Author: Uri Blass
Date: 19:24:44 12/07/05
Go up one level in this thread
On December 07, 2005 at 11:49:03, Uri Blass wrote: >On December 07, 2005 at 11:36:06, Laszlo Gaspar wrote: > >>On December 07, 2005 at 10:43:28, Uri Blass wrote: >> >>>On December 07, 2005 at 10:01:33, Laszlo Gaspar wrote: >>> >>>>On December 07, 2005 at 07:25:26, Tord Romstad wrote: >>>> >>>>>On December 07, 2005 at 05:31:51, Barry Culp wrote: >>>>> >>>>>>On December 06, 2005 at 23:11:38, Uri Blass wrote: >>>>>> >>>>>>>results of more matches >>>>>>> >>>>>>>Rybka(2 plies)-Movei(4 plies) 12.5-7.5 >>>>>>>Rybka(3 plies)-Movei(5 plies) 7.5-12.5 >>>>>>>Rybka(4 plies)-Movei(6 plies) 11-9 >>>>>>>Rybka(5 plies)-Movei(7 plies) 10.5-9.5 >>>>>> >>>>>>What's the point of running these handicap matches ? Please explain >>>>> >>>>>It is commonly believed that the difference between a search to >>>>>depth N and a search to depth (N+d) is less significant when N is >>>>>big. This seems intuitively plausible, but as far as I know we >>>>>don't have much experimental support for this hypothesis. >>>>> >>>>>Experiments like Uri's could help us to learn more about this. >>>>> >>>>>Tord >>>>Dear Tord and others, >>>> >>>>This is my first post on this forum but I would like to share my experience in >>>>this topic, since I also have an own engine and researched a bit in this >>>>direction. Interesting but I found that every ply increase in search gives an >>>>almost same ELO increase. This means that my weak engine which is let's say 450 >>>>ELO weaker than Fruit at depth=4 is almost equal at depth=6 (when Fruit remained >>>>at 4 ply) and better with depth=7 and so on. This increase seems to be different >>>>but constant in relation of two specific engine and it's around 200ELO, which is >>>>quite big. >>>>I think this is very important to know and it has quite a few consequenses: >>>> >>>>1. Any dumb engine can be champion on a fast hardware. >>> >>>I disagree about it. >>>I do not believe tscp can be a champion on a fast hardware. >>> >>>>2. You can improve your engine quite a lot if you improve the evaluation >>>>function only or tune your extensions. >>> >>>I agree. >>>>3. You can neglect the evaluation function and concentrate on search only to >>>>reach higher depth. >>> >>>You can do it and get improvement but you will miss possible improvement. >>> >>>>4. Not important to search everything to high depth just the important >>>>lines(extensions, reductions). >>> >>>I agree. >>>>5. Time management is important. >>> >>>I agree. >>>>6. Testing development version of an engine can be more simple, since a fixed, >>>>low depth test match can be done quickly. >>> >>>I agree. >>>>7. In my opinion the ELO rating can be estimated for any time control by simple >>>>maths though still I didn't figure out the exact formulation (we have to >>>>calculate the average search depth for a specific TC, which is not hard if we >>>>know the average branching factor). >>> >>>I do not think that it is so simple. >>> >>>> >>>>In the view of these I consider Uri's data very interesting and useful although >>>>the number of games is low to justify the aboves. >>>> >>>>Best regards, >>>>László >>> >>>I run now a match of movei against tscp >>>I think that I believe that I can let tscp outsearch movei by more plies if the >>>depth is bigger and still win but we are going to see(note that movei does mopre >>>pruning and more extensions(checks in the qsearch) so I may change movei to use >>>similiar algorithm as tscp for comparison later. >>> >>>first match was >>>Tscp(5 plies)-Movei(4 plies) >>> >>>Result is 15.5-4.5 for Movei >>> >>>second match is tscp(6 plies)-Movei(4 plies) >>>tscp is leading 5.5-1.5 for tscp at this moment(no wins for movei so far). >>> >>>Unfortunately I am afraid that more plies for tscp may take a long time. >>> >>>I expect movei to lose with 4 plies against 6 plies of tscp but to win with >>>6 plies against 8 plies of tscp but I may be wrong about it. >> >>Hi Uri, >> >>When I sad in my first statement that "on a fast hardware", I meant fast enough >>:-)! It can be 100 or 1000 or more times than the opposition's one. > >100 or 1000 is not enough for tscp to beat other programs even if you remove >some problems that make it unable to search more than 32 plies and you need >probably more speed. > > And if it is >>fast enough it will play good chess...This is important because relatively big >>differences can be equalized by speed. > >I agree that enough speed can compensate. > >>I think your test against TSCP will also justify it. (I 'm glad you do it!) But >>it is not sure at all that if TSCP wins in the 6 to 4 ply match then the same >>result would happen in case of 8 to 6. But the 9 ply TSCP will be better again >>than the 6 ply Movei and so on. > >I agree about this. > > This is because the ELO increase constant can be >>different for TSCP and Movei in their relation, let's say 180 for TSCP and 250 >>for Movei and Movei improves faster. >> >>I' m waiting for your findings with interest! >> >>Best regards, >>László > >6 vs 4 plies was 14-6 for tscp >7 vs 5 plies is running at this moment and result so far is 0.5-0.5 > >Uri results so far Tscp wins 6 vs 4 14-6 7 vs 5 15-5 Movei wins 4 vs 5 15.5-4.5 this seems too optimistic so I played the match that is considered by chessbase as the nunn2 match(not the original one) and the result in it was 29.5-20.5 so the total result is 45-25 5 vs 6 13-7 It seems that tscp can win also 8 vs 6 based on the data but I suspect that movei does more pruning than extensions at high depthes so I may try version with no pruning. Uri
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.