TSP — aprender-tsp Local Optimization
Recipes for aprender-tsp v0.31.2 — Traveling-Salesman-Problem solver that uses personalized .apr models to bias edge selection from user history (delivery routes, daily commutes, etc.). All-local, no network.
Closes the ≥3 recipes per sister crate requirement from expand-cookbooks/subcrate-coverage.md.
Recipes
| # | Recipe | What |
|---|---|---|
| TSP.1 | tsp_solve_with_tabu | 10-city Euclidean TSP solved by TabuSolver with seed-deterministic output |
| TSP.2 | tsp_distance_matrix_explicit | 5-city symmetric distance matrix (non-Euclidean) via TspInstance::from_matrix + jagged/empty rejection |
| TSP.3 | tsp_compare_tabu_vs_genetic | Same instance solved by TabuSolver vs GaSolver with matched iteration budget |
API surface exercised
aprender_tsp::instance::TspInstance::{from_coords, from_matrix}aprender_tsp::solver::{TabuSolver, GaSolver, TspSolver, Budget}TabuSolver::with_seed(u64)— deterministic output for IIUR
Provenance
Added during PMAT-080 (expand-cookbooks initiative, v6.1.0).