Heuristic — local search
Stochastic Hill Climbing
| Alias | stochastic_hill |
| Type | Heuristic — local search |
| Auto-seeds from | shuffle (random tour) |
Description
Iterative improvement with random restarts. Each step applies a random perturbation to the current tour; the new tour is accepted if it is better. When no improvement is found for platoo_epochs consecutive steps, the solver restarts from a fresh random tour. The best tour seen across all restarts is returned.
Auto-expands to pipeline(shuffle, stochastic_hill).
Options
| Flag | Description | Default |
|---|---|---|
--epochs |
Maximum iterations total (0 = unlimited) | 0 |
--platoo_epochs |
Steps without improvement before restart | — |
Usage
teeline solve stochastic_hill -i ./data/tsplib/berlin52.tsp
teeline solve stochastic_hill -i ./data/tsplib/berlin52.tsp --epochs=1000
teeline solve stochastic_hill -i ./data/tsplib/berlin52.tsp --platoo_epochs=50
References
- AIMA, Chapter 4.1 — Local Search and Optimization Problems
- Hill climbing (Wikipedia)