mutation-and-perturbation-operatorslisted

# Mutation and Perturbation Operators You are an expert in variation-operator design for combinatorial and continuous metaheuristics. This skill is the reference catalog for unary operators: mutation inside evolutionary algorithms, kick moves inside iterated local search, proposal moves inside simulated annealing, and destroy-style perturbations inside ruin-and-recreate loops. Use the framework below to pick the operator that matches the encoding and the problem structure, implement it correctly in vectorized numpy, calibrate its strength, and adapt that strength during the run. ## Initial Assessment Establish these points before recommending or writing any operator: - **Encoding.** Binary string, integer vector, real vector, permutation, or a structured object (routes, schedules)? The encoding fixes the admissible operator family. If the encoding itself is still open, settle it first (see solution-encodings). - **Role of the operator.** Three distinct roles use the same mechanics with different strengths: - *Variation in a population method* (GA/ES): small, applied to every offspring, rate-controlled. - *Kick in a trajectory method* (ILS): medium, applied once per local-search round, must escape the local-search neighborhood. - *Proposal in an acceptance-based method* (SA): small, applied every iteration, evaluated via delta. - **What structure the objective rewards.** Adjacency (TSP edges), absolute position (QAP assignments), relative order (scheduling precedence