Author: Koundinya Veluri
Date: 18:37:34 04/13/03
Go up one level in this thread
On April 13, 2003 at 11:35:23, Uri Blass wrote: >On April 13, 2003 at 11:05:07, Robert Hyatt wrote: > >>On April 13, 2003 at 03:22:27, Tim Foden wrote: >> >>>On April 13, 2003 at 00:56:04, Robert Hyatt wrote: >>> >>>>On April 12, 2003 at 22:01:10, Uri Blass wrote: >>>>> >>>>>Yes but if the pawn is not captured the program in the search the program has no >>>>>reason to let the rook trade. >>>>> >>>>>simple evaluation that give bonus to every passed pawn is enough. >>>> >>>>You have _got_ to be a better chess player than that. You are saying all >>>>passed pawns are equal. They most definitely are not. Pawns on the a,b c >>>>file for both sides. My passer is on the h file. Yours is on the e file. >>>>You _still_ think that a bonus for a passer is enough? I have a passer already, >>>>you can trade rooks to create your passed e pawn. Smart move? I don't think >>>>so... >>> >>>I realise it's not clear from his answer at this point in the message, but I >>>beleive Uri is talking _specifically_ about the single position posted. He >>>believes that distant passed pawn info is not required to solve _this_ position. >> >>I believe it depends. In this position, you trade rooks into a lost position. >>The position is lost first because black has a distant passer, and then because >>that distant passer is given up to lure the king away from the pawns leaving >>black with a won ending. If you can't search deep enough in this position to >>see that the passer has to be given up to make progress, then you simply won't >>see the solution. If you can search deeply enough to see that the passer has >>to be given up, then you need the knowledge about king location to realize that >>the ending is _still_ won with the distant passer removed. >> >>> >>>If you had read to the bottom of Uri's reply before replying, I believe that you >>>would have understood this :) See below... >>> >>>>>I agree that evaluation of distant passed pawn is necessary but the position was >>>>>not a good example because program with only small bonus for passed pawns and >>>>>knowledge to evaluate pawns relative to the king can also solve it. >> >> >>I don't see how, which was the point of my reply. White is increasing the >>value of the black distant passer by trading rooks. White needs to know >>that the value of a distant passer increases as pieces are removed. > >White does not need to know it in this case. >The point is that white can get equality by other means > >White threats the pawn at b6 in the root position and even a short search of 4 >plies can reveal that white can without trading rooks do one of the following: >1)force repetition by perpetual check or (if black choose not to threat check) >2)trade the passed pawn of black for another pawn. > >The only reason for white to choose trading rooks is if white can see more than >equality by trading rooks and it happens only if the search can see winning the >passed pawn. > >Uri After trading rooks, it is two different things for the search to see winning the passed pawn, and for the search to realize that white loses after winning the passed pawn. I have explained this in another post. White will try to avoid capturing the passed pawn for as long as possible, because as soon as it is captured, the black king will be allowed into white's territory. To allow the trade of rooks, the search only has to see that white can capture the pawn, regardless of how long it takes to capture the pawn. So the search will settle for a very short variation where the white king goes directly for the passed pawn after rooks are traded. However, this is not the variation that will be played. The white king instead will delay capturing this pawn until it is forced to capture it, and the variation in which the white king is forced to capture the passed pawn will be much longer than the variation that allowed trade of rooks without penalty. Until search can see that white is *forced* to capture the passed pawn, it won't realize that black is winning, and the king close to pawns evaluation won't even be executed. So just because the program is allowing the rook trade does *not* mean that it can see white being forced to capture the passed pawn. Koundinya
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.