a question about thinking output

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 20 June 2004 05:48:48:

I think to include information not only about the brute force depth but also about the selective depth that is the maximal depth that movei searches in the qsearch and the question is how to do it in a way that will cause no problem not only under winboard but also under arena or fritz.
I see no instructions in
http://www.tim-mann.org/xboard/engine-intf.html for reporting selective depth.
Is the following output ok as analysis of the opening position?
12/30 16 18274 31249512 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
12/33 16 20771 35307194 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
13/33 24 36010 61287362 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
13/36 24 40063 68092606 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
14/36 23 70733 121696318 e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3 c7c5 f1e2 b8c6 d2d3
d8c7 c3b5 c7b8
Uri

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 20 June 2004 17:58:49:
Als Antwort auf:/In reply to: a question about thinking output geschrieben von:/posted by: Uri Blass at 20 June 2004 05:48:48:

I think to include information not only about the brute force depth but also about the selective depth that is the maximal depth that movei searches in the qsearch and the question is how to do it in a way that will cause no problem not only under winboard but also under arena or fritz.
I see no instructions in
http://www.tim-mann.org/xboard/engine-intf.html for reporting selective depth.
Is the following output ok as analysis of the opening position?
12/30 16 18274 31249512 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
12/33 16 20771 35307194 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
13/33 24 36010 61287362 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
13/36 24 40063 68092606 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
14/36 23 70733 121696318 e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3 c7c5 f1e2 b8c6 d2d3
d8c7 c3b5 c7b8
Uri

I find out that this is not correct output(at least winboard does not read it correctly and show strange np in the analysis)
The question is if there is a way to print correct pv with this information under winboard.
Uri

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 20 June 2004 18:20:48:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 20 June 2004 17:58:49:

I think to include information not only about the brute force depth but also about the selective depth that is the maximal depth that movei searches in the qsearch and the question is how to do it in a way that will cause no problem not only under winboard but also under arena or fritz.
I see no instructions in
http://www.tim-mann.org/xboard/engine-intf.html for reporting selective depth.
Is the following output ok as analysis of the opening position?
12/30 16 18274 31249512 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
12/33 16 20771 35307194 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
13/33 24 36010 61287362 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
13/36 24 40063 68092606 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
14/36 23 70733 121696318 e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3 c7c5 f1e2 b8c6 d2d3
d8c7 c3b5 c7b8
Uri

I find out that this is not correct output(at least winboard does not read it correctly and show strange np in the analysis)
The question is if there is a way to print correct pv with this information under winboard.

The actual PV is free format. So you can do (for example)

14 23 70733 121696318 {36} e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3
I suspect, that commercial GUIs would have problems with this, however. But it is not wrong. The problem is that the GUIs want to interprete the PV and have to assume some common format.
Regards,
Dieter

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 20 June 2004 18:42:27:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Dieter Bürßner at 20 June 2004 18:20:48:

I think to include information not only about the brute force depth but also about the selective depth that is the maximal depth that movei searches in the qsearch and the question is how to do it in a way that will cause no problem not only under winboard but also under arena or fritz.
I see no instructions in
http://www.tim-mann.org/xboard/engine-intf.html for reporting selective depth.
Is the following output ok as analysis of the opening position?
12/30 16 18274 31249512 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
12/33 16 20771 35307194 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 d2d4 e5d4 f3d4 f8c5 d4c6
b7c6
13/33 24 36010 61287362 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
13/36 24 40063 68092606 e2e4 e7e5 b1c3 b8c6 g1f3 g8f6 f1b5 f8c5 e1g1 c6d4 f3e5
d4b5 c3b5 f6e4
14/36 23 70733 121696318 e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3 c7c5 f1e2 b8c6 d2d3
d8c7 c3b5 c7b8
Uri

I find out that this is not correct output(at least winboard does not read it correctly and show strange np in the analysis)
The question is if there is a way to print correct pv with this information under winboard.

The actual PV is free format. So you can do (for example)

14 23 70733 121696318 {36} e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3
I suspect, that commercial GUIs would have problems with this, however. But it is not wrong. The problem is that the GUIs want to interprete the PV and have to assume some common format.
Regards,
Dieter

Maybe it is better to put {36} at the end of the pv so hopefully commercial programs will have no problem to interprete the pv.
I do not like to write something only to discover that some interface has problems with it.
Uri

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 20 June 2004 22:03:30:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 20 June 2004 18:42:27:

The actual PV is free format. So you can do (for example)

14 23 70733 121696318 {36} e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3
I suspect, that commercial GUIs would have problems with this, however. But it is not wrong. The problem is that the GUIs want to interprete the PV and have to assume some common format.
Regards,
Dieter

Maybe it is better to put {36} at the end of the pv so hopefully commercial programs will have no problem to interprete the pv.

Perhaps. YOu have to try it out. We cannot guess, how clever the parsers of all the GUIs are. WB has of course no problem, because it just displays the string sent by the engine. A slightly clever parser could seperate the PV into white space seperated tokens, and try every token, if it can be interpreted as a legal move (this would automatically skip {36} as well as other things like move numbers). In my experience, GUIs have problems with this. Perhaps, the chance is better, when the offending token is at the end. Disadvantage: you won't see it in general with WB (because the "PV-display-area" is too small in normal game mode).
BTW. I misused the PV for some "real-on-the-board-tournament-rules". Yace can send a PV like "0 0 0 0 My internal clock is at 13:34, Please adjust if needed" (or something similar), and then don't send a new PV for some seconds. This to be in accordance with some tournament rules, which demand that the operator must be asked for a time adjustment. In this mode, Yace will also ignore time commands from WB (there is no way to adjust the WB-clock) and only use the internal time book keeping.
In the end, I always operated in "console mode", so this feature was really never needed.
Regards,
Dieter

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 20 June 2004 22:04:07:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 20 June 2004 18:42:27:

The actual PV is free format. So you can do (for example)

14 23 70733 121696318 {36} e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3
I suspect, that commercial GUIs would have problems with this, however. But it is not wrong. The problem is that the GUIs want to interprete the PV and have to assume some common format.
Regards,
Dieter

Maybe it is better to put {36} at the end of the pv so hopefully commercial programs will have no problem to interprete the pv.

Perhaps. YOu have to try it out. We cannot guess, how clever the parsers of all the GUIs are. WB has of course no problem, because it just displays the string sent by the engine. A slightly clever parser could seperate the PV into white space seperated tokens, and try every token, if it can be interpreted as a legal move (this would automatically skip {36} as well as other things like move numbers). In my experience, GUIs have problems with this. Perhaps, the chance is better, when the offending token is at the end. Disadvantage: you won't see it in general with WB (because the "PV-display-area" is too small in normal game mode).
BTW. I misused the PV for some "real-on-the-board-tournament-rules". Yace can send a PV like "0 0 0 0 My internal clock is at 13:34, Please adjust if needed" (or something similar), and then don't send a new PV for some seconds. This to be in accordance with some tournament rules, which demand that the operator must be asked for a time adjustment. In this mode, Yace will also ignore time commands from WB (there is no way to adjust the WB-clock) and only use the internal time book keeping.
In the end, I always operated in "console mode", so this feature was really never needed.
Regards,
Dieter

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 20 June 2004 22:22:11:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Dieter Bürßner at 20 June 2004 22:04:07:

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 20 June 2004 23:11:25:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Dieter Bürßner at 20 June 2004 22:04:07:

The actual PV is free format. So you can do (for example)

14 23 70733 121696318 {36} e2e4 e7e6 g1f3 d7d5 e4e5 c8d7 b1c3
I suspect, that commercial GUIs would have problems with this, however. But it is not wrong. The problem is that the GUIs want to interprete the PV and have to assume some common format.
Regards,
Dieter

Maybe it is better to put {36} at the end of the pv so hopefully commercial programs will have no problem to interprete the pv.

Perhaps. YOu have to try it out. We cannot guess, how clever the parsers of all the GUIs are. WB has of course no problem, because it just displays the string sent by the engine. A slightly clever parser could seperate the PV into white space seperated tokens, and try every token, if it can be interpreted as a legal move (this would automatically skip {36} as well as other things like move numbers). In my experience, GUIs have problems with this. Perhaps, the chance is better, when the offending token is at the end. Disadvantage: you won't see it in general with WB (because the "PV-display-area" is too small in normal game mode).
BTW. I misused the PV for some "real-on-the-board-tournament-rules". Yace can send a PV like "0 0 0 0 My internal clock is at 13:34, Please adjust if needed" (or something similar), and then don't send a new PV for some seconds. This to be in accordance with some tournament rules, which demand that the operator must be asked for a time adjustment. In this mode, Yace will also ignore time commands from WB (there is no way to adjust the WB-clock) and only use the internal time book keeping.
In the end, I always operated in "console mode", so this feature was really never needed.
Regards,
Dieter

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.
Uri

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 20 June 2004 23:15:18:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 20 June 2004 23:11:25:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 21 June 2004 00:27:52:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Dieter Bürßner at 20 June 2004 23:15:18:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

Yes
I may decide to play with arena interface.
With console mode I do not see the pieces on the board and it increase the chance that I will not notice some typing error.
Uri

[Peter Berger]

Geschrieben von:/Posted by: Peter Berger at 21 June 2004 00:55:35:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 21 June 2004 00:27:52:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

Yes
I may decide to play with arena interface.
With console mode I do not see the pieces on the board and it increase the chance that I will not notice some typing error.
Uri

One question I was wondering about:
The rules are not really clear to wether it is allowed to setup another time control than the official one. I have no idea if this has ever been an issue.
Peter

[Peter Berger]

Geschrieben von:/Posted by: Peter Berger at 21 June 2004 01:01:28:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Peter Berger at 21 June 2004 00:55:35:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

Yes
I may decide to play with arena interface.
With console mode I do not see the pieces on the board and it increase the chance that I will not notice some typing error.
Uri

One question I was wondering about:
The rules are not really clear to wether it is allowed to setup another time control than the official one. I have no idea if this has ever been an issue.
Peter

I should explain why I think this is a real question. Guess the program of two friends are playing each other. They decide to make the internal time control game/5 to be able to have a drink together later. Or _one_ of them does for that matter ..
I hope there are no rules concerning this as I came to similar conclusions regarding easiest way to deal with operation time .
Peter

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 21 June 2004 01:26:17:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Peter Berger at 21 June 2004 01:01:28:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

Yes
I may decide to play with arena interface.
With console mode I do not see the pieces on the board and it increase the chance that I will not notice some typing error.
Uri

One question I was wondering about:
The rules are not really clear to wether it is allowed to setup another time control than the official one. I have no idea if this has ever been an issue.
Peter

I should explain why I think this is a real question. Guess the program of two friends are playing each other. They decide to make the internal time control game/5 to be able to have a drink together later. Or _one_ of them does for that matter ..
I hope there are no rules concerning this as I came to similar conclusions regarding easiest way to deal with operation time .
Peter

I think that if the time is the same for all the games there should be no problem.
After all if people want to play blitz they can tell their program to use
exactly 2 seconds for every move regardless of the time control..
It is also illogical to tell the program exactly the remaining time because suppose that the program has 10 seconds for the 60 move and use 9.9 seconds for it.
The operator may lose on time so to prevent it he may in the beginning tell the program that it has less time.
I think to use at least 20 minutes as operator time because I have examples when games between computer could reach near 300 moves so even assuming 20 minutes of operator time means taking a small risk of losing on time and maybe it is even better if I use 30 minutes to be careful so I can use 2 hours for all the game under winboard.
Uri

[Peter Berger]

Geschrieben von:/Posted by: Peter Berger at 21 June 2004 01:52:08:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Uri Blass at 21 June 2004 01:26:17:

I guess that in WCCC I will simply use 20 minutes less time than the time that the program have(less 15 minutes in the first 60 moves and less 5 minutes in the rest of the game) and not adjust the clock during the game.
The alternative may lead to bugs and I do not like the idea to play in console mode when I do not like to spend time on testing it.

If you don't want to play in console - which interface will you use? Did you recognize, that it is impossible to give the time control of WCCC to Winboard?
Regards,
Dieter

Yes
I may decide to play with arena interface.
With console mode I do not see the pieces on the board and it increase the chance that I will not notice some typing error.
Uri

One question I was wondering about:
The rules are not really clear to wether it is allowed to setup another time control than the official one. I have no idea if this has ever been an issue.
Peter

I should explain why I think this is a real question. Guess the program of two friends are playing each other. They decide to make the internal time control game/5 to be able to have a drink together later. Or _one_ of them does for that matter ..
I hope there are no rules concerning this as I came to similar conclusions regarding easiest way to deal with operation time .
Peter

I think that if the time is the same for all the games there should be no problem.
After all if people want to play blitz they can tell their program to use
exactly 2 seconds for every move regardless of the time control..
It is also illogical to tell the program exactly the remaining time because suppose that the program has 10 seconds for the 60 move and use 9.9 seconds for it.
The operator may lose on time so to prevent it he may in the beginning tell the program that it has less time.
I think to use at least 20 minutes as operator time because I have examples when games between computer could reach near 300 moves so even assuming 20 minutes of operator time means taking a small risk of losing on time and maybe it is even better if I use 30 minutes to be careful so I can use 2 hours for all the game under winboard.
Uri

15/5 is very conservative IMHO, especially when you use a graphical GUI - as clocks are only turned on when one opponent is out of book in the main event. But I see your point: 30 minutes instead of 20 minutes certainly doesn't make a major difference in performance, but is even safer.
But related questions have actually been a controversial topic in the past. Recently I looked up some old discussions wondering about allowed answers if the engine asks for time on clock. Can you forward an operator reserve to later answers ? I think in this case the answer is probably no.
There is no clear logical answer to all these questions. Of course you can always play game/5 internally whatever time control you set up in the interface but setting up a different time control is still something else. Fortunately these questions used to be _much_ more controversial in the past, and I don't think people care too much these days.
Peter

[Thomas Mayer]

Geschrieben von:/Posted by: Thomas Mayer at 21 June 2004 12:01:58:
Als Antwort auf:/In reply to: Re: a question about thinking output geschrieben von:/posted by: Peter Berger at 21 June 2004 00:55:35:

Hi Peter,

One question I was wondering about:
The rules are not really clear to wether it is allowed to setup another time
control than the official one. I have no idea if this has ever been an
issue.

If I remember correctly there is a rule that you must adjust the clock of the chess engine 'near' to the board clock. Whatever near means...
I would definitely recommand to the near entrants of the WCCC to play at least ONE test match under tournament conditions, at best you invite a good friend and let him play with another program at a real board. That is at least what I did. (Anyway there is still a bug in my tournament time control, it has sometimes little problems when a fail low occurs in move 36, 37, 38 or 39)
Greets, Thomas