Author: Steve Coladonato
Date: 10:26:25 11/22/04
Go up one level in this thread
On November 22, 2004 at 12:24:28, Laurens Winkelhagen wrote: >On November 22, 2004 at 11:20:19, Steve Coladonato wrote: > >>On November 22, 2004 at 09:16:58, Laurens Winkelhagen wrote: >> >>> >>> >>>Hi Steve, >>> >>>I'm not sure if I understand you correctly, so I don't know if this reply will >>>make sense to you, but: >>>tablebases only tell only one side of the story: there is a KRPKBP tablebase but >>>there won't be a KBPKRP tablebase. The board is flipped appropriately. There are >>>obviously positions where KBP will win over KRP, which is why there are both >>>maxDTM for 1-0 and for 0-1. There wouln't need to be 0-1 maxDTM scores when >>>there would be a seperate KBPKRP tablebase, but there isn't;-) >>> >>>Hope I make sense, Laurens. >> >> >>Thanks Laurens. >> >>It's not totally clear yet. I will have to think over the concept of flipping >>the board to figure out why tablebases provide the statistics for either side >>winning. >> >>In the example I gave, white had the rook and black had the bishop. Hence, I >>don't think there is a need for flipping the board and the tablebases should >>determine whether white wins, loses, or draws. Is that correct? >> >>Thanks. >> >>Steve > >Indeed. But I feel I have obfuscated the issue: > >One possible source of confusion occured to me while I was browsing through some >EGTB-probe code: When I said flipping the board, I actually meant inverting the >color of the pieces. > >Let's use some of the Crafty Code Comments to clarify things: > >/* > ************************************************************ > * * > * yes, finish setting up to probe the tablebase. if * > * black is the "winning" side (more pieces) then we need * > * to "invert" the pieces in the lists. * > * * > ************************************************************ > */ > >In your example, white is the "winning" side, hence the board need not be >flipped. > >Now, the next step is clear: > >/* > ************************************************************ > * * > * now check to see if this particular tablebase for this * > * color to move is registered. * > * * > ************************************************************ > */ > >Probing the tablebases will now give one of three possible types of outcome: >0 = it's a draw >positive = I (site-to-move) am gonna win in xxx half-moves >negative = I'm gonna lose in xxx half-moves >Actually, I didn't even see your first post until now. > >So, in actuality there are two KRPKBP tablebases, one for wtm and one for btm. >Both give both 1-0 and 0-1 results. This is why we can distinguish four maxDTM >values. Inverting the pieces is only relevant in case black has KRP and white >KBP. > >Regarding 1/2-1/2 results: "maxDTM" for 1/2-1/2 is by definition pointless, >which is why the EGTB's return the value "0" for such positions. > >I Hope this clarifies things, Laurens, author of Jan Willem (with a big help >from kind Prof. Hyatt regarding the EGTB-probe code, thanx again, Robbert >Hyatt:-) Laurens, Thanks for hanging in there. It makes sense. Steve
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.