+- Ludii Forum (https://ludii.games/forums)
+-- Forum: Suggestions (https://ludii.games/forums/forumdisplay.php?fid=10)
+--- Forum: Submit Your Games (https://ludii.games/forums/forumdisplay.php?fid=23)
+--- Thread: Nodal Chess (/showthread.php?tid=1203)
Nodal Chess - dale walton - 09-03-2022
This is a chess variant that is played on the intersections.
The script seems to work, but the move counting function is probably all messed up....
Please see how you like the game.
Also, I was able to isolate one of the bugs that I worked around.
Please note the difference when loading the trial using the correct NodalChess.lud or using NodalChess1.lud
The only script difference is in the later version, I have the default (to...) for the slide scripted explicitly - which causes strange behavior in this particular case.
(try backing up a couple moves and relocating the queen one space in the buggy game-over version... - The trial shouldn't be checkmate...)
I also had difficulty with using "Forwards" for diagonal + straight forward on this board, among other direction problems.
On the ai - in the end game the king should be moving aggressively, not running away if it has the advantage. how could that behavior be encouraged (in other parts of the game as well, the AI sometimes seems to be simply trying to mark time.)
-------------------------
And a different version where I defined the board by adding diagonal edges, didn't allow for the suppression of diagonals in display simultaneously with showing colors on the board. In that version Orthogonal direction using vertices included the diagonals = adjacent, and (difference Adjacent Diagonal) and Forwards didn't behave well. I might be able to find it if you want to look at those issues, but not with the bugs.
RE: Nodal Chess - dale walton - 09-05-2022
Another Bug in the AI:
AB doesn't see my checkmate. Thinks it is 5 moves away.
RE: Nodal Chess - Eric Piette - 09-06-2022
Hi,
1) I am not sure what you mean by the "move counting function"?
2) For the slide bug you mention. It is not. The default value of (to ..) in slide is just out of date, I removed it from the doc.
But having (to if:(is In (to) (sites Empty))) in the slide move makes the slide move to stop as soon as it reachs an empty site, which is not the same has having no condition for the (to ....) meaning that only the between condition (sliding on empty sites) will be used.
3) For the directions, did you use the (directions ...) ludeme? Which allow you to decide the relative directions according to different relation type? Because for games played on vertices that's often necessary for them to work as wished.
4) I will add this game to Ludii
5) I do not think the ai is buggy, it just did not see it with the time you allocated to it. But I will ask Dennis to have a look, just in case.
RE: Nodal Chess - DennisSoemers - 09-06-2022
The checkmate in that picture was indeed 5 moves away:
124. Black passes
125. White king from E5 to E6
126. White passes
127. Black pawn from A7 to A6
128. White bishop from D5 to A8
RE: Nodal Chess - dale walton - 09-07-2022
Thanks for adding it to Ludii. Note I may need to update the move counting function, but this should not affect the essense of the game very much.
RE 1) I am not sure what you mean by the "move counting function"?
I meant setting the counting of moves until a draw is declared - I am pretty sure I messed it up, but I have a bit of work left in deciding how that should work for this version of chess anyway.
2) Glad you updated the documentation, hope you added to it the other info you give above.
3) I did for some ways I tried to deal with it, but eventually All and Diagonal started to work, and I came to this solution. I think when I used Forwards it was not inside the (direction) ludeme. Is there a place that this is addressed in the documentation?
-- Also the documentation should mention
- that cells and regions cannot receive a color when using Edge-based graphics, (or make it possible to do so) and
- that all defined edges are in the category of Orthogonal, whether or not they are orthogonal. And
- that edges across cells won't show if a Cell style BoardStyle such as Chess style board is used, unless the non-edge diagonals are also shown.
RE: The checkmate in that picture was indeed 5 moves away:
But my checkmate was 1 move away - so why didn't it see the one move away checkmate first?
RE: Nodal Chess - Eric Piette - 09-07-2022
Hi,
All edges are orthogonal to each other. I am inviting you to have a look at: http://piette.info/eric/papers/acg21-1.pdf
For the direction question, that's just explained in the doc of the parameter of the (directions ...) one.
For the graphics questions, I will ask Matthew to answer to you.
Eric
RE: Nodal Chess - DennisSoemers - 09-07-2022
> But my checkmate was 1 move away - so why didn't it see the one move away checkmate first?
In your screenshot, the two final messages I see printed by the AI are:
1) proven loss at depth 5
2) completed search of depth 0
Judging from the trial file, the final sequence of 5 moves was:
124. Black passes
125. White king from E5 to E6
126. White passes
127. Black pawn from A7 to A6
128. White bishop from D5 to A8
The AI only prints messages during its own turns, not during your turns. So, the second-to-last message about the proven loss at depth 5 was printed at step 124 (at that point this was correct), and the last message about completing a depth-0 search was printed at step 127. A depth-0 search means it didn't search at all, which is because it only had a single legal move at that point, so running a search would've been a waste of time.
RE: Nodal Chess - dale walton - 09-08-2022
OK, Thanks I understand what you are saying now.
RE: Nodal Chess - Updated Script - dale walton - 09-11-2022
I have now implemented the anti-stalling rule (turn count limits).
Please update the script to this version
RE: Nodal Chess - Eric Piette - 09-12-2022
Hi,
Ok, I did it for the next release.
Regards,
Eric