What are the improvements of Fritz 11?

[mercury]

I noticed that Fritz 11 is much weaker than Fritz 10 in the endgames so I wonder what are Fritz 11's improvements?
http://www.talkchess.com/forum/viewtopic.php?t=19849
In which part of the game do you find Fritz 11 stronger than Fritz 10?

[Uri Blass]

I noticed that Fritz 11 is much weaker than Fritz 10 in the endgames so I wonder what are Fritz 11's improvements?
http://www.talkchess.com/forum/viewtopic.php?t=19849
In which part of the game do you find Fritz 11 stronger than Fritz 10?

I do not have Fritz11 but your link does not prove that Fritz11 is weaker than Fritz10 in endgames.

Endgames are not about solving test positions when there is some sacrifice but mainly about positional moves.

The only way to prove that program A is better than program B in endgame is to let them play games from practical endgame positions and I see no games from endgame positions in the link that you give.

Uri

[mercury]

Of course in my link there are only 2 positions but now I finished to test 120 endgame positions and I find Fritz 11 weaker than Fritz 10.
http://pagesperso-orange.fr/echecs88/cl ... t.html#gen
At the moment I didn't add Fritz 11 but it will be at the 9th position with 66,1 points.

[Uri Blass]

Of course in my link there are only 2 positions but now I finished to test 120 endgame positions and I find Fritz 11 weaker than Fritz 10.
http://pagesperso-orange.fr/echecs88/cl ... t.html#gen
At the moment I didn't add Fritz 11 but it will be at the 9th position with 66,1 points.

Test positions are usually about cases when one move is significantly better than the rest of the moves or about a sacrifice.
They are about tactics.

In endgames you need to take correct positional decisions to slightly improve your position and test suites are not going to tell you which program is better in these decisions.

The only way to find which program is better in endgame is to let programs play endgames.

Test suites are simply not going to give you information.
I do not know if Fritz11 is better or worse than Fritz10 in endgame(I do not have these programs) but if people are interested to test it they can use specific endgames and I remember that specific endgame position with games were used

Here is an old link and again epd of positions and I wonder if Fritz10 performs better than Fritz11 from these positions

http://www.stmintz.com/ccc/index.php?id=473739

8/1ppr1kp1/p1p4p/8/8/5P2/PPP1RKPP/8 w - - 0 1
8/1pk3pp/p7/3p1p2/3K4/6P1/PP2PP1P/8 b - - 0 1
8/4k3/r4pp1/pR6/P4KP1/5P2/8/8 w - - 0 1
8/p1p1k3/1p1p2p1/3Pq2p/2P1P3/1P3QPK/P7/8 w - - 0 1
8/p3ppk1/1p4pp/3nN3/6PP/1P3P2/P3PK2/8 w - - 0 1
7k/pp4pp/2n5/8/8/P7/1P4PP/2K1B3 b - - 0 1
8/8/p4Bp1/1pPb2P1/1P2kp2/P7/5K2/8 w - - 0 1
8/4k3/p1B4p/2K5/1P4bP/8/8/8 b - - 0 1
3R4/1p6/2b5/2P1k2p/p3p2P/P6r/1P2KB2/8 b - - 0 1
8/5b2/1k2p1p1/2NpP1P1/1K1P4/8/8/8 w - - 0 1
8/5p1k/r5pp/P7/3R3P/6P1/5PK1/8 w - - 0 1
4b3/5k2/4p1p1/3pP2p/2pP1P1P/2P5/6N1/2K5 w - - 0 1
8/1p1b2k1/p1p1p1p1/2P1P2p/1P3P1P/P2B4/5K2/8 w - - 0 1
6k1/2p3np/1p1p2p1/3P4/1PPK1R2/6PB/7P/4r3 w - - 0 1
1n6/4k2p/p3ppp1/1pPp4/3P1PP1/3NP3/P3K2P/8 w - - 0 1
3r4/7p/Rp4k1/5p2/4p3/2P5/PP3P1P/5K2 b - - 0 1
8/4k3/1p2b1pp/p1p1p3/P1P1P3/1P3P1P/5KB1/8 b - - 0 37
2r5/1p1k1p2/p2p2p1/P1PPnp1p/1P1K1N1P/5P2/4R1P1/8 w - - 0 1
8/6b1/4k2p/p1p1p1p1/2P1K1P1/PP3N1P/8/8 w - - 0 1
2r3k1/4pp1p/3p2p1/p2R4/1r6/1P4P1/P3PP1P/R5K1 b - - 0 1


Uri

[Oscar L]

Hi Ludovic

Thanks for your endgame rating list!

About Fritz 11, It could be interesting, if you have time, to test Fritz 11 with the parameter null move unchecked. By default it is checked, probably this is good for middlegame to get more depth in a given time, but it is not good for endgame positions where understanding of zugzwang is important.

But perhaps you only plan to test engines with default settings?

[mercury]

So far I tested engines with default parameters.
I found the definition of null move:
The Null Move is a special search mechanism which reduces the size of the search tree, so that the engine can look deeper. If one side makes two moves in a row ("null move" for opponent) and does not get a big advantage, the engine assumes that the first move is bad. This is usually highly efficient. However in certain endgame positions it can overlook Zugzwang.
Actually I noticed Fritz 11 missed a lot of positions with zugzwang. It could be due to this nullmove parameter. I'm not convinced this setting is an improvement for Fritz. Has somebody examples to show me the intererest of this option?

[Oscar L]

Fritz 11 has gained around 60-100 elo points in the different rating lists when you compare it with Fritz 10. AFAIK Fritz 11 has only been tested with default settings in these lists, so we can not be sure of the impact of null move parameter. I suppose that if the engine author decided that it was good, must be because it allowed to gain elo points in the internal testing before the release, and as you say it is because is looking deeper. Rybka was too far away!


If you noticed that Fritz 11 missed a lot of positions with zugzwang, sure it is because of nullmove parameter. Again, it would be quite interesting that you run your test with the option uncheched to compare the results, but of course it is up to you


In the wikipedia there is a good explanation, I think, about this, but perhaps Uri or other engine developer can comment something more:

http://en.wikipedia.org/wiki/Null-move_heuristic

http://en.wikipedia.org/wiki/Zugzwang

[mercury]

I began the test with nullmove unchecked. With only 10 positions it is already much better
Thank you Oscar for indicating me the null move option

[Jouni]

Hi Ludovic,

Your test is interesting! It can't be totally chance, that Rybka is the best from engines I tested in this test as well as in EET . I tested WITH tablebases of course - Rybka gains more than others from them. And your suite has one bishop promotion, which is too much for Rybka.

Jouni

[Oscar L]

Great

Waiting for the final results