Help us make Cute Chess GUI better

[hgm]

I was also asked to add support for Chessbase opening books, and while that's not as impossible, it's still difficult and frustrating enough to cause severe hair loss and brain damage.

Better to have a book converter that could do CTG -> Polyglot conversion. That would solve the problem once and for all.

[Don]

I was also asked to add support for Chessbase opening books, and while that's not as impossible, it's still difficult and frustrating enough to cause severe hair loss and brain damage.

Better to have a book converter that could do CTG -> Polyglot conversion. That would solve the problem once and for all.

Some conversions would only be approximate in the cases where a book has information that another book doesn't support. But of course there is nothing to be done about that, there really should be tools to make all these conversions.

[mike angel]

I use timecontrol 40 moves in 5 min repeating
over 200 engines in use

what would be very nice - if Cute GUI could recognize time forfeit games so it could be delete easily with the GUI and also for replay these games one only had to push a botton

thanx in advance
mike

[Ajedrecista]

Hi again:

I have more proposals (sorry if they are implemented and I have not noticed them):

Regarding cutechess-cli, in a direct match between only two engines, I see something like:

Code: Select all

Finished game 2499 (Depth_10 vs Depth_9): 1/2-1/2 {Draw by adjudication}
Score of Depth_9 vs Depth_10: 191 - 1189 - 1120  [0.30] 2500
ELO difference: -147
Finished match

The score is only rounded up to 1%, but I think that could be more precise (rounded up to 0.1% or 0.01%); other thing is that the Elo difference is calculated at the end... if this Elo gap is calculated after each finished game (rounded up to 1 Elo or even up to 0.1 Elo), would it slow down cutechess-cli a lot? I guess that adding error bars for a given confidence interval (e.g. 95%) is a crazyness, so error bars should be discarded; the same applies with LOS (likelihood of superiority): too much work.

I also see that cutechess-cli explains why a game is drawn (50-move rule, stalemate...) or lost (checkmate, by adjudication, on time...). I find this info very useful, but once the match is finished, this info 'disappears' when you close the command line interface (it can be found in the output PGN file), but I find interesting to summarize this info just like LittleBlitzer GUI (that GUI also offers info of average time per move, average depth and average nodes/second of each engine).

Another interesting feature (maybe more suitable in CuteChess GUI) is estimating the expected remaining time from the current elapsed time after each finished game, for example with a lineal approach (just like LB again, if I am not wrong). I hope no typos:

Code: Select all

t: current elapsed time after x finished games.
T: estimated total time.
T - t: estimated remaining time.
n: number of games of the whole match.
x: number of finished games at time t.

T = (n/x)*t; T - t = (n/x - 1)*t

-----------------------

Example: t = 10526 sec. = 2h55'26"; n = 2500 games; x = 746 games.

T - t = (2500/746 - 1)*10526 ~ 24748.8 ~ 24749 sec. remaining = 6h52'29" remaining.

Thanks in advance for your attention. CuteChess team's work is greatly appreciated here! I wait for 'pause and resume' feature.

Regards from Spain.

Ajedrecista.

[ilari]

Hi again:

I have more proposals (sorry if they are implemented and I have not noticed them):

Regarding cutechess-cli, in a direct match between only two engines, I see something like:

Code: Select all

Finished game 2499 (Depth_10 vs Depth_9): 1/2-1/2 {Draw by adjudication}
Score of Depth_9 vs Depth_10: 191 - 1189 - 1120  [0.30] 2500
ELO difference: -147
Finished match

The score is only rounded up to 1%, but I think that could be more precise (rounded up to 0.1% or 0.01%); other thing is that the Elo difference is calculated at the end... if this Elo gap is calculated after each finished game (rounded up to 1 Elo or even up to 0.1 Elo), would it slow down cutechess-cli a lot? I guess that adding error bars for a given confidence interval (e.g. 95%) is a crazyness, so error bars should be discarded; the same applies with LOS (likelihood of superiority): too much work.

Of course I can make the score percentage more precise if that's needed. And no, it doesn't slow down anything to calculate the ELO gap after each game. In fact you can already do it by setting the "-ratinginterval" option to 1.

I also see that cutechess-cli explains why a game is drawn (50-move rule, stalemate...) or lost (checkmate, by adjudication, on time...). I find this info very useful, but once the match is finished, this info 'disappears' when you close the command line interface (it can be found in the output PGN file), but I find interesting to summarize this info just like LittleBlitzer GUI (that GUI also offers info of average time per move, average depth and average nodes/second of each engine).

When we have a file format for the tournaments then we can focus on creating some interesting tournament statistics.

Another interesting feature (maybe more suitable in CuteChess GUI) is estimating the expected remaining time from the current elapsed time after each finished game, for example with a lineal approach (just like LB again, if I am not wrong). I hope no typos:

Code: Select all

t: current elapsed time after x finished games.
T: estimated total time.
T - t: estimated remaining time.
n: number of games of the whole match.
x: number of finished games at time t.

T = (n/x)*t; T - t = (n/x - 1)*t

-----------------------

Example: t = 10526 sec. = 2h55'26"; n = 2500 games; x = 746 games.

T - t = (2500/746 - 1)*10526 ~ 24748.8 ~ 24749 sec. remaining = 6h52'29" remaining.

Something like this is also planned.

Thanks in advance for your attention. CuteChess team's work is greatly appreciated here! I wait for 'pause and resume' feature.

Regards from Spain.

Ajedrecista.

Thanks for using Cute Chess (CLI) and helping us improve it. You'll get your pause and resume once we're done with the tournament file format. It's our goal that you could start a tournament in the GUI, stop it, and later resume it in cutechess-cli. Or vice versa of course.

[hgm]

It's our goal that you could start a tournament in the GUI, stop it, and later resume it in cutechess-cli. Or vice versa of course.

Would such starting and stopping include resuming the interrupted game? Or would it have to be replayed? I was wondering if I should build a feature in WinBoard to resume interrupted games. Normally an interrupted game would be written to the PGN file with a '*' result. But the clock settings could be stored in the result comment, so that on loading the game the clocks could be restored, and the game continued. E.g it could say something like

* {adjourned; white time: 85.3, black time 61.7}

where 'adjourned' could perhaps be replaced by a more specific reason, like 'xboard exit', 'game aborted'...

[Ajedrecista]

Hi Ilari:

It is good to know that some features already exist, such as -ratinginterval... I think I read it in the readme file, but I forgot about it! Thanks for the reminder.

Thanks for using Cute Chess (CLI) and helping us improve it. You'll get your pause and resume once we're done with the tournament file format. It's our goal that you could start a tournament in the GUI, stop it, and later resume it in cutechess-cli. Or vice versa of course.

This feature is very much appreciated. I am running an ambitious fixed depth test between two copies of Quazar 0.4 w32, one at depth d, the other at depth (d - 1). Each match consists in 2500 games because I want to narrow the error bars. I started (depth 1) versus (depth 2) and the last match was (depth 9) versus (depth 10)... (so, I have played 22500 games now, which is a huge amount for me, impossible of doing something similar without cutechess-cli), but this last match lasted almost ten hours in my Dual Core! Extrapolating the elapsed time for the next match (depth 10 versus depth 11), it would take more than twenty hours, and this amount of time is impossible for me. I seldom have access to a quad (an Intel i5-760) that saves time in somewhat more than a 50% in comparison to my Dual Core... but still I do not have access of eight or ten consecutive hours to that quad. So this feature will enable to follow my fixed depth testing a little more (maybe two more 2500-game matches: depth 10 versus depth 11 and depth 11 versus depth 12). At the moment, my results are sometimes the expected ones, sometimes not exactly what I expected. I will share the results once I consider that my experiment is finished.

A last question that can be a little pretentious: could you give a rough estimate for the date release of CuteChess GUI 1.0? Somewhat like October of 2012, November of 2012...? Thanks in advance once again for your patience and your great work! I stay tunned.

Regards from Spain.

Ajedrecista.

[ilari]

It's our goal that you could start a tournament in the GUI, stop it, and later resume it in cutechess-cli. Or vice versa of course.

Would such starting and stopping include resuming the interrupted game? Or would it have to be replayed? I was wondering if I should build a feature in WinBoard to resume interrupted games. Normally an interrupted game would be written to the PGN file with a '*' result. But the clock settings could be stored in the result comment, so that on loading the game the clocks could be restored, and the game continued. E.g it could say something like

* {adjourned; white time: 85.3, black time 61.7}

where 'adjourned' could perhaps be replaced by a more specific reason, like 'xboard exit', 'game aborted'...

It would be preferable to be able to continue interrupted games, so saving the state of the clocks is important. I was thinking about saving that kind of information in the tournament file, not the PGN file.

[ilari]

A last question that can be a little pretentious: could you give a rough estimate for the date release of CuteChess GUI 1.0? Somewhat like October of 2012, November of 2012...? Thanks in advance once again for your patience and your great work! I stay tunned.

Regards from Spain.

Ajedrecista.

If I can continue to work on Cute Chess at the current pace, then I guess we could have a product ready in a couple of months. But then again, we're working on Cute Chess on our free time, which is limited, so it's impossible to set an accurate release date. Usually software projects tend to take more time than planned, and Cute Chess is no exception.

[hgm]

It would be preferable to be able to continue interrupted games, so saving the state of the clocks is important. I was thinking about saving that kind of information in the tournament file, not the PGN file.

That would certainly be an alternative. But not every game might be part of a tournament. E.g. if someone is playing against an engine, and is called away or runs out of battery power. And putting all info in the PGN would allow you to also finish the game in another context.