Post-Solve Robustness framework audits MILP trustworthiness under perturbations

Yi-Xiang Hu's position paper tackles a critical gap in optimization pipelines: the assumption that a solved MILP plan remains valid under real-world perturbations. The proposed 'post-solve robustness layer' audits a solved incumbent and returns certified evidence about its trustworthiness. Two central concepts are formalized: an ε-near-optimal feasible neighborhood in parameter space (capturing perturbations where the solution stays feasible and near-optimal) and solution smoothness in decision space (measuring whether nearby alternatives with small combinatorial edits remain competitive).

The paper synthesizes tools from sensitivity analysis, robust optimization, neighborhood search, adversarial testing, and learning-based methods to build a unified framework. It calls for certified inner approximations around incumbents, probabilistic robustness estimation with calibrated uncertainty, adversarial robustness margins, and learning-based prediction aligned with solver-backed verification. A compact reporting template and evaluation protocol are proposed to make robustness a first-class output of decision engines, directly benefiting high-stakes applications like supply chain, energy, and logistics where small shifts can invalidate optimal plans.