Author: Robert Hyatt
Date: 08:58:11 01/13/03
Go up one level in this thread
On January 13, 2003 at 06:23:29, Marc Bourzutschky wrote: >On January 13, 2003 at 00:28:13, Robert Hyatt wrote: > >>On January 12, 2003 at 15:56:30, Marc Bourzutschky wrote: >> >>>On January 12, 2003 at 14:24:24, Rafael Andrist wrote: >>> >>>>On January 11, 2003 at 14:46:30, Marc Bourzutschky wrote: >>>> >>>>>I have some code that might be cobbled together to do the conversion. However, >>>>>I agree with Guy that it might be easier to just generate DTC from scratch, >>>>>especially since the Thompson set is incomplete (particularly missing endings >>>>>with more than one pawn) and does not have wtm. I'm also not sure how many >>>>>people still have the Thompson data. >>>> >>>>As it seems not many people are interested in DTC data. Therefore you may be >>>>right that it is easier to generate DTC from scratch using modified code of >>>>Nalimovs EGTB generator. On the other hand it would be easier for people who >>>>still have the Thompson data to just convert it. I think I will do some >>>>experiments with Thompsons code - if it is so slow that a conversion is not much >>>>faster than re-generation, I will better use the Nalimov code. >>>> >>>>regards >>>>Rafael B. Andrist >>> >>>Conversion will definitely be faster than re-generation, especially if you use >>>the Nalimov code :-) For Thompson data the main thing you'll have to write is a >>>function that decodes a whole block at a time, otherwise it will be very slow. >>>code.c on the Thompson disks only has code to read a single position at a time. >>> >>>-Marc >> >>I don't see how you are going to "convert" the data. You are going to have to >>do the same sort of work starting at endpoints and working backward, to find >>out how many moves pass before a pawn is moved or you translate into another >>tablebase file... > >By conversion Rafael just means translating the existing Thompson data to use >the same indexing scheme as Eugene, but keep their DTC values. This would >mainly involve reducing the Thompson data a bit to use like piece symmetries and >eliminate certain types of illegal positions like Eugene does. Having done a >similar exercise before, I think this would probably take a few minutes for a >typical 5-man ending. But putting together working code might take a lot longer >unless you have most of the components already, and you might be better off >re-generating the DTC data from scratch which would take a few days for the >complete 5-man set. OK... then I misinterpreted what he meant. I interpreted that to mean take a Nalimov table and convert it from DTM to DTC... Which should be doable, but it would be a _lot_ of computation, probably about as hard as just computing a DTC from scratch.
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.