01In progress
Searching a mapPathfinding
A*, Dijkstra and friends on a grid: five ways to pick the next cell, small enough to watch.
Open the project→Games and search
How does a program choose a move?
Each game is its own project, with its own rules. They share one question: given a position, which move is best, and how far ahead must the search look? More games will be added.
The series
02In progress
Searching a gameTic-tac-toe
Minimax with alpha-beta pruning: the same idea of a search, but against an opponent who answers back.
Open the project→03Shipped
A connection gameHEX
A shortest-path evaluation feeds an alpha-beta search. Raise the depth and watch the positions examined.
Open the project→04Shipped
Flipping a boardOthello
Corners, edges and mobility decide the game, so the evaluation matters as much as the search. Pick the algorithm that plays you.
Open the project→05Shipped
Five opponents, one boardConnect Four
Few moves per turn, so the searches can go deep. Compare minimax, alpha-beta and Monte Carlo on the same position.
Open the project→06In progress
A game with phasesMills
Placing, moving and flying change the rules mid-game. Finished in C++; the browser version is next.
Open the project→Why they belong together
Each game needs the same three things: a way to represent a position, a way to score it, and a way to skip what is not worth searching.