Combinatorial Game Theory
who wins, and by how much
Concept

Exact evaluation — where it appears

Computing a position's value from the rules rather than estimating it, which is what every number on this site is. It is what makes a claim here refutable: a reader who beats one of these figures has found an error in the recursion.

Named by 25 essays across 5 fields — each of them below, with the objects they name alongside it.

Which questions are answerable. The theory is exact and much of it is expensive. Values are computable by definition; computing one for a position of any size is a different matter, and deciding the winner of a generalised board game is complete for PSPACE — as hard as anything solvable in polynomial space.

How hard is it

Every theorem on this site stays true at any size. The answers stop being reachable long before the games get interesting — deciding the winner of a generalised board game is PSPACE-complete, and an exact evaluator gives out after a few dozen moves.

complexity · Complexity A formula, drawn as a game. A token on a directed graph. A move slides it along an edge to a vertex not yet visited, and a player who cannot move loses. That is the whole game, and deciding who wins it is as hard as anything decidable in polynomial space — which is the strongest hardness claim anybody makes about a combinatorial game.

Hard, proved

A game is as hard as a logical formula when the formula can be drawn as the game. Here is the drawing — a quantified formula turned into a graph with a token on it — with every formula over three variables played both ways and required to agree.

complexity · Complexity What it costs to decide the winner. Three families of game, each at several sizes, with the number of distinct positions an exhaustive solver must evaluate beside the work its closed form does. The bars are logarithmic. Hardness is not about the size of the board or the length of the rules — Domineering has the shortest rule here — it is about whether anybody has found the shortcut.

The game with the shortest rule is the hard one

Deciding a generalised board game is PSPACE-complete, which is a statement about families and encodings rather than about size. Nim in the same subject is settled by one pass over the input at any size, and green Hackenbush by one pass over the edges — while Domineering, whose rules take a single line, has no shortcut anybody has found.

complexity · Complexity a loop with a way out: what the backward analysis settles. A position graph in which the moves can lead back to where they started. The labels are the order in which a backward analysis settles each position, starting from the ones where a player has already run out of moves. Positions the analysis never reaches are drawn — and there is no test for that; being unreachable is what a draw is.

Start at the end and work backwards

When play can return to where it started there is no bottom for the recursion to stand on. What replaces it begins at the positions where somebody has already lost and propagates outwards — and the positions it never reaches are exactly the draws. There is no test for a draw, and there does not need to be.

limits · Loopy One node per route, one node per position. For each board, the number of nodes in the recursion tree a solver with no memo table would walk, beside the number of distinct positions that tree contains, beside the longest run of moves in it. The first number is the cost of forgetting; the second is the size of the table that avoids it; the third is the stack, and it stays small however the other two grow.

A position reached eleven ways is one position

A 4×4 Domineering board has 5,700 positions in it and 6,257,129 routes through them. Three heaps of 7, 11 and 13 have 480 positions and 7.6 × 10¹⁶ routes. The gap between those two numbers is not an optimisation — it is the difference between a search that finishes and one that does not.

complexity · Search a cycle of three: what the backward analysis settles. A position graph in which the moves can lead back to where they started. The labels are the order in which a backward analysis settles each position, starting from the ones where a player has already run out of moves. Positions the analysis never reaches are drawn — and there is no test for that; being unreachable is what a draw is.

An outcome with no value behind it

Retrograde analysis labels positions in rounds, outward from the ones already lost. Whatever is still blank when nothing more can be deduced is a draw — and there is no separate test for a draw, because a draw is exactly the residue the method never reaches.

limits · Loopy A board in pieces costs the sum, not the product. A Domineering board with squares blocked out, so that it falls into regions no domino can span. The number of positions in the whole board is exactly the product of the numbers in its regions — which is why evaluating the regions separately, and adding the values, is an exponential saving rather than a tidier way of writing the same search.

The board falls apart, and the arithmetic changes

A 4×5 Domineering board with a wall down the middle has 2,916 positions in it, and that number is exactly 54 × 54 — the product of its two halves. Solving the halves separately costs 108. Decomposition is the one saving in this subject that turns a product into a sum.

complexity · Decomposition Amazons, after the arrows have cut the board in 2. An amazon moves like a queen and then shoots an arrow, also like a queen, which burns the square it lands on. Late in a game the burnt squares cut the board into regions no amazon can cross — and from that moment the position is a sum of independent games, which is the shape the whole theory was built for, arrived at by the play rather than assumed.

Amazons, and when a position becomes a sum

Every technique on this site starts from a position already broken into independent parts. Amazons does not begin that way — the board is one fight until the arrows cut it, and the moment of cutting is something the play produces rather than the analyst assumes.

positions · Amazons The same position, and two rulers to measure it with. Nim positions with the length of their input under two encodings — the heap sizes in binary, and the counters themselves — beside the work the nim-sum does. The work never changes. Which of the two lengths it is compared against decides whether the same algorithm reads as linear or as exponentially fast, and hardness claims are always made against one particular encoding.

Nim is easy, in binary

Three heaps of a thousand counters take thirty bits to write down and three thousand counters to lay out. The nim-sum does three exclusive-ors either way. Whether that counts as fast depends entirely on which of the two numbers the work is compared against.

complexity · Complexity The thermograph of {6 | {5 | {4 | 0}}}. Temperature runs up the page and value across it. Each wall is where a player is willing to move once a tax of that much is charged per move; above the temperature at which they meet, neither wants to move and the position is worth its mean value. The height of the meeting point is what is at stake.

A thermograph is built from its options'

The diagram is not measured, it is computed — each wall is an extremum over the options' opposite walls, shifted by the tax. That construction is why a hot follow-up lowers the temperature instead of raising it, and why two fights with the same swing can differ by a factor of two in what is at stake.

temperature · Thermograph Toads and frogs. Toads move right and frogs move left, one square into a gap or hopping over exactly one opponent. A player unable to move loses. It can be played on squared paper by anybody, and its values are immediately stranger than the game looks.

The strip nobody has a formula for

Some toads, a gap, some frogs. Two counts and a spacing is the whole description, and the values that come out of it are integers, stars, switches with eighth-point options and a down — four classes inside one two-parameter family, which is why nobody has written the formula.

positions · Toads and Frogs the temperature that bounds the loss. Temperature runs up the page and value across it. Each wall is where a player is willing to move once a tax of that much is charged per move; above the temperature at which they meet, neither wants to move and the position is worth its mean value. The height of the meeting point is what is at stake.

A rule that is never right and cannot be far wrong

Playing the hottest component is not optimal — over 440 measured lines it costs something on 17 of them. What makes it worth having is that the cost is bounded by the temperature, provably, and that the same test run with the ordering reversed breaks the bound on 54.

complexity · Approximation The thermograph of {{5 | {3 | 1}} | 0}. Temperature runs up the page and value across it. Each wall is where a player is willing to move once a tax of that much is charged per move; above the temperature at which they meet, neither wants to move and the position is worth its mean value. The height of the meeting point is what is at stake. The marks partway up the walls are the bends: the heights at which the option holding a wall up stops holding it up. Each stands at the temperature of a follow-up. Every one of them is below the temperature of the position itself, which is where the two walls meet.

A thermograph with two bends

A wall bends where the option holding it up stops holding it up, and most drawn thermographs bend once. { {5 | {3 | 1}} | 0} bends twice, at 1 and 3/2, and neither height is its temperature of 7/4 — every bend lies strictly below where the walls meet. Over a census of 17,255 hot positions two levels deep, a second bend in one wall never happened once.

temperature · Thermograph The thermograph of {5 | 1}. Temperature runs up the page and value across it. Each wall is where a player is willing to move once a tax of that much is charged per move; above the temperature at which they meet, neither wants to move and the position is worth its mean value. The height of the meeting point is what is at stake. The two marks on the base line are the stops — what each player gets by moving first and playing the fight out with no tax charged at all.

Where the fight stops

Keep taking the biggest thing on offer until somebody is left facing a number, and the number reached is a stop. Every position has two of them, one for each player moving first, and between them they say how a fight ends — except for the part of it no number can reach.

values · Stops Small Domineering boards and what they are worth. Every value here was computed from the moves rather than looked up. Even on boards this small the values are switches and infinitesimals rather than numbers, which is the ordinary situation for a partizan game and the reason the theory needs more than arithmetic.

The values of every small board

Thirty Domineering rectangles, every value computed from the moves rather than looked up. The 1×n row obeys a formula and the 2×n row does not: its outcomes run L N N R three times over and then 2×13 comes out worth exactly 0, and its temperatures climb to 19/16 and fall back without settling.

positions · Domineering A hot position, split into its mean and what is left. Each row is a position cooled by exactly its own temperature — the tax at which it stops being worth moving in. The result is the mean value with something small still attached, and the last column is that something, obtained by subtracting the mean from the cooled game rather than by inspection.

A number and a fight

Charge a position exactly what it is worth fighting over and the fight disappears, leaving the mean value — with something still attached to it. Over all 1,122 hot values born by day three the residue is smaller than every positive number, it is a star in 942 of them, and it is never nothing. So a hot game is its mean plus a fight plus a remainder that no number reports, and the remainder is what decides close games.

temperature · Dissociation What a deeper position does to the shape. Thermographs side by side, two of them, with temperature running up each panel and value across it: {{6 | 2} | {1 | −3}}, with a bend where an option's own fight cools out; {4 | −1}, straight-walled. A wall that runs straight has nothing changing hands below the meeting point; a bend is an option's own fight cooling out at a lower temperature than this position's, and it is where a decision passes from one player to the other. The two marks on each base line are the stops — what each player gets by moving first with no tax charged.

The switch a player is imagining

Every account of a hot position ends up as "worth about m, and worth t to move in", which is the switch {m+t | m−t}. For a plain fight that summary is the position exactly. For a fight with anything behind it the leftover is not a rounding error — on one position here it is a whole second fight of temperature two.

values · Switches The same strips, with no jumping. Elephants and Rhinos: toads move right and frogs move left, one square into an empty one, and nothing may hop over anything. The pieces keep their order for ever, and the values are computed by the same recursion as the game with the jump in it.

The same strip without the jump

Delete one clause from Toads and Frogs — the hop over an opponent — and the game is Elephants and Rhinos. Over the same 3,279 strips the values do not become simpler in the way a reader would guess: every value that is a number becomes an integer, against 172 fractions with the jump in, and the count of positions worth fighting over nearly doubles. Removing a move made the game hotter.

positions · Toads and Frogs End-Nim: a player at each end of the row. Rows of heaps in which Left may take from the leftmost heap and Right from the rightmost. The value beside each row was computed by the game recursion and reduced to canonical form; the outcome beside it says who wins. A single heap is a Nim heap, because both players may take from it — and that is the last thing about this game that looks like Nim.

Taking from the ends

End-Nim is Nim's board with a player at each end, and it takes one sentence to state. Not one of its 5,460 small positions is worth a non-zero number — the game is all-small, so zero is the only number any of them can reach — and there are 2,693 distinct values between them. The outcome says a great deal more: 4,738 of those positions are won by the same player whoever moves, and on two heaps the rule is that the larger end wins.

positions · End-Nim Clobber: every value smaller than every number. Blue and red stones on a small board. A move takes one of your own stones onto an orthogonally adjacent enemy stone, which is removed. Because adjacency is symmetric, a player has a move exactly when the opponent does — so no position can ever be worth a whole move to anybody, and every value that comes out is an infinitesimal.

One row of Clobber

Every string of blue, red and empty squares up to eight long — 9,840 rows — carries one of only 111 values, and every one of them is infinitesimal. A third of the rows are worth exactly zero. Six alternating stones are worth zero and eight are worth a form that takes four lines to print, so the values do not simplify as the row grows: they explode, while the row stays trivial to describe.

positions · Clobber Shove strips, and what each is worth. A shelf of positions with the value the recursion returns beside each. Every one is a number: Shove has no hot positions at all, which is unusual for a partizan game and is the first of the essay's three claims.

Nothing worth fighting over

Shove is a strip of coins beside a cliff, and both players have completely different moves. Every one of its 728 positions is worth a number, so nobody ever wants to move; the winner is the owner of the coin furthest from the cliff, in all 728; and the number the board is worth is not the sum of its coins — that reading is exact on 126 strips and wrong on 588 of the other 602.

positions · Shove Three questions about the same board. For each sum of two positions: the cost of deciding who wins each part alone, of deciding who wins the whole sum by search, and of computing what each part is worth. The middle question is in the middle on seven of the eight, and the exception is the sum whose two parts are identical.

The question in the middle

Between knowing who wins each part and knowing what each part is worth sits the question a player actually has: who wins the board. Priced on sums of two it lands between the other two on seven of eight, cheaper than the values by up to eight times. On sums of three, with nothing repeated, it is dearer than the values on five of six — because a component multiplies a search and only adds to a value.

complexity · Value cost Right, wrong, and right again. A 4 × 5 Domineering position with Right to move, which Right loses, beside what a search cut at each depth from 0 to 9 says about it when it guesses that the player with more placements wins. The guess alone is right, a search one move deeper is wrong, and every deeper search is right.

A verdict that changes with the depth

A who-wins search of 4 × 5 Domineering cut at a fixed depth, guessing that the player with more placements wins where it stops, is right about 72.5 per cent of positions at depth 0 and about every one of them by depth 7. On the way, 4,697 positions are right at one depth and wrong at a deeper one. With a guess that knows nothing, going one move deeper makes the search worse — and its errors alternate in kind with the parity of the depth, so that half its verdicts are proofs.

complexity · Search Two depths that agree. A 4 × 5 Domineering position with Right to move, which Right wins, beside what a search to each depth from 0 to 6 says under the guess that any mover wins. The verdicts alternate until depths 2 and 3 agree, which certifies the answer 3 moves before the longest line.

Where a search may stop

A search deepened until two consecutive depths agree carries a proof of its verdict, and on 4 × 5 Domineering it stops before the longest line on 17,589 of 48,670 positions. It also costs three times what the search that simply finishes costs. The rule that pays is the other one. Search on wherever the two players' counts of placements are within one, and at depth 2 the wrong verdicts fall from 2,140 to 86 for about a quarter more work per search.

complexity · Search

The bound names the hottest part and the cost does not

Moving in the hottest component costs at most the largest temperature on the board, and that bound is attained: 100 lines of 4,240 pay exactly it. It is still the wrong quantity. Across four pools and boards of two, three and four parts the cost is nothing on 90.8% of lines and otherwise takes one of two values — half a point or one — on boards whose largest temperature runs to three, and it exceeds the coolest component on 13 lines and twice it on none.

complexity · Approximation

Named alongside it

The objects these essays reach for when they reach for this one.

Exhaustive searchTemperatureOutcome classComplexityMean valueSwitchThermographCanonical formDomineeringPartizanHot gameInfinitesimal

All concepts