Skip to content

v0.1.0 — Initial release

Choose a tag to compare

View all tags
@apiad apiad released this 02 Jun 19:50
v0.1.0
685bbb2

First public release of PATHOS — a problem-centric library of classical AI search algorithms for Python.

You declare your problem's structure (via decorator hooks on a Space); the auto-solver picks the most powerful compatible algorithm. No machine learning — pure search.

Highlights

Anytime delivery by default. space.solver().solve() runs a cascade [GreedyBestFirst → WeightedAStar(5,3,2,1.5) → AStar], keeping the best incumbent across phases. Optimal if the budget allows; best-effort otherwise — never not_found when a feasible path exists. Default budget is 1 hour; set solver(timeout=…) to bound it tighter.

Cooperative cancellation across every algorithm. A lightweight CancelToken is checked inside the main loop of every metaheuristic and path-search algorithm. Set a timeout on a GeneticAlgorithm and you get the best individual seen so far when time runs out — not not_found.

Three execution modes:

  • mode="auto" (default) — anytime cascade.
  • mode="exact" — single-shot admissible algorithm (pre-cascade behaviour).
  • mode="approximate" — single-shot bounded-suboptimal A* variant.

Quality bound exposed. SearchResult.epsilon distinguishes proven-optimal (1.0) from ε-bounded (>1.0) from unbounded (inf). result.optimal is the derived convenience boolean.

Coverage

  • Uninformed: BFS, DFS, IDDFS, UCS
  • Informed: A*, IDA*, WeightedA*, BidirectionalA*, GreedyBestFirst
  • Meta: AnytimeAStar (the cascade above, registered as a normal Algorithm)
  • Local search: HillClimbing, TabuSearch, LocalBeamSearch
  • Evolutionary / Metaheuristic: SimulatedAnnealing, GeneticAlgorithm, DifferentialEvolution, ParticleSwarm
  • Adversarial: Minimax, AlphaBeta, Negamax, MCTS
  • CSP: Backtracking, ForwardChecking, MinConflicts

Specialized spaces: GraphSpace, CSPSpace, TourSpace, GameSpace.

Installation

pip install pathos-ai

Python 3.11+. MIT license.

Documentation

Assets 2
Loading