From cube-lovers-errors@mc.lcs.mit.edu Thu Apr 8 13:14:00 1999 Return-Path: Received: from sun28.aic.nrl.navy.mil (sun28.aic.nrl.navy.mil [132.250.84.38]) by mc.lcs.mit.edu (8.9.1a/8.9.1-mod) with SMTP id NAA29818 for ; Thu, 8 Apr 1999 13:13:59 -0400 (EDT) Precedence: bulk Errors-To: cube-lovers-errors@mc.lcs.mit.edu From: Andrew John Walker Message-Id: <199904080113.LAA02581@wumpus.its.uow.edu.au> Subject: Solvers To: cube-lovers@ai.mit.edu Date: Thu, 8 Apr 1999 11:13:10 +1000 (EST) Regarding Mike Reid's program, I compiled it fine with DJGPP, an MSDOS compiler, but it didn't run. Any ideas? I'll probably get a windows compiler soon. Anyway, two other points I was thinking of recently. Firstly, do any of the kociemba algorithm search programs use the fact that you can perform a depth n search by 3 depth n-1 searches using the 3 orthogonal orientations? (if my logic is correct!) This is because if you are using the group for the final phase, the last move of any depth n sequence must end in a square move, in which case the n-1 will easily find it, or else a quarter turn in which case the three orientations are required to make it found in the second phase. Unfortuneately I doubt the n-1 searches could be replaced by n-2 searches. Also when a cube is being scrambled adjacent cubes tend to stay together for a while. Has this been of any use in search methods? (eg. to help prune the search tree). Obvously a sequence like F2 B2 U2 D2 L2 R2 separates all adjoining pairs, but there is still a high degree of order with next to adjacent cubes, so maybe they could be used as well. Andrew Walker