Toga II 3.1.2SE

[pedrox]

Hi Elco,

You must test these changes on history prunning with games, do so only with positions is dangerous, in some positions will be better and others worse.

[mariaclara]

wow

thanks for this latest toga version...

can hardly wait to get it goin and chessing on Arena...

[Eelco de Groot]

Pedro wrote:

Hi Elco,

You must test these changes on history prunning with games, do so only with positions is dangerous, in some positions will be better and others worse.

wow

thanks for this latest toga version...

can hardly wait to get it goin and chessing on Arena...

Thanks people! And Pedro, yes, the proposed code seemed to have very serious horizon limitations if I start at depth 6. I cannot recommend it especially the double/triple reductions but I'm trying something slightly different now.

Some HOT News:

Due to the magnificent work of Shaun Brewer, who played no less than 607 testgames with Toga against some very strong opponents, I can report that Toga Checkov Beta 3 now has an official CCRL rating!

http://kirr.homeunix.org/chess/discussi ... 1186d0d3ce

Single engines, apart from Toga II 1.2.1par:

Toga II 1.4 beta5 2921 +23 −23 58.7% −59.5 37.7% 640
Toga II 3.1.2SE 2909 +22 −22 54.8% −32.6 37.8% 703
Toga II Checkov_beta 3.1.2SE 2894 +24 −23 54.7% −31.9 33.6% 607
Toga II 1.4 beta4a 2884 +20 −20 50.0% +0.8 37.3% 785
Toga II 1.2.1par 2CPU 2884 +32 −32 48.3% +11.5 34.9% 318

Wonderful Shaun! How can I thank you!

One small thing I can maybe say in return is that, inspired by this result, I sat down again to do some code and the Fischer (LMR) Reductions are now officially an existing item!

Only thing is I don't know yet exactly what the Fischer Reductions are doing!

This is the result for Toga Checkov Beta with FR, Fischer LMR Reductions, in Alekhine - Sterk:


[d]r4rk1/p4ppp/qp2p3/b5B1/n1R5/5N2/PP2QPPP/1R4K1 w - -

Engine: Toga Checkov Beta 4j (256 MB)
by Thomas Gaksch and Fabien Letouzey


7/22 0:00 +1.41++ 1.Qf1 h6 2.Be3 g6 3.Rxa4 Qxf1+
4.Rxf1 (61.566)

8/22 0:00 +1.36 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Bxc5 Rxc5 6.Rxc5 bxc5 (77.289)

9/24 0:00 +1.30 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Ne5 Nb3 6.Rf4 (141.148)

10/33 0:00 +1.42 1.Qf1 Nc5 2.b4 Bxb4 3.Rcxb4 Qxf1+
4.Rxf1 Rac8 5.Rd1 f6 6.Bf4 Rfd8
7.Bd6 Nd3 (365.867)

11/33 0:01 +1.29 1.Qf1 Nc5 2.b4 Bxb4 3.Rcxb4 Qxa2
4.Ra1 Qc2 5.Be7 Rfc8 6.Rc4 Qf5 7.Nd4 Qe5 (828.724) 570

12/40 0:05 +1.30 1.Qf1 Nc5 2.b4 Bxb4 3.Rcxb4 Qxa2
4.Ra1 Qc2 5.Be7 Rfc8 6.Rc4 Qf5 7.Nd4 Qd5
8.Bxc5 Rxc5 (3.114.851) 570

13/40 0:08 +1.33 1.Qf1 Nc5 2.b4 Bxb4 3.Rcxb4 Qxa2
4.Ra1 Qc2 5.Be7 Rfc8 6.Rc1 Qd3
7.Rbc4 Qf5 8.Nd4 Qe5 9.Bxc5 Rxc5
10.Rxc5 bxc5 (4.644.341) 566

13/40 0:10 +1.54 1.Nd4 Nc5 2.b4 Ne4 3.Be7 Rfe8 4.bxa5 Nd2
5.Qxd2 Qxc4 6.Rc1 Qd5 7.Rc7 Qxa5
8.Qxa5 bxa5 (5.743.787) 565

14/40 0:17 +1.51 1.Nd4 Nc5 2.b4 Ne4 3.Bf4 Rae8 4.Nb5 Qxb5
5.Rxe4 Qxe2 6.Rxe2 Bxb4 7.Rxb4 Rd8
8.Re1 Rd5 9.Rc4 Rfd8 (9.998.447) 563

14/45 0:41 +1.94++ 1.Bf6 Rfc8 2.Qe5 Bc3 3.bxc3 Rxc4
4.Qg5 b5 5.Bxg7 h5 6.Bh6+ Kh8 7.Rxb5 Rg4
8.Qf6+ Kh7 9.Qxf7+ Kh8 10.Qxh5 Nxc3
11.Qxg4 Nxb5 (23.851.969) 572

15/62 3:17 +3.14++ 1.Bf6 Rfc8 2.Qe5 Bb4 3.Rxb4 Qd3
4.Re1 Qg6 5.Be7 Nc5 6.Nh4 Qc2 7.Rg4 g6
8.Nf3 Nd3 (116.681.002) 590

16/76 8:05 +3.66++ 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qh4 h5 7.Qg5 Kh7
8.Qe5 Qb7 9.b4 Bc7 10.Ng5+ Kg8 (285.460.231) 588

17/76 9:50 +3.38 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qh4 Qd6 7.Be5 Qd3
8.Rc1 Qf5 9.b4 Bb6 10.Qh6 f6 (349.314.037) 591

18/76 18:14 +3.71 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qa3 b4 7.Qa4 Qd3
8.Rd1 Qf5 9.Be7 Kg7 10.Qd7 Qc2
11.Qd4+ Kg8 (645.790.620) 590

19/86 33:19 +3.89 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qa3 b4 7.Qa4 Qd3
8.Rd1 Qf5 9.Qc6 Rf8 10.Be7 Rb8
11.Nd4 Qf4 (1.186.428.468) 593


Not bad at all I think, while the solution is found at a greater depth, 14 ply, than with Toga Checkov Beta 4 and just a little bit slower, but a possible offset is that mysteriously for me the solution depth 14 is reached in a very short time. And the PV with 1.Bf6 Rfc8 2.Qe5 Rxc4 is right this time, found at 8 minutes, it's not a record mind you, but also not very bad. This time somehow no terrible horizon effects are visible...
Fischer LMR reductions already start at depth = 5 How will this end This cannot possibly go right!

Eelco

[Eelco de Groot]

Sadly, this position Toga Checkov Beta 4j now can't solve anymore in twenty seconds going over Arasan 7 testsuite. I hope that this phenomenon only occurs if the position seems already lost. For Mr. Tueschen; the new reductions are now called Fischer History Reductions, not to be confused with Fail High Reductions from Ed I Believe, and Random Transpositions is another phenomenon displayed mainly by Fritz, as Christopher is trying to explain in another post. You almost got it right though. Just working on getting the randomness under control.

[d]q2r2k1/1bpn1ppp/1p2p3/4P1N1/r2P4/b5P1/1BQ2PBP/3RR1K1 b - -

Engine: Toga Checkov Beta 4j (128 MB)
by Thomas Gaksch and Fabien Letouzey


6/27 0:00 -1.00 1...Bxg2 2.Qxh7+ Kf8 3.Bxa3+ Rxa3
4.Qh5 g6 5.Qh7 Rf3 6.Kxg2 Rd3+ 7.Kg1 (82.117)

6/27 0:00 -0.81 1...Bxb2 2.Bxb7 Qxb7 3.Qxa4 h6 4.Ne4 h5
5.Re3 (108.898)

7/35 0:00 -1.77-- 1...Bxb2 2.Bxb7 Qxb7 3.Qxa4 h6 4.Ne4 Qa8
5.Qc2 Qa2 6.Qxc7 (227.735)

8/41 0:02 -14.06-- 1...Bxb2 2.Qxh7+ Kf8 3.Qh5 g6 4.Qh7 Nxe5
5.dxe5 Rf4 6.gxf4 Rd7 7.Qh8+ Ke7
8.Rxd7+ Kxd7 9.Rd1+ Bd5 10.Qxa8 (1.320.727) 640

9/59 0:08 -M10-- 1...Bxb2 2.Qxh7+ Kf8 3.Qh5 Nxe5
4.dxe5 Rf4 5.gxf4 Kg8 6.Qxf7+ Kh8
7.Qh5+ Kg8 8.Qh7+ Kf8 9.Qh8+ Ke7
10.Qxg7+ Ke8 11.Qf7+ (5.147.130) 602

9/59 0:09 -13.52++ 1...Bxg2 2.Bxa3 Rxa3 3.Ra1 (5.486.736) 602

10/59 0:13 -5.89++ 1...Bxg2 2.Qxh7+ Kf8 3.Qh5 Qf3
4.Nxf3 Bxf3 5.Qxf3 Bxb2 6.Qb3 Rxd4
7.Qxb2 Rxd1 8.Rxd1 Kg8 9.h4 (8.671.348) 622

11/59 0:17 -6.17 1...Bxg2 2.Qxh7+ Kf8 3.Qh5 Qf3
4.Nxf3 Bxf3 5.Qxf3 Bxb2 6.Qb3 Rxd4
7.Qxb2 Rc4 8.Kg2 Kg8 9.Qb5 Rc5 (10.962.701) 630

11/59 0:17 -5.49++ 1...g6 2.Bxb7 Qxb7 3.Nxe6 Bxb2
4.Nxd8 Qa8 5.Qxb2 Qxd8 6.e6 (11.148.853) 630

12/59 0:22 -4.16++ 1...g6 2.Bxb7 Qxb7 3.Qxa4 Be7 4.Ne4 b5
5.Qb3 c6 6.Ra1 Nb6 7.Ra2 Nd5 8.Rea1 Bb4
9.Ra7 (13.994.355) 622

13/59 0:29 -4.41 1...g6 2.Bxb7 Qxb7 3.Qxa4 Be7 4.Ne4 b5
5.Qb3 c6 6.Ra1 Nb6 7.Ra2 Nd5 8.Rea1 Rd7
9.Bc3 Rd8 (18.167.543) 614

13/59 0:33 -3.76++ 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re2 h6 5.Nxf7 Kxf7 6.d5 exd5
7.Qf4+ Ke6 8.Qg4+ Ke7 9.Qxg7+ Ke6
10.Qxh6+ Kf5 (20.680.489) 617

14/59 1:14 -1.48++ 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ra3 7.Rb3 Rxb3
8.Qxb3 Qa8 9.Kg2 Qa3 10.Qc4 Qe7 (44.833.789) 603

15/59 1:33 -1.38 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Qc4 Ng6
8.d5 exd5 9.Rxd5 c5 10.Qe4 Nf8
11.Rd6 Qxe4 12.Rxe4 b5 13.Kg2 (56.781.339) 605

16/59 2:39 -1.51 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Kg2 Ra7
8.Qb3 Nh7 9.d5 Ng5 10.Nxg5 hxg5
11.Ree1 Qa8 (96.949.742) 609

17/59 4:43 -1.38 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Kg2 Ra7
8.Qc4 Qa8 9.h4 Ra2 10.h5 Qa4 11.Qxa4 Rxa4
12.Rc1 (171.988.728) 606

18/59 10:17 -1.66 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 Qa8 5.Qxa8 Rxa8 6.Nf3 Ra5
7.Rb1 Ba1 8.Rd3 Ra7 9.Rdd1 Bc3
10.Rb3 Ba5 11.Kg2 c5 12.dxc5 bxc5
13.Rc1 Bb4 (376.772.082) 610

19/63 19:39 -1.75 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 Qa8 5.Qxa8 Rxa8 6.Nf3 Ra5
7.Rb1 Ba1 8.Rd3 Ra7 9.Kg2 h6 10.h3 Ra2
11.Rdd1 Bc3 12.Rbc1 Ra3 13.Rd3 Bb2
14.Rxa3 Bxc1 (723.944.217) 613

Almost, almost Hiarcs 11.1 was beaten here in Arasan 7

Eelco

[Rolf]

Sadly, this position Toga Checkov Beta 4j now can't solve anymore in twenty seconds going over Arasan 7 testsuite. I hope that this phenomenon only occurs if the position seems already lost. For Mr. Tueschen; the new reductions are now called Fischer History Reductions, not to be confused with Fail High Reductions from Ed I Believe, and Random Transpositions is another phenomenon displayed mainly by Fritz, as Christopher is trying to explain in another post. You almost got it right though. Just working on getting the randomness under control.

[d]q2r2k1/1bpn1ppp/1p2p3/4P1N1/r2P4/b5P1/1BQ2PBP/3RR1K1 b - -

Engine: Toga Checkov Beta 4j (128 MB)
by Thomas Gaksch and Fabien Letouzey


6/27 0:00 -1.00 1...Bxg2 2.Qxh7+ Kf8 3.Bxa3+ Rxa3
4.Qh5 g6 5.Qh7 Rf3 6.Kxg2 Rd3+ 7.Kg1 (82.117)

6/27 0:00 -0.81 1...Bxb2 2.Bxb7 Qxb7 3.Qxa4 h6 4.Ne4 h5
5.Re3 (108.898)

7/35 0:00 -1.77-- 1...Bxb2 2.Bxb7 Qxb7 3.Qxa4 h6 4.Ne4 Qa8
5.Qc2 Qa2 6.Qxc7 (227.735)

8/41 0:02 -14.06-- 1...Bxb2 2.Qxh7+ Kf8 3.Qh5 g6 4.Qh7 Nxe5
5.dxe5 Rf4 6.gxf4 Rd7 7.Qh8+ Ke7
8.Rxd7+ Kxd7 9.Rd1+ Bd5 10.Qxa8 (1.320.727) 640

9/59 0:08 -M10-- 1...Bxb2 2.Qxh7+ Kf8 3.Qh5 Nxe5
4.dxe5 Rf4 5.gxf4 Kg8 6.Qxf7+ Kh8
7.Qh5+ Kg8 8.Qh7+ Kf8 9.Qh8+ Ke7
10.Qxg7+ Ke8 11.Qf7+ (5.147.130) 602

9/59 0:09 -13.52++ 1...Bxg2 2.Bxa3 Rxa3 3.Ra1 (5.486.736) 602

10/59 0:13 -5.89++ 1...Bxg2 2.Qxh7+ Kf8 3.Qh5 Qf3
4.Nxf3 Bxf3 5.Qxf3 Bxb2 6.Qb3 Rxd4
7.Qxb2 Rxd1 8.Rxd1 Kg8 9.h4 (8.671.348) 622

11/59 0:17 -6.17 1...Bxg2 2.Qxh7+ Kf8 3.Qh5 Qf3
4.Nxf3 Bxf3 5.Qxf3 Bxb2 6.Qb3 Rxd4
7.Qxb2 Rc4 8.Kg2 Kg8 9.Qb5 Rc5 (10.962.701) 630

11/59 0:17 -5.49++ 1...g6 2.Bxb7 Qxb7 3.Nxe6 Bxb2
4.Nxd8 Qa8 5.Qxb2 Qxd8 6.e6 (11.148.853) 630

12/59 0:22 -4.16++ 1...g6 2.Bxb7 Qxb7 3.Qxa4 Be7 4.Ne4 b5
5.Qb3 c6 6.Ra1 Nb6 7.Ra2 Nd5 8.Rea1 Bb4
9.Ra7 (13.994.355) 622

13/59 0:29 -4.41 1...g6 2.Bxb7 Qxb7 3.Qxa4 Be7 4.Ne4 b5
5.Qb3 c6 6.Ra1 Nb6 7.Ra2 Nd5 8.Rea1 Rd7
9.Bc3 Rd8 (18.167.543) 614

13/59 0:33 -3.76++ 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re2 h6 5.Nxf7 Kxf7 6.d5 exd5
7.Qf4+ Ke6 8.Qg4+ Ke7 9.Qxg7+ Ke6
10.Qxh6+ Kf5 (20.680.489) 617

14/59 1:14 -1.48++ 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ra3 7.Rb3 Rxb3
8.Qxb3 Qa8 9.Kg2 Qa3 10.Qc4 Qe7 (44.833.789) 603

15/59 1:33 -1.38 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Qc4 Ng6
8.d5 exd5 9.Rxd5 c5 10.Qe4 Nf8
11.Rd6 Qxe4 12.Rxe4 b5 13.Kg2 (56.781.339) 605

16/59 2:39 -1.51 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Kg2 Ra7
8.Qb3 Nh7 9.d5 Ng5 10.Nxg5 hxg5
11.Ree1 Qa8 (96.949.742) 609

17/59 4:43 -1.38 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 h6 5.Nf3 Ra8 6.Qc2 Ba1 7.Kg2 Ra7
8.Qc4 Qa8 9.h4 Ra2 10.h5 Qa4 11.Qxa4 Rxa4
12.Rc1 (171.988.728) 606

18/59 10:17 -1.66 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 Qa8 5.Qxa8 Rxa8 6.Nf3 Ra5
7.Rb1 Ba1 8.Rd3 Ra7 9.Rdd1 Bc3
10.Rb3 Ba5 11.Kg2 c5 12.dxc5 bxc5
13.Rc1 Bb4 (376.772.082) 610

19/63 19:39 -1.75 1...Nf8 2.Bxb7 Qxb7 3.Qxa4 Bxb2
4.Re3 Qa8 5.Qxa8 Rxa8 6.Nf3 Ra5
7.Rb1 Ba1 8.Rd3 Ra7 9.Kg2 h6 10.h3 Ra2
11.Rdd1 Bc3 12.Rbc1 Ra3 13.Rd3 Bb2
14.Rxa3 Bxc1 (723.944.217) 613

Almost, almost Hiarcs 11.1 was beaten here in Arasan 7

Eelco

Err, whaT do you mean with working? Do you plan to create your own engine or what is this all about? Not that I would want to disturb any kind of hobby, but why not putting that energy in something of your own? I'm just trying to understand. Perhaps that part of the topic (motivation) was already discussed somewhere else?

[Eelco de Groot]

My deleted reply in the other thread, I hope this post can remain, sorry to the persons whose fair posts in the other thread got removed that I called for moderation, I explained my objections there.

Rolf wrote:

Eelco de Groot wrote:

[...irrelevant post here...]

Because you cant answer the apparent questions? Or because you can count on mods who are supporting such activities as clone testing. Is this really so clearly and finally discussed that you couldnt stop it at all and dont want to hear any questions? I swear that I came to your example with the Toga output in the Alekhine position and just published my thoughts. I'm not acting for other people, and a short reply would suffice about your reasons for all this work. Thanks.

Well Rolf maybe an analogy would be jazz music, where people take a well known tune that they like very much, a 'standard' tune if you like, and give it their own twist, their own flavour. It is old and yet it is new also, it is repetition and yet never the same, in the copying there is praise and the recognition that there are some things that can not be improved, because they are universal. I think that Fabien Letouzy's work in Fruit is one of those universal works of art, like Michelangelo's statue of David maybe, that you just can not get around anymore.

It is like asking a chess programmer to write a top chess program only he can't use mini-max. It is that universal. So it will keep coming back no matter how every programmer rewrites it and in this case it makes 0% difference what code they make of it because the ideas behind it are so universally applicable. It's like when you are a sculptor and try to make a human likeness in a statue, try getting around studying and analyzing human proportions first. You can't. That is my deepest reason for studying Fruit and I think I'm not alone in thinking this.

Of course it is nice if I could build such a program totally from the ground up, but I can't do that I readily admit it, you have to be an accomplished programmer first, like Fabien and even he had the experience of years writing Othello programs first before he started Fruit. Not that I won't want to learn that, but I'm just starting with the portions in a chess program that I know best. There is no need to participate in a tournament as if I had created this program all by myself, I would not dream of it. It is just another category of chessprograms, where everyone can add their own little bit just as they like as long as they keep to the GPL license, and it has absolutely zero to do with clones, I wish you had that figured out by now Rolf, but I am sure that you will eventually. Certain other persons will not I am afraid.

Thanks, [...irrelevant here...]

Eelco

[Rolf]

My deleted reply in the other thread, I hope this post can remain, sorry to the persons whose fair posts in the other thread got removed that I called for moderation, I explained my objections there.

My answer to you there was as follows:


You are on the right track. I expected much less from people who come with their "own" creation. Understzood that they know how the ideas work in such super marvels like Fruit, I would hope that many used the ideas now. Because then it would depend on the novelties that they would add. That is progress in my eyes. However a slightly modified fruit in Toga with the same ideas and now after four years a modification more - where is the beef of the new ideas? What has this to do with programming if only certain parameters are tested, what normally also testing users could do? Of course the same question would exist for the many CM personalities. But the CM always remains the same. However here with Toga we do not speak of a personality aka Fruit but people beg that this IMO clone because it's Fruit, should participate in the CCT. To take your own example Michelangelo, a somehow tuned Fruit, now called Toga, doesnt add something to the art of David. However in Jazz it's different. There the interpretation by always different musicians is the message. Not the original tune. But a tournament is something something else again. There we have different programs all playing their (hopefully) own music. Many of them who already have digested the *ideas* of Fruit too. But Toga is NOT about a plus of new ideas to the old Fruit. (Therefore I asked for your work. You could well have planned to come out with your own program on the basis of all the Fruit ideas PLUS some of your own. But alas, I'm really sorry and I dont want to nag you, but if you dont have NEW ideas then you just shouldnt waste so much time with the Toga tuning. And nevertheless always talking of doing some code here and there or of programming stuff. Why not taking all your ideas to some new project of your own - this was my main question.)

===========================


And after it was raised that Toga would have 150 more *strength* than Fruit 4 years ago, I wrote the following note:

I would like to object against such a short cut. Just take a theoretical example. Then you optimize the code for over 4 years and you test the different new versions against original well know programs, but these programs can never be tuned against this say Fruit clone of the 24th generation, then it might well be that you get such a boost, but that cant be called *strength*. It's a statistical test artefact. I dont know if my objection could be made clear. We can well discuss it. A test takes for granted that all the members have the same chance to develop but if you come from the outside and create extra conditions for a singular participant then you cant interprete the results as if it has participated under the same conditions as all others.

Another no no could be seen in a sudden tornament participation of such extra tunes clone. What xould be the motivation. I have a choice. I get it as a form of Rybka prevention measure to somehow weaken the favorite program... I am certain that this is the motif for Toga cluster clones and such some. That shouldnt be allowed. Neither the so called testing tolerated...


[I was told that I should repeat these messages because everything had been deleted in the other thread but not because it was directed against my opinions. Thanks.]

[slobo]

My deleted reply in the other thread, I hope this post can remain, sorry to the persons whose fair posts in the other thread got removed that I called for moderation, I explained my objections there.

My answer to you there was as follows:


You are on the right track. I expected much less from people who come with their "own" creation. Understzood that they know how the ideas work in such super marvels like Fruit, I would hope that many used the ideas now. Because then it would depend on the novelties that they would add. That is progress in my eyes. However a slightly modified fruit in Toga with the same ideas and now after four years a modification more - where is the beef of the new ideas? What has this to do with programming if only certain parameters are tested, what normally also testing users could do? Of course the same question would exist for the many CM personalities. But the CM always remains the same. However here with Toga we do not speak of a personality aka Fruit but people beg that this IMO clone because it's Fruit, should participate in the CCT. To take your own example Michelangelo, a somehow tuned Fruit, now called Toga, doesnt add something to the art of David. However in Jazz it's different. There the interpretation by always different musicians is the message. Not the original tune. But a tournament is something something else again. There we have different programs all playing their (hopefully) own music. Many of them who already have digested the *ideas* of Fruit too. But Toga is NOT about a plus of new ideas to the old Fruit. (Therefore I asked for your work. You could well have planned to come out with your own program on the basis of all the Fruit ideas PLUS some of your own. But alas, I'm really sorry and I dont want to nag you, but if you dont have NEW ideas then you just shouldnt waste so much time with the Toga tuning. And nevertheless always talking of doing some code here and there or of programming stuff. Why not taking all your ideas to some new project of your own - this was my main question.)

===========================


And after it was raised that Toga would have 150 more *strength* than Fruit 4 years ago, I wrote the following note:

I would like to object against such a short cut. Just take a theoretical example. Then you optimize the code for over 4 years and you test the different new versions against original well know programs, but these programs can never be tuned against this say Fruit clone of the 24th generation, then it might well be that you get such a boost, but that cant be called *strength*. It's a statistical test artefact. I dont know if my objection could be made clear. We can well discuss it. A test takes for granted that all the members have the same chance to develop but if you come from the outside and create extra conditions for a singular participant then you cant interprete the results as if it has participated under the same conditions as all others.

Another no no could be seen in a sudden tornament participation of such extra tunes clone. What xould be the motivation. I have a choice. I get it as a form of Rybka prevention measure to somehow weaken the favorite program... I am certain that this is the motif for Toga cluster clones and such some. That shouldnt be allowed. Neither the so called testing tolerated...


[I was told that I should repeat these messages because everything had been deleted in the other thread but not because it was directed against my opinions. Thanks.]

So, Rolf, you think they must ban Toga cluster and prevent it to beat Rybka?

[Eelco de Groot]

Sadly, this position Toga Checkov Beta 4j now can't solve anymore in twenty seconds going over Arasan 7 testsuite.

Almost, almost Hiarcs 11.1 was beaten here in Arasan 7

Eelco

Really a pity that the tenth position could not be solved.
I would like to make changes to the evaluation function of Toga so these testresults are not saying very much, because they will change with any changes you make, but I was happy with this result anyway, maybe just one or two positions more should be possible one day...:

Toga Checkov Beta 4j 128 Mb HT 20 sec per pos. Version dated 22-01-08
7:49, with Fischer History Reductions

Code: Select all

Right until now: 180 of 227  ;  26:49m
I/O Fault!

  1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
 ---------------------------------------------------------------------------------
  0   0   0   1  10   3   2   0   3   -   1   0   0   0   0  14   5   1   0   3
  0   -   -   -   -   2   0   0   2   -   2   0  12   3   5   0   4   8   0   1
  1   -   2   0   0   -   0   3   1   0   0   0   -   0   4   4   0   -   -   0
  0   8   0   6   5   0   7   -   2  15  10   0   1   0   2   0   -   3   -   3
  -   3  11   9   0   3   0  15   0   0  14   0   0   7   6   0   -  12   0   1
  0   0   0   0  13   2   0   1   8   0   4   1   0   4   -  11   9   0   -   0
  0   -   0   1   5  19   2   4  10   -   -   1   1   3   1   2   -   0   -   1
  17  0   7   0   6   0   6   -   -   -   -   0   1   0   -   1   0   -   0   -
  6   5  16   -   2   4   1   0   -   0   3   0   -   -   -   1  14   -   3   1
  0   7   -   0   2   0   -   1   -   -   1   0   3   1   0   0   -   0  11   1
  0   -   -   0   0   -  18   6   1  11   0  16   2  19   2   1   -   0  18   -
  3   4   -  11   0   0  14

   1 sec ->  74/227
   2 sec -> 100/227
   3 sec -> 114/227
   4 sec -> 128/227
   5 sec -> 136/227
   6 sec -> 141/227
   7 sec -> 147/227
   8 sec -> 151/227
   9 sec -> 154/227
  10 sec -> 156/227
  11 sec -> 159/227
  12 sec -> 164/227
  13 sec -> 166/227
  14 sec -> 167/227
  15 sec -> 171/227
  20 sec -> 180/227
  n/s: 699.940  
  TotTime: 76:00m    SolTime: 26:49m

For comparison this is the current leader in my list:

Code: Select all

HIARCS 11.1 Default
Goed tot nu toe: 182 van 227  ;  25:40m
I/O fout!

 1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
 ---------------------------------------------------------------------------------  0  2   1   0   -   8   0   0   5   0   1   2   0   0   0  13   -   0   0   1
 5   -   6   0   -   2   0   1  10   8   0   0   -  18   0   1   5   2   0   1
 0   3   0   0   1  12   8   2   9   0   0   0   -   0   7   -   0   4   -   0
 0   6   2   8   -   1   0   -   -   1   0   0  14   0   0   0  10   3   8   -
 12  -   2   4  19   0   0  10   0   1   1   0   0  13  11   2   1   4   0   0
 0   0   0  19   0   0  13   0   -   2   1   0   1  17   -   1   -   2   -   3
 0   -   1   -   0   -   0   -   3   -   3   1   2   1   0   3   -   -   0   2
 0   -   -   5   0   -   -   0   3   -   0   0   1   0   -   1   0   3   1   5
 5   3   -   -   3   2  14  13   5   0   0   5   6   -   6   0   0   -   0   0
10   -   -   7   3   0   5   -  13   -  15   3   0   2   0   -   5   2   4   0 
16   4   8   0   0   0   -  19   0   1   0   1   0   1   0   0   -   0   0   -
 -   4   -   0   0   -   1

   1 sec ->  80/227
   2 sec -> 105/227
   3 sec -> 120/227
   4 sec -> 132/227
   5 sec -> 138/227
   6 sec -> 148/227
   7 sec -> 152/227
   8 sec -> 154/227
   9 sec -> 160/227
  10 sec -> 161/227
  11 sec -> 165/227
  12 sec -> 166/227
  13 sec -> 168/227
  14 sec -> 173/227
  15 sec -> 175/227
  20 sec -> 182/227
  n/s: 295.491  
  TotTime: 74:28m    SolTime: 25:40m

Okay I know, just showing off
Eelco

[M ANSARI]

Can I ask if this is a single core version or the MP version of Toga. If this is the single core version is the MP version available. Thanks