---

name: genetic-algorithms description: "Use scikit-opt GA-family APIs for Gray-coded, elitist, and real-coded optimization." disable-model-invocation: true metadata: disco-role: operating license: MIT

genetic-algorithms

Use this sub-skill for bounded optimization with GA, EGA, and RCGA, including integer or mixed-precision search, continuation with repeated run(max_iter) calls, history extraction, and safe register()-based operator swaps.

Read these bundled files

• references/api-reference.md — exact constructor signatures, parameter meanings, result/history fields, and the GA_TSP handoff note. Read this when you need precise argument names or return values.
• references/workflows.md — which GA-family class to choose, how to continue a run, how to set integer or mixed precision, and when to route problem workflows elsewhere. Read this before planning a solution.
• references/custom-operators.md — SkoBase.register, operator signatures, built-in ranking/selection/crossover/mutation helpers, and a safe custom selection pattern. Read this when replacing defaults.
• references/troubleshooting.md — fixes for even-population assertions, objective shape issues, bound/precision mismatches, penalty surprises, custom operator errors, and stochastic convergence. Read this when a run fails or behaves oddly.
• scripts/smoke_genetic_algorithms.py — tiny deterministic import-and-run smoke for GA, EGA, RCGA, and register() behavior. Run this after changes or before handoff.

Route elsewhere when needed

• Route DE, PSO, SA, and AFSA continuous optimizers to continuous-optimizers.
• Route full TSP or routing workflows to routing-and-combinatorial; treat GA_TSP only as a handoff marker here.
• Route run-mode, vectorization, cache, multiprocessing, and GA.to(device) details to objective-functions-and-speedups.

This sub-skill stays small on purpose: the bundled references carry the exact API and workflow detail, while the runtime tree remains self-contained.