Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit, trying to answer various missed posts

Author: Uri Blass

Date: 07:07:34 03/15/04

Go up one level in this thread


On March 15, 2004 at 06:22:09, Fabien Letouzey wrote:

>
>Hello,
>
>I missed many posts during the week-end.
>Sorry that I am putting the answers into a single post.
>I think it's useless to answer old posts here (they get unnoticed, or?).
>
>Please only quote selected parts of this post if you reply, thanks!
>
>Fabien.
>
>---
>
>On March 12, 2004 at 16:14:50, Uri Blass wrote:
>
>>On March 12, 2004 at 14:49:28, Steve Maughan wrote:
>>
>>>Amazing!
>>>
>>>I'm amazed at Fruit's score.  I've downloaded it and it seems like a robust
>>>engine but I note that it's quite slow and according to the author has little
>>>knowedge above and beyond piece tables.  Monarch is my engine.  It's nowhere
>>>near as strong as Fruit - I'd estimate Fruit is +200 ELO stronger than Monarch.
>>>Like Fruit, Monarch also has just piece tables and I believe that Monarch is
>>>probably even a little faster than Fruit.  Like Fruit Monarch also has Null
>>>move, check extensions and hash table (2 probes) i.e. on paper they seem to be
>>>close in spec.  So the question is what makes Fruit so strong?
>>>
>>>a) Fewer bugs - quite possible!
>>>b) ETC - Monarch doesn't have this - Fruit does
>>>c) No PV cut-off's in Fruit
>>>d) Better piece square tables
>>>e) Combination of b, c & d
>>>f) ?
>>>
>>>Of course another explanation is that Fruit is much more sophisticated than the
>>>author claims but I see no reason to think this.
>>>
>>>Any thoughts?
>>>
>>>Steve
>>
>>It has some boolean features.
>>
>>I understood that it also has some pawn structure evaluation.
>>
>>Uri
>
>a) maybe; but Fruit is untested.
>b) I only expect effects in very few positions, but I think it doesn't hurt.
>c) makes the search more stable, but I think it hurts overall (much slower
>"next-move search" startup)
>d) no
>...
>
>Uri is right;
>
>1) I use a "static" pawn eval (no interaction with pieces).  Only basic pawn
>types (inc. passed) + a "bogus" definition of candidates (some people call those
>"hidden passed pawns"?).
>2) There are boolean features for sliders, they are penalysed when "behind" "own
>pawns" (e.g. behind for bishops means "is there a 'own pawn' along forward
>diagonals").
>
>It never crossed my mind to write a piece-square-table program.
>Pawn eval is one of my main goals in the long term.
>
>---
>
>On March 12, 2004 at 18:02:56, Uri Blass wrote:
>
>>It may be interesting to compare Crafty with fruit's evaluation to Fruit to see
>>if Fruit really has big advantage relative to Crafty in search rules or I simply
>>overestimated the value of Crafty's evaluation.
>>
>>Uri
>
>I don't understand at all, of course.
>1) I consider Crafty to be at least 200 elo points stronger than Fruit, making
>any comparison a loss of time.

We are going to see later.
There is some evidence that the difference from Crafty in the nunn list may be
more than 100 elo based on nunn top tournament but I still expect less than 200
elo difference.

I think that fruit can be improved by better weights(probably by increasing the
values for knight and bishop to be better than rook and pawn)

See the following game when fruit overestimated Rook and pawn against knight and
bishop.

It showed a score of +2.34 at move 38 that is unrealistic score
I believe that it could draw later by 62.R7h6+ but it did not want a draw and
evaluated itself as better.

(5853) Fruit 0.98b - Pharaon 2.62 [A48]
Nunn Top Pos.3 MyTown (2.5), 15.03.2004


1.d4 Nf6 2.Nf3 g6 3.Bf4 Bg7 4.e3 d6 5.h3 0-0 6.Be2 c5 7.c3 b6 8.0-0 0.43/10 11
Bb7 (Nc6) 0.12/8 6 9.Nbd2 0.43/10 11 Nbd7 (Nc6) 0.07/8 10 10.Nc4 (Bc4) 0.42/8 7
10...Re8 (d5) 0.02/8 14 11.Qb3 (Rc1) 0.46/9 11 11...Bd5 (Qc7) -0.05/8 8 12.Rfd1
(Rad1) 0.39/8 10 12...Qc7 (Qb8) -0.12/8 8 13.Rac1 0.28/8 8 e5 -0.14/8 8 14.Bh2
(Bg3) 0.23/9 9 14...a6 (Qc6) -0.22/8 8 15.dxe5 (Qa3) 0.65/10 7 15...Bxc4 (dxe5)
1.05/9 6 16.Bxc4 1.16/12 6 Nxe5 0.91/11 8 17.Bxe5 1.11/12 9 dxe5 0.80/11 6
18.Ng5 0.93/11 9 Rf8 (Re7) 0.76/10 6 19.a4 0.84/11 9 Rab8 (Bh6) 0.68/10 7
20.Bxa6 1.16/10 7 e4 0.70/11 7 21.Be2 (h4) 1.13/10 7 21...h6 0.58/11 7 22.Nxf7
1.20/12 11 Rxf7 0.64/11 5 23.Bc4 (Rd8+) 1.16/12 6 23...Rbf8 (Bh8) 0.64/11 7
24.Rd2 (Bxf7+) 1.36/12 8 24...Kh7 (Bh8) 0.79/11 7 25.Bxf7 1.27/13 7 Rxf7 0.86/11
4 26.Rcd1 (Rd8) 1.33/13 8 26...Bf8 0.88/10 7 27.Qe6 1.32/12 7 Kg7 0.99/10 7
28.Rd8 1.35/11 6 Qa7 (Qb7) 1.04/10 7 29.Qc6 (b3) 1.33/10 6 29...Rc7 (h5) 0.77/10
5 30.Ra8 (Qb5) 1.86/12 4 30...Qb7 1.74/12 6 31.Qxb7 2.06/14 7 Rxb7 1.47/13 5
32.Rdd8 2.01/14 5 Nd7 (Rf7) 1.18/12 6 33.Re8 2.15/13 7 Kf7 1.14/12 4 34.Rxe4
2.16/13 4 Nf6 1.13/12 6 35.Rf4 2.12/13 7 Bd6 1.13/11 5 36.Rc4 1.96/12 7 Ke6
(Bb8) 1.14/10 8 37.b3 (f3) 2.16/11 6 37...Nd5 (b5) 1.14/9 6 38.Rh8 (e4) 2.34/9 4
38...h5 1.07/9 4 39.Rg8 (g4) 2.31/10 4 39...Ne7 (Kf7) 1.08/11 6 40.Ra8 2.16/11 6
Nd5 (Nc6) 1.05/9 6 41.e4 (Rd8) 2.10/11 6 41...Nf4 0.98/10 5 42.Rg8 1.67/10 6 Be7
(b5) 0.80/10 5 43.b4 1.31/10 4 Kf7 (b5) 0.15/11 6 44.Rh8 1.08/12 6 Nd3 0.02/11 6
45.e5 (bxc5) 0.52/11 5 45...Nxe5 (Bf8) 0.00/10 6 46.Re4 0.86/12 6 Kf6 0.00/10 8
47.Re8 0.81/11 7 cxb4 0.00/10 6 48.cxb4 0.59/11 5 Kf7 0.00/10 6 49.Rh8 0.36/13 5
Kg7 0.00/11 4 50.Rc8 (Re8) 0.15/13 5 50...Nd3 0.00/11 6 51.Rcc4 (Re8) 0.17/12 4
51...Nb2 (Rd7) 0.06/11 9 52.Rc6 0.71/13 10 Nxa4 -0.02/11 4 53.Ree6 (Rxe7+)
0.36/12 5 53...g5 -0.07/12 5 54.Rh6 0.24/12 5 Rd7 -0.16/11 5 55.Rxh5 0.16/12 5
Rd1+ -0.20/10 4 56.Kh2 0.73/1 0 Rd2 (Bd6+) -0.15/11 5 57.f3 0.19/11 4 Bf6 (Rb2)
-0.10/10 6 58.f4 (b5) 0.24/10 3 58...Nc3 (gxf4) -0.09/9 8 59.h4 (Rxb6) 0.34/11 5
59...gxh4 (gxf4) -0.13/10 5 60.Rc7+ (Rxb6) 0.03/11 5 60...Kg6 -0.21/11 4 61.Rch7
0.00/12 3 Bd4 (Bg7) -0.17/10 4 62.Rxh4 (R7h6+) 0.12/10 2 62...Kf5 (Nd5) -0.24/9
4 63.Rf7+ (R7h5+) 0.10/10 4 63...Ke4 (Ke6) -0.35/9 4 64.Rh6 (Re7+) 0.15/10 3
64...Nd5 -0.50/10 4 65.Re6+ 0.07/11 4 Kd3 -0.29/11 6 66.Kh3 0.09/11 4 Rf2 (Kc4)
-0.34/10 4 67.g3 0.00/11 2 Nxb4 -0.45/10 3 68.Rd7 0.00/12 4 Kc4 -0.42/10 4
69.Kg4 (Rc7+) -0.02/12 4 69...Nc2 (Nd5) -0.41/9 4 70.Kf5 (Kg5) -0.11/11 4
70...b5 -0.62/9 4 71.Kg5 (g4) -0.11/10 3 71...b4 -0.89/9 4 72.Rc6+ -0.27/10 4
Kd3 -0.89/10 12 73.g4 (Rb7) -0.34/11 4 73...Rg2 -1.04/10 3 74.Rb6 (Kh5) -0.43/11
3 74...Ke4 -1.58/9 2 75.Rh6 -1.27/11 4 b3 (Rg3) -2.27/9 3 76.Re7+ -1.91/10 4 Kd3
-1.84/9 3 77.Rd7 (Rb7) -1.97/11 4 77...b2 -3.17/9 3 78.Rb6 -2.31/11 4 Kc3 (Ke4)
-3.16/8 3 79.Rc7+ (Rb8) -2.32/10 4 79...Kd2 -3.09/10 4 80.Rb3 -2.87/10 9 Kc1
(Ne3) -3.49/9 3 81.Rcb7 -2.29/11 2 Ne3 -3.67/9 3 82.f5 (Kh5) -3.03/10 3
82...Rxg4+ -4.83/7 3 83.Kh5 -3.56/11 3 Rf4 (Rg1) -4.65/7 3 84.f6 (Kg6) -3.03/10
2 84...Rf5+ (Rxf6) -5.38/7 3 85.Kg6 -6.05/11 3 Rxf6+ -5.38/7 3 86.Kh5 (Kg5)
-6.12/11 3 86...Kc2 -7.90/8 3 87.R3b4 -6.21/12 3 Bc3 -8.18/7 3 88.Re4 -6.15/11 2
Rf5+ -8.18/7 9 89.Kg6 -6.32/12 3 Rf3 -8.36/7 2 90.Rh4 (Kg5) -6.80/12 3 90...Rg3+
-8.38/7 3 91.Kh5 -6.82/12 3 b1Q -8.36/7 2 92.Rxb1 -6.91/13 2 Kxb1 -8.47/7 1 0-1




>2) I don't believe a second that Fruit's search is superior to Crafty's in any
>way.
>
>Uri, you have already mentionned ETC.
>Maybe programmers can post here result of matches, same program w/ and w/o ETC.
>I would be surprised if it affected match results at all, although I like to
>think it can "save the day" in endgame now and then.

I also think that checks in the first ply of the qsearch together with
R=3 may be superior to crafty.

It may be also interesting for me if you use hash in the qsearch and if yes how
do you do it(today I do not hash the qsearch).

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.