search-graph

Data structure for optimization on a discrete space

Go to file

Stu Black d61b946d35 Documentation, add missing methods, and merge EdgeAdder with MutChildList.		2016-01-24 18:04:29 -05:00
src	Documentation, add missing methods, and merge EdgeAdder with MutChildList.	2016-01-24 18:04:29 -05:00
.gitignore	Factored out into its own thing, with code more neatly organized.	2016-01-22 20:56:07 -05:00
Cargo.toml	Factored out into its own thing, with code more neatly organized.	2016-01-22 20:56:07 -05:00
LICENSE	Factored out into its own thing, with code more neatly organized.	2016-01-22 20:56:07 -05:00
README.org	Factored out into its own thing, with code more neatly organized.	2016-01-22 20:56:07 -05:00

Overview

This package provide a rollout-based, DAG-like data structure. Its intended use is building game state graphs for general game-playing or similar AI tasks.

Vertices correspond to game states. They are addressed by unique game states (which may be de-duplicated as if by transposition table by implementing std::hash::Hash and std::cmp::Eq appropriately for your game state type).

Edges correspond to game-modifying moves. They are bidirectional, so paths may be traced up and down the graph.

The structure is rollout-based in that edges are expanded lazily. This should make it easy to implement planning algorithms like A* and Monte Carlo tree search on top of the DAG structure.

It is DAG-like in that it nominally maintains a cycle-free topology by checking for cycles at edge expansion time and noting which edges would create cycles when they are expanded. Such edges are added to the graph as half-edges, which are only present in one direction. Unfortunately, checking for cycles leads to O(|V|) running time when adding vertices.

At present, the graph topology may be grown but not pruned. Vertex and edge deletion may be supported in a future version.

Copyright

Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.