Author: Mridul Muralidharan
Date: 15:03:58 01/17/05
Go up one level in this thread
Hi Chandler Yergin,
Just out of curiosity , do you see this approach been taken by chessbase
programs when they play the game ??
No - but then you will have a large number of reasons to give for that !
Let us take a simple example :
[D]r2qk2r/ppp1b1pp/2n1p3/3pP1n1/3P2b1/2PB1NN1/PP4PP/R1BQK2R w KQkq - bm Nxg5; id
"NOLOT Position 3";
I dont think the required move will be listed in first 10 moves even after a 10
min search (assuming that this is the case - been ages since i fooled with this
position on top programs).
The point being - until they see the win after Nxg5 - they are blind to it ...
so searching 5 or 10 best move pv's will not help.
Most positions discussed in this forum are typically of this nature ... most ,
not all.
So the programmers try to analyze and find out why they did not find it faster ,
better and how to improve.
Let us do that math ... 5 moves (conservative) at root , for each of the 5 moves
, 5 best replies .... and it goes on. (exponential growth - notice it ??)
If you knew anything about chess programming - this would be categorised as very
very unsound pruning (pruning near the root ! :)) )
(You describe how humans play - that does not work for comps : has been tried in
various forms and always failed)
Anyway - hope this and other insight's that people have tried to give you will
help you in understanding the problem better.
Chessbase having written a module does not mean anything - I dont see any world
class chess programs by chessbase.
(The programs that you see are by authors who are affiliated with chessbase -
atleast I dont think frans morsch sits and codes the UI and the "module" that
you mention :D ).
Chessbase programmers might know less about programming than any decent
programmer out there considering the quality of their offering ...
Most of the programmers who post here are more expierenced and better at
programming in general and chess programming and its aspects in particular.
Among us , might have differences in opinion about various aspects : and some
of these might get heated up too (hence the long discussions in the archives
about when to probe EGT's , parallel search , etc) - but we all know and
understand what we talk about , and respect though might disagree with eachother
opinions.
Unless you learn the basic vocab , it will be quiet stupid for you to argue with
these statemeents here.
It is like trying to fight against "travelling salesman problem is NP complete"
You can try to prove it to be not so ... but unless you understand the math and
reasoning behind it , you might as well try to bust through a brick wall with
your head (Note this is an analogy - dont take it literally and hurt yourself !)
I am not saying this to convince you or argue with you - I am seeing quiet a
lot of "you guys are idiots" posts by you in particular.
The very reason that most have tried to convince you is that they want to make
you understand what you are truely saying.
Nobody really cares if you agree or not (atleast after all this vitriol that you
have spewed) - dont you think so ?
Does it really matter whether you agree ? Will next version of crafty , zappa ,
ferret , fritz really change based on your statements here ? (IF you make valid
points - definitely they might !!)
NOTE : all these are from a chess programmer's perspective and most of the posts
here are from that perspective - and since we wrote it in the first place : we
"might" know it better than others .... valuable inputs do not take the form "Do
it my way moron".
Anyway , enjoy chess and have fun !!
Mridul
PS : Let the flames begin !
On January 17, 2005 at 16:57:20, chandler yergin wrote:
>On January 17, 2005 at 11:32:35, Robert Hyatt wrote:
>
>>On January 16, 2005 at 15:33:38, chandler yergin wrote:
>>
>>>On January 16, 2005 at 13:10:08, Dieter Buerssner wrote:
>>>
>>>>On January 16, 2005 at 07:55:49, Yar wrote:
>>>>
>>>>>How long does it take for your engine to find kf3 ?
>>>
>>>DOES IT MATTER?
>>>The King has 6 "Possible" Moves....
>>>
>>>Use Multiple Lines... USE the TOOLS!
>>>Create 6 pV's in Parallel SIMULTANEOUSLY!
>>
>>Yes, and it takes 6x longer to compute those 6 pvs, which means the program
>>searches about 1.5-2.0 plies _shallower_ than it would normally search showing
>>only one PV.
>>
>>What exactly would this prove, therefore???
>>
>>You don't quite understand alpha/beta search yet, obviously...
>>
>>
>>>
>>>Each Move generates a Line..
>>>SEE how the Computer Evaluates ALL of them! At the SAME TIME!
>>>New game,
>>>[D]8/7R/1N2k1p1/P1p1b3/P3r3/7P/4K3/8 w - - 0 1
>>>
>>>Analysis by Shredder 8:
>>>
>>>1. +- (2.78): 1.Kf3 Rb4 2.Rb7
>>>2. +- (1.92): 1.Kd1 Rd4+ 2.Kc1 Rb4 3.Rb7 Kd6 4.Rb8 Kc6 5.Rc8+
>>>3. +- (1.78): 1.Kd2 Rb4 2.Kd3 Rd4+ 3.Kc2 Rb4 4.Kc1 Bf4+ 5.Kc2 Be5 6.Kc1 Bf4+
>>>7.Kc2 Be5 8.Kc1
>>>4. +- (1.73): 1.Kd3 Rd4+ 2.Kc2 Rb4 3.Kc1 Bf4+ 4.Kc2 Be5 5.Kc1 Bf4+ 6.Kc2 Be5
>>>7.Kc1
>>>5. = (0.03): 1.Kf1 c4 2.Rd7 c3 3.Rd8 Bd4 4.Re8+ Kf5
>>>6. µ (-0.74): 1.Kf2 c4 2.a6 Bd4+ 3.Kg2 Re2+ 4.Kf1 Rf2+ 5.Ke1 Bxb6 6.Rb7 Bc5 7.a7
>>>c3 8.a8Q Rf8 9.Qxf8 c2 10.Qxc5
>>>
>>>(, 16.01.2005)
>>>
>>>
>>>THIS, is the ‘absurdity' that I have with Programmers.
>>
>>It is "absurdity" of course, but _not_ with the programmers...
>>
>>
>>>
>>>You ‘wait'for 3 Hours?
>>>
>>>WHY?
>>>
>>>Of Course the Program finds 1. Kf3 Evaluates it as Best...
>>>So.. THEN Focus on the one Line!
>>>It Will Scroll that pV, and you haven't 'lost' any information.
>>>
>>>Indeed, you have gained information.
>>>
>>>
>>>
>>>>>8/7R/1N2k1p1/P1p1b3/P3r3/7P/4K3/8 w - - 0 47
>>>
>>>
>>>
>>>
>>>>
>>>>Yace almost immediately, loses it after some seconds. Finds it again after 3
>>>>hours (when I lost patience). Is Kf3 the only winning move?
>>>
>>>
>>>If you'd have used Multiple Lines, you'd Know; and could use those 3 Hours
>>>a bit more productively!
>>
>>Totally wrong conceptually.
>
>
>Wrong! I won't let you get away with this either!
>You think Chessbase doesn't understand their Modules?
>You think Chessbase Programmers don't understand Programming?
>You, have a lot to learn!
>
>Using the Tools THEY suggest, and the Method THEY suggest, is
>exactly what I have Posted!
>
>I get a lot of "Trash Talking" over here; and I'll "respond" in kind,
>to the willful ignorance of those who attack me with garbage!
>
>
>> The point is that the engine sees the right move
>>quickly, but then doesn't understand why it is right so that the move was just
>>lucky to be first. After a _long_ search the program discovers the tactical
>>reason that makes the move really best...
>
>How long is long....? Subjective huh?
>
>I said, and I say again..
>"USE the TOOLS" at your disposal!
>
>Using Multiple lines, generating 6, because the King ONLY has 6 possible
>moves... wait a decent amount of time.. 10 minutes, which is a Depth of search
>to about 18 or 20 Plies, then, FOCUS on the Main pV, and search THAT one in
>depth. If, in the deeper search the Program finds a better move, the line will
>be overwritten, however I doubt it will switch.
>
>There is NOTHING wrong with my 'Concept', it is the CHESSBASE concept.
>
>This is not Rocket Science!
>>
>>No program will understand this type of move at a shallow search unless it has
>>specific chess knowledge. In that case it would not need to search that deeply
>>to see the idea behind the move.
>>
>>This entire discussion is so ridiculous that any student participating in it
>>should be given a grade of "F" for not understanding basic concepts yet arguing
>>with those that do...
>
>I agree.... you get an "F-"
>
>Because obviously you, don't understand what is going on.
>
>
>>>
>>>>
>>>>Regards,
>>>>Dieter
>>>>
>>>> 107976929 2:02.5 1.12 14. 47.Kd1 Rb4 48.Rb7 Rd4+ 49.Ke2 Re4+ 50.Kf3 Rf4+
>>>> 51.Kg2 c4 52.a6 c3 53.a7 c2 54.a8=Q c1=Q
>>>> 55.Qe8+ Kf5 56.Rf7+ Bf6 57.Rxf6+ Kxf6 58.Nd5+
>>>> Kf5 {-80}
>>>> 196012611 3:43.2 1.15 15t 47.Kd1 Rb4 48.Rb7 Kd6 49.Nc8+ Kc6 50.Rxb4 cxb4
>>>> 51.Ne7+ Kb7H 52.Nxg6H Bc3H 53.Kc2H Ka6H 54.h4H
>>>> Bd2H {HT} {150}
>>>> 250601620 4:50.1 1.15 15. 47.Kd1 Rb4 48.Rb7 Kd6 49.Nc8+ Kc6 50.Rxb4 cxb4
>>>> 51.Ne7+ Kb7 52.Nxg6 Bc3 53.Kc2 Ka6 54.h4 Bd2
>>>> {HT} {150}
>>>> 466173985 9:02.1 1.20 16t 47.Kd1 Rb4 48.Rb7 Rd4+ 49.Kc1 Kd6 50.Rb8 Kc6
>>>> 51.Rc8+ Bc7 52.Na8 Rc4+ 53.Kd2 Kb7 54.Rxc7+
>>>> Kxa8 55.Rc6 Rd4+ 56.Kc3 Rh4 {HT} {80}
>>>> 653549812 12:56.4 1.20 16. 47.Kd1 Rb4 48.Rb7 Rd4+ 49.Kc1 Kd6 50.Rb8 Kc6
>>>> 51.Rc8+ Bc7 52.Na8 Rc4+ 53.Kd2 Kb7 54.Rxc7+
>>>> Kxa8 55.Rc6 Rd4+ 56.Kc3 Rh4 {HT} {80}
>>>>1088869997 20:39.7 1.60 17++ 47.Kd1 Rb4 48.Rb7 Rd4+ 49.Kc2 c4 50.a6 c3
>>>> 51.Rd7 Rxd7 52.Nxd7 Bd4 53.Nf8+ Kf7 54.Nh7 Bb6
>>>> 55.Kxc3 Kg8 56.a7 Bxa7 57.Ng5 {HT} {70}
>>>>1623915033 30:01.4 1.62 17t 47.Kd1 Rd4+ 48.Kc1 Rb4 49.Rb7 Kd6 50.Rb8 Ke6
>>>> 51.Re8+ Kf5 52.a6 Rxb6 53.a7 Ra6 54.a8=Q Rxa8
>>>> 55.Rxa8 c4 56.Rc8 Bf4+ 57.Kc2 {HT} {150}
>>>>1953247137 36:32.4 1.62 17. 47.Kd1 Rd4+ 48.Kc1 Rb4 49.Rb7 Kd6 50.Rb8 Ke6
>>>> 51.Re8+ Kf5 52.a6 Rxb6 53.a7 Ra6 54.a8=Q Rxa8
>>>> 55.Rxa8 c4 56.Rc8 Bf4+ 57.Kc2 {HT} {150}
>>>>2878266080 53:20.6 1.65 18t 47.Kd1 Rd4+ 48.Kc1 Rb4 49.Rb7 Kd6 50.Rb8 Ke6
>>>> 51.Re8+ Kf5 52.a6 Rxb6 53.a7 Ra6 54.a8=Q Rxa8
>>>> 55.Rxa8 c4 56.Re8 Ke4 57.Kd2 c3+ {HT} {150}
>>>>3788563414 1:12:24 1.65 18. 47.Kd1 Rd4+ 48.Kc1 Rb4 49.Rb7 Kd6 50.Rb8 Ke6
>>>> 51.Re8+ Kf5 52.a6 Rxb6 53.a7 Ra6 54.a8=Q Rxa8
>>>> 55.Rxa8 c4 56.Re8 Ke4 57.Kd2 c3+ {HT} {150}
>>>>2054909735 2:05:09 1.59 19t 47.Kd1 Rd4+ 48.Kc1 Rb4 49.Rb7 Kd6 50.Rb8 Kc6
>>>> 51.Rc8+ Kb7 52.Rxc5 Bh2 53.Rc4 Rb3 54.Rh4 Ka6
>>>> 55.Kc2 Rf3 56.Nc4 Rf2+ 57.Kd3 Bf4 {HT} {150}
>>>> 568389920 3:04:14 1.60 19t+ 47.Kf3 Rf4+ 48.Kg2 Rb4 49.Rb7 Rb2+ 50.Kf3 Rb3+
>>>> 51.Ke2 Bf6 52.a6 Rg3
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.