ChompPlayer-ImplementationDescription.docVIP

CS Workshop Chomp Player - Project Design May 29th 2003 Daniel Glasner 034495036 Omer Kulka 033370669 Alon Benari 033315615 Chomp Player - Implementation Description General: The game graph: Let us look at a game starting with an M row by N column board, (w.l.o.g MN). Consider all the possible intermediate moves, the game graph G = (V,E) is a directed graph whose vertices are all possible game positions. An edge (u,v) exists iff there is a legal move from position u to position v, thus the graph is acyclic. We can also compute the size of V to be: and if then . We now define the groups(Previous player wins or Next player wins). . From here on, we will be concerned with an algorithm that determines if a given position v is in P or in N. G can be viewed as a layer graph consisting of M*N layers, in which, the classification is such that every layer i, for, contains all the positionssuch that S(v) = i , where S(v) is the area of v, or the number of squares in the position. Thus, the first and MNth layers contain one vertex each - the beginning and ending positions respectively. We denote that for any edge (u,v) layer(u)layer(v). Representing a position: We can think of chomp positions as integer partitions. A partition of n is a finite, non-decreasing sequence of positive integers such that: . By limiting rM andN we achieve the desired correspondence. It is now possible to represent a position v using two integers (i,j), where, i indicates a layer in the above mentioned classification of the graph and j is the index of the partition of I, which corresponds to v (we use an indexing scheme to number the distinct partitions of i). The transposition of a position is a reflection of that position along the primary diagonal. Obviously lies in P iff v does, as a result we need only deal with half the positions. Algorithm: Our chomp player relies on a data base of P positions which is generated using a dynamic