![]() StartingPoints Paths Referenced by
|
tree-representation
Path: CGTPath · Prev: GameTheoryInterface · Next: SurrealNumbers Other than showing a board, pieces and so on, there are several commonly-used ways of describing games. The tree-representation has plenty going for it, especially in computing terms. In fact it corresponds to writing games as conventional data structures [1]. Unfortunately it is the worst way to display data using a keyboard. The normal convention uses the slash | as notation for the basic tree-forming operation, and {(*) and (*)} for the left- and right-hand separators. Inside them (*) stands for any unordered list of games. Therefore {G,...|H,...} is to be read as a crunched-onto-one-line notation for a tree with a root, left-hand 'branch' a list G,... of games and right-hand 'branch' another list H,... of games. Since each of G, H and so on may in turn be represented as a tree, with enough paper one can draw the entire tree. Books on CGT follow the universal computer science convention that trees have a root at the top and leaves below. In this notation, we read G and so on as Left's options ('moves') and H,... as Right's options. Left isn't allowed to play (choose) in the right-hand side, though there is no reason at all that some of the games on the right of the slash can't co-incide with some on the left. In fact in games like Nim which are impartial that's exactly what happens: the lists on the left and on the right are always the same. People use || and ||| and so on to save on some { and }: the more ||| the closer the slash is to the root. A bit deplorable, that, in my view. Readers of Winning Ways will know that an alternative approach is to invent an algebra of games, with operations and notations for common games. This seems much less useful for the purposes of go than for the general theory (which is indeed extremely general, and aims at unifying phenomena over a broad sweep of ludic territory). [1] This is natural if one assumes a certain attitude to combinatorics, a flourishing if junior branch of mathematics. Combinatorics can be considered to deal with the representative behaviour of typical data structures of some type. This would be my definition of the subject matter: combinatorialists tend to argue in terms of proofs and techniques for those. The reduction to canonical form is a typical (successful) combinatorial procedure, combing out 'data' not needed in a game for the usual purposes. Data structures without explicit constraints on their structure tend to be 'ragged', and the number of possibilities grows fast in typical cases (the 'combinatorial explosion'). What this might have to do with go is that the explosion of possibilities is familiar to players; but strong players in general feel that go is 'tidy' and 'equitable' - you might say this corresponds to ideas of shape and exchange which are well-researched working concepts. Path: CGTPath · Prev: GameTheoryInterface · Next: SurrealNumbers This is a copy of the living page "tree-representation" at Sensei's Library. ![]() |