Interactive simulations of classic games with perfect-play algorithms. Watch unbeatable AI in action, explore game-theoretic solutions, and analyze win conditions in real time.
A fully solved game. Minimax with alpha-beta pruning plays perfectly — the best you can do is draw. Includes complete decision tree analytics.
Strongly solved in 1988 — first player wins with perfect play. AI uses Minimax + Alpha-Beta with window scoring and center-column preference.
Minimax + Alpha-Beta with positional weights, mobility scoring, and corner-capture bonuses. The AI prioritizes corners and minimizes opponent mobility.
Alpha-Beta Minimax with mandatory capture enforcement, multi-jump chains, king promotion, and back-row/mobility heuristics on a full 8×8 canvas board.
Knuth's 5-guess algorithm guarantees a solution in at most 5 guesses for any 4-peg 6-color code. Minimax over remaining possibilities each turn.
Gaussian elimination over GF(2) finds the minimum button presses to solve any board configuration. Selectable 3×3 through 6×6 grid sizes.
Constraint propagation + subset analysis solver. Deduces safe cells and mine positions logically, with probabilistic fallback when forced to guess.
Mathematically perfect play via the Sprague-Grundy theorem — the AI computes the Nim-sum (XOR) of all heap sizes and always leaves a losing position.
Expectimax AI with a Snake heuristic weight matrix. Scores board by tile values, smoothness, monotonicity, and empty cells. Reliably reaches 2048+.
Tile-rummy simulation with an optimal greedy set-cover solver using dynamic programming to maximize tile placement. Supports 1–4 players.
4-player territory control with all 21 pieces and all 8 transforms (rotations + flips). AI uses greedy corner-expansion heuristics with lookahead and blocking.