Stockfish code tweak request

[dollyknot]

Hello,

For many years I have used Crafty for analysis, thinking it was the strongest open source engine around, then I discovered Glaurung, which in engine matches with Crafty nearly always won, but I did not like using Glaurung for analysis, because of the different way it reported which side was ahead, Crafty consistently had white ahead with a '+' and black ahead with a '-' and also I never managed to compile Glaurung.

I could never make sense of which side Glaurung calculated to be ahead, so I went back to using Crafty.

Just recently I found out about Stockfish and things started to get very interesting.

I don't know whether anyone has kept up with the story of the French player accused of cheating.

Details here.

http://stevegiddins.blogspot.com/2011/0 ... l?spref=fb

Giddins thinks that the game in question is

[White "Howell, David W L"]
[Black "Feller, Sebastien"]

And the key move 19...Nc2

So I decided to analyse it with Crafty.

1. e4 e6 2. d4 d5 3. Nd2 Nf6 4. e5 Nfd7 5. Bd3 c5 6. c3 Nc6 7. Ne2 cxd4 8. cxd4Qb6 9. Nf3 f6 10. exf6 Nxf6 11. O-O Bd6 12. b3 O-O 13. Bb2 Bd7 14. Nc3 Rac8 15.Na4 Qa5 16. Ne5 Nb4 17. Bb1 Bxa4 18. bxa4 Qc7 19. a3

At this point I left Crafty analysing for hours and hours and it never found 19...Nc2 which was played by Fellar, the game was played out by Howell and Fellar with these moves.

20. Ra2 Bxe5 21. dxe5 Ne4 22. Bc1 Nxf2! 23. Qxc2 Nh3+ 24. gxh3 Qb6+ 25. Kg2 Rxc2+ 26. Bxc2 Rxf1 27.Kxf1 Qc7 28. Rb2 d4 29. Ke2 Qxe5+ 30. Kd1 Qh5+ 31. Kd2 Qg5+ 32. Kd1 Qg1+ 33. Kd2 Qxh2+ 34. Kd3 Qg1 35. Rb1 Qg6+ 36. Kd2 Qg2+ 37. Kd3 Qg6+ 38. Kd2 Qg2+ 39.Kd3 e5 40. Rb2 Qf1+ 41. Kd2 Qf2+ 42. Kd3 Qf3+ 43. Kd2 e4 44. Bb3+ Kf8 45. Ke1 Qh1+ 46. Kd2 Qg2+ 47. Ke1 Qg1+ 48. Kd2 Qf2+ 0-1

Instead Crafty 19.19 opted for 19...Nc6 which with Crafty playing both sides and 30 minutes on the clock led to

20. Qe2 Kh8 21. Nf3 Nh5 22. Ng5 Bxh2+ 23. Kh1 g6 24. Nxe6 Qe7 25. Kxh2 Rce8 26. Bd3 Qxe6 27. Qxe6 Rxe6 28. g3 Nf6 29. Bc3 Ne4 30. Rae1 Rf3 31. Bxe4 Rxe4 32. Rxe4 dxe4 33. Rc1 Kg8 34. d5 Ne7 35. d6 Nc6 36. Kg2 Rd3 37. Bb4 a5 38. Bc5 Rd5 39. g4 Kf7 40. Be3 Rxd6 41. Rb1 Rd7 42. Rb5 Kf6 43. g5+ Ke6 44. Kh2 Rd3 45. Bc1 Rc3 46. Bb2 Rf3 47. Kg2 Rf5 48. Bc3 Rd5 49. Kf1 Rf5 50. Ke2 Rxb5 51. axb5 Ne5 52. Bxa5 Nc4 53. Bb4 Kd5 54. a4 b6 55. Bd2 Kd4 56. Bb4 Kd5 57. Bd2 Kc5 58. Bc3 Kd5 {Draw agreed} 1/2-1/2

Now I do not know whether the French lad was cheating or not, that is not why I am writing this, but because I find it very interesting that Stockfish found 19...Nc2 all most immediately, whereas Crafty never found it after hours of calculating.

So I'm going to use Stockfish for analysis, from now on.

I managed to get Stockfish to compile on my machine, which gave me a great deal of pleasure and also the ability to tweak the code.

What I would like Stockfish to do is, to report which side is ahead the same way as Crafty does ie Black ahead with a '-' and White ahead with a '+'.

Any help with this project gratefully received, like where specifically in the source code does it need to be changed and to what?

Regards,

Peter

http://dollyknot.com

[bob]

One major question, why crafty v 19.19? Current is 23.4. 19.19 is several years old and things have changed a lot on the computer chess scene since then...

Hello,

For many years I have used Crafty for analysis, thinking it was the strongest open source engine around, then I discovered Glaurung, which in engine matches with Crafty nearly always won, but I did not like using Glaurung for analysis, because of the different way it reported which side was ahead, Crafty consistently had white ahead with a '+' and black ahead with a '-' and also I never managed to compile Glaurung.

I could never make sense of which side Glaurung calculated to be ahead, so I went back to using Crafty.

Just recently I found out about Stockfish and things started to get very interesting.

I don't know whether anyone has kept up with the story of the French player accused of cheating.

Details here.

http://stevegiddins.blogspot.com/2011/0 ... l?spref=fb

Giddins thinks that the game in question is

[White "Howell, David W L"]
[Black "Feller, Sebastien"]

And the key move 19...Nc2

So I decided to analyse it with Crafty.

1. e4 e6 2. d4 d5 3. Nd2 Nf6 4. e5 Nfd7 5. Bd3 c5 6. c3 Nc6 7. Ne2 cxd4 8. cxd4Qb6 9. Nf3 f6 10. exf6 Nxf6 11. O-O Bd6 12. b3 O-O 13. Bb2 Bd7 14. Nc3 Rac8 15.Na4 Qa5 16. Ne5 Nb4 17. Bb1 Bxa4 18. bxa4 Qc7 19. a3

At this point I left Crafty analysing for hours and hours and it never found 19...Nc2 which was played by Fellar, the game was played out by Howell and Fellar with these moves.

20. Ra2 Bxe5 21. dxe5 Ne4 22. Bc1 Nxf2! 23. Qxc2 Nh3+ 24. gxh3 Qb6+ 25. Kg2 Rxc2+ 26. Bxc2 Rxf1 27.Kxf1 Qc7 28. Rb2 d4 29. Ke2 Qxe5+ 30. Kd1 Qh5+ 31. Kd2 Qg5+ 32. Kd1 Qg1+ 33. Kd2 Qxh2+ 34. Kd3 Qg1 35. Rb1 Qg6+ 36. Kd2 Qg2+ 37. Kd3 Qg6+ 38. Kd2 Qg2+ 39.Kd3 e5 40. Rb2 Qf1+ 41. Kd2 Qf2+ 42. Kd3 Qf3+ 43. Kd2 e4 44. Bb3+ Kf8 45. Ke1 Qh1+ 46. Kd2 Qg2+ 47. Ke1 Qg1+ 48. Kd2 Qf2+ 0-1

Instead Crafty 19.19 opted for 19...Nc6 which with Crafty playing both sides and 30 minutes on the clock led to

20. Qe2 Kh8 21. Nf3 Nh5 22. Ng5 Bxh2+ 23. Kh1 g6 24. Nxe6 Qe7 25. Kxh2 Rce8 26. Bd3 Qxe6 27. Qxe6 Rxe6 28. g3 Nf6 29. Bc3 Ne4 30. Rae1 Rf3 31. Bxe4 Rxe4 32. Rxe4 dxe4 33. Rc1 Kg8 34. d5 Ne7 35. d6 Nc6 36. Kg2 Rd3 37. Bb4 a5 38. Bc5 Rd5 39. g4 Kf7 40. Be3 Rxd6 41. Rb1 Rd7 42. Rb5 Kf6 43. g5+ Ke6 44. Kh2 Rd3 45. Bc1 Rc3 46. Bb2 Rf3 47. Kg2 Rf5 48. Bc3 Rd5 49. Kf1 Rf5 50. Ke2 Rxb5 51. axb5 Ne5 52. Bxa5 Nc4 53. Bb4 Kd5 54. a4 b6 55. Bd2 Kd4 56. Bb4 Kd5 57. Bd2 Kc5 58. Bc3 Kd5 {Draw agreed} 1/2-1/2

Now I do not know whether the French lad was cheating or not, that is not why I am writing this, but because I find it very interesting that Stockfish found 19...Nc2 all most immediately, whereas Crafty never found it after hours of calculating.

So I'm going to use Stockfish for analysis, from now on.

I managed to get Stockfish to compile on my machine, which gave me a great deal of pleasure and also the ability to tweak the code.

What I would like Stockfish to do is, to report which side is ahead the same way as Crafty does ie Black ahead with a '-' and White ahead with a '+'.

Any help with this project gratefully received, like where specifically in the source code does it need to be changed and to what?

Regards,

Peter

http://dollyknot.com

[hgm]

I think this is a GUI matter rather than an engine matter. UCI protocol prescribes engines should report score from your own Point Of View. It is the GUIs task to decide how it is presented to the engine.

You can simply emulate Crafty's (non-standard) way of reporting scores by telling the GUI the engine's score is 'absolute' (i.e. from white POV) to make an engine whose score was not absolute (like Stockfish) report like its score was absolute (i.e. like Crafty).

[dollyknot]

I think this is a GUI matter rather than an engine matter. UCI protocol prescribes engines should report score from your own Point Of View. It is the GUIs task to decide how it is presented to the engine.

You can simply emulate Crafty's (non-standard) way of reporting scores by telling the GUI the engine's score is 'absolute' (i.e. from white POV) to make an engine whose score was not absolute (like Stockfish) report like its score was absolute (i.e. like Crafty).

Interesting, I'm using xboard as my GUI which BTW I've managed to compile the new version that you have kindly written, after battling my way out of dependency hell, so I will try what you suggest and report back.

[hgm]

In XBoard this option can only be set from the command line, as most WinBoard menu dialogs have not been ported yet. The option you need is

-firstScoreAbs true

It is a volatile option,so you would have to specify it every time. Unless you add it in /etc/xboard.conf at the end (behind -settingsFile),for 4.5.0.

[dollyknot]

One major question, why crafty v 19.19? Current is 23.4. 19.19 is several years old and things have changed a lot on the computer chess scene since then...

Hello Bob,

It has been a while 19.19 is the binary version supplied with Mandriva, the distro I use.

Although I have managed to compile Crafty in the past, advancing senility seems to have defeated me, in terms of compiling current versions. If I can type ./configure and install missing libraries I can still get there sometimes. But looking at the bewildering number of options in the crafty make file, made me give up. If you like I will try again, but you will have to hold my hand.

Given that I can compile a working binary from the latest version, I will contact Mandriva and ask them to include it in their repository.

[dollyknot]

In XBoard this option can only be set from the command line, as most WinBoard menu dialogs have not been ported yet. The option you need is

-firstScoreAbs true

It is a volatile option,so you would have to specify it every time. Unless you add it in /etc/xboard.conf at the end (behind -settingsFile),for 4.5.0.

Yeah that works fine with 4.5.0 and as I am firing up Stockfish from a bash script, not a problem, I just inserted it into the bash script and now I'm clear who is in front!

I like watching live matches and it will be like having a supercharged super GM at my elbow that speaks English.