CHOMP

CHOMP
Title	CHOMP
Developer	Unknown
Publisher	Unknown
Platforms	Multiplatform
Released	Unknown
Genre	Abstract strategy
Modes	Single-player, multiplayer

CHOMP CHOMP is an abstract two-player strategy game characterized by a diminishing grid, asymmetric move options, and a last-move-loses condition common to impartial combinatorial play. Players alternately remove contiguous portions of a rectangular array, creating progressively smaller configurations; play ends when a designated poisoned cell is taken. The game has influenced researchers and hobbyists across combinatorial game theory, recreational mathematics, and artificial intelligence.

Definition and overview

CHOMP is defined on a finite rectangular grid of cookies or squares with a designated poisoned cell, typically the upper-left corner. On each turn a player chooses one cookie and removes that cookie and all cookies to its right and below, producing a new skyline shape; the player forced to take the poisoned cookie loses. The game is impartial in the combinatorial sense but asymmetric in board geometry, making it a standard illustrative example alongside Nim, Kayles (game), Green Hackenbush, Treblecross, and Wythoff's game in texts by authors such as Elwyn Berlekamp, John Conway, and Richard Guy. CHOMP boards are commonly named by dimensions (m × n) as in tournaments involving Mathematical Association of America, American Mathematical Monthly problems and International Mathematical Olympiad-style recreational problems.

History and development

CHOMP emerged in the late 20th century within the community of recreational mathematics and theoretical computer science. Early informal descriptions circulated in puzzle columns and seminars led by figures associated with Bell Labs, MIT, and Princeton University; formal analyses appear in papers inspired by the work of Conway and the development of combinatorial game theory at institutions like University of Cambridge and University of California, Berkeley. Interest expanded through conferences such as the Symposium on Theory of Computing and workshops organized by SIAM and IEEE on algorithmic game theory. Influential expositions appeared in collections edited by Martin Gardner and referenced by authors at Harvard University and Stanford University when teaching algorithmic strategy topics.

Gameplay and mechanics

A standard play session begins on an m × n grid with the top-left cookie designated as poisoned; typical family and research variants use sizes such as 3×3, 4×7, or 8×8. A legal move selects a cell (i,j) and removes every cell (x,y) satisfying x ≥ i and y ≥ j, analogous in effect to choosing a suffix of rows and columns. The combinatorial structure yields P-positions and N-positions analyzed via the Sprague–Grundy theorem and mex functions, similar to computations used in analyses of Nimble and Turning Toads and Frogs. Optimal play is known for some small rectangles and special classes (e.g., 1×n and 2×n) through pairing strategies and symmetry arguments invoked in lectures by scholars at University of Oxford and Caltech. The strategy-stealing argument, attributed to techniques in papers by Stewart Coffin and others, proves that the first player has a non-losing strategy on rectangular boards but does not construct it explicitly, paralleling existence proofs in combinatorial problems discussed in Annals of Mathematics Studies.

Variants and adaptations

Many variants adapt removal rules, poisoned-cell positions, or scoring mechanisms. One variant, "Misère CHOMP," inverts the end condition akin to misère play in Nim, altering strategy classification studied by researchers at University of Waterloo and University of Toronto. Higher-dimensional CHOMP generalizations replace the grid with 3D arrays and hyperrectangles, drawing parallels with problems treated at DIMACS workshops. Stochastic variants introduce chance elements reminiscent of work on Markov decision processes at Columbia University, while partisan adaptations assign different move sets to players analogous to constructs in Partizan Dawson's Kayles. Implementations appear in puzzle columns of Scientific American and at programming competitions hosted by ACM and Google Code Jam.

Cultural impact and reception

CHOMP enjoys a steady presence in mathematical outreach, recreational literature, and classroom demonstrations. It features in expository articles in Mathematical Gazette and problems posed in Putnam Competition-style training. Educators at University of Michigan and Yale University use CHOMP to introduce concepts from combinatorial game theory and algorithmic complexity, and popularizers like Ian Stewart and Persi Diaconis have referenced its pedagogical utility. Open-source communities on platforms such as GitHub host implementations and analyses, while blogs maintained by contributors associated with Wolfram Research and Ars Technica discuss its surprising combinatorial richness. CHOMP-inspired art and puzzles have appeared at maker fairs and exhibitions organized by Museum of Mathematics and Science Museum outreach teams.

Technical implementations and algorithms

Algorithmic treatments implement recursive search, memoization, and symmetry pruning to analyze positions; state encoding commonly uses bitboards similar to techniques applied in Chess programming and Go (game) engines. Exact solving for modest boards employs dynamic programming and canonicalization via row-wise skyline compression; more advanced attempts apply alpha-beta search, transposition tables, and heuristic evaluation functions akin to those in Monte Carlo tree search research at University College London and University of Alberta. Computational complexity results connect CHOMP decision problems to PSPACE-hardness analogues explored in papers from Cornell University and Carnegie Mellon University, while experimental machine-learning efforts use reinforcement learning frameworks popularized by DeepMind to approximate strategies for larger boards.

Category:Abstract strategy games