit large variability that is not explained by basic properties alone. Notably, this experiment proved optimality for the open instance 'neos-5045105-creuse' in two SCIP runs. Using static descriptors and root-level features, a tree-based predictor with a gated tail uplift achieves a mean absolute error of 0.058 and Spearman rank correlation of 0.65 on a sealed test split. Feature-importance analysis highlights root degeneracy, matrix structure, coefficient ranges, and presolve reductions as the main drivers of high WVS. Because variability is systematic and partially predictable from inexpensive early-run features, our findings offer a practical tool for preemptive tuning and lay a foundation for fairer evaluation of new algorithms.
Related
- Mitigating Performance Variability in Mixed-Integer Programming Solvers via LP State Steering
- Log-Driven Semantic Tuning for Mixed Integer Programming Solvers
- Comparative Performance Analysis of Optimization Solvers for Linear Support Vector Machines: Impact of Preconditioning and Dimensionality An Empirical Study Using the Benchopt Framework
- Quantifying and Predicting Pitch Sequence Effectiveness Using Visual Deception Metrics
- Comparative Analysis of Large-Scale SVM Solvers: Convergence, Resource Consumption, and Energy Footprint
BibTeX
@misc{adegbindin2026when,
title = {{When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming}},
author = {Mourchid Adegbindin and Brenda Dietrich and Jamol Pender and Andrea Lodi},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.12876},
eprint = {2610.12876},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.12876},
}
RIS
TY - UNPB
TI - When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming
AU - Mourchid Adegbindin
AU - Brenda Dietrich
AU - Jamol Pender
AU - Andrea Lodi
PY - 2026
DA - 2026/10/09
PB - arXiv
DO - 10.48550/arxiv.2610.12876
UR - https://arxiv.org/abs/2610.12876
AB - Mixed-integer programming (MIP) solvers are widely used, yet their performance can vary substantially even under seemingly neutral perturbations. Such variability complicates benchmarking, and the interpretation of computational studies. We introduce the Weighted Variability Score (WVS), a robust multi-component metric that aggregates dispersion in normalized runtime, final optimality gap, LP iterations, and branch-and-bound nodes across repeated runs. On a large subset of MIPLIB 2017 instances, each solved in 100 randomized runs under a four-hour limit, WVS reveals a right-skewed distribution with a substantial upper tail: many instances are essentially stable, but some exhibit large variability that is not explained by basic properties alone. Notably, this experiment proved optimality for the open instance 'neos-5045105-creuse' in two SCIP runs. Using static descriptors and root-level features, a tree-based predictor with a gated tail uplift achieves a mean absolute error of 0.058 and Spearman rank correlation of 0.65 on a sealed test split. Feature-importance analysis highlights root degeneracy, matrix structure, coefficient ranges, and presolve reductions as the main drivers of high WVS. Because variability is systematic and partially predictable from inexpensive early-run features, our findings offer a practical tool for preemptive tuning and lay a foundation for fairer evaluation of new algorithms.
ER -