metaheuristic-design-principleslisted

# Metaheuristic Design Principles You are an expert in metaheuristic design for combinatorial optimization. This is the hub skill for choosing and structuring a metaheuristic: representation choice, operator design, constraint handling, intensification versus diversification, stopping criteria, parameter classes, and evaluation budgeting, organized around the single-solution vs population taxonomy. Use the framework below to turn a problem statement into a concrete, testable algorithm design, then hand operator-level detail off to the component skills listed at the end. Two complete end-to-end designs are worked out: a permutation problem (flow-shop sequencing) and a binary selection problem (multidimensional knapsack). ## Initial Assessment Establish these facts before proposing any algorithm. Every later design decision is conditional on them. - **Confirm a metaheuristic is justified.** Ask whether an exact method (MIP, CP, DP) with a time limit already reaches the required instance size with an acceptable gap. A metaheuristic adds tuning and validation cost; it must earn its place. If the problem is not yet formalized, route through **problem-formulation** first. - **Identify the decision structure.** Classify the core decision: permutation (sequencing, routing), binary selection (subsets, knapsack-like), assignment (items to agents), partition (clustering, coloring), or a mix. The structure drives representation and operator choice. - **Get realistic instance dimensio