Author: Guido
Date: 02:13:58 05/01/03
Go up one level in this thread
On April 30, 2003 at 19:13:54, GuyHaworth wrote: > >I forgot to say that you must have done some significant programming to >implement both: > >a) the 'forward-pass' Edwards/Nalimov full-EGT repeated-trawl, and > >b) the 'retro' Thompson/Wirth/Wu-Beal EGT generation. > >The only 'trick' in the WU-Beal code is that it only needs a bitvector-based >approach for DTM as 'mate in N' is only backed up on the Nth cycle. > >g I use a bit vector only for 'lost in N' and not for 'mate in N'. Probably I use a different approach from WU-Beal. An interesting thing is that in verifying retro generated tables I use the 'forward-pass' already implemented in the forward generation. This gives a better, also if not mathematical, guarantee of the correctness of the tables, apart the possibility to compare directly tables generated by the two different ways. The main problem in my code remains the optimization vs CPU time. My indexing and de-indexing is very slow compared to other programs, for example tbgen, but the factor found in reduction of CPU time with retro generation is equally correct as only the algorithm has been modified. Guido
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.