Chess variant tournament: Shatranj

[hgm]

Unrealistic eval does not give you a sudden score jump. That is the hallmark of search bugs.

[enhorning]

Btw, this source says you would have to start counting at '6' in the KRRNK case, where draw is reached at '8', so that white would really only have 2 moves to checkmate you. It also says the counting only is to be done when you have no Pawns, even if the opponent has bare King.

www.ancientchess.com - That's where I bought my Makruk set from:

...and Shatranj... same board, different pieces:

... as well as a few other ones.

So, for Makruk, seeing as how the majority of program don't support counting, and there does not seem to be an agreement about exactly how it works, I am planning on running the tournament without the counting rules.

[Evert]

Unrealistic eval does not give you a sudden score jump. That is the hallmark of search bugs.

Well, yeah, ish.
I was thinking "bad evaluation leading to bad pruning decisions", which would have the symptoms of a search problem (bad pruning) but the cause would be bad evaluation.

That's not saying there can't be search bugs, of course, but there cannot be Shatranj *specific* search bugs (at least not with material for both sides on the board), so they'd have to show up in other variants. There can, however, easily be Shatranj specific evaluation bugs/misevaluations.

[Ferdy]

I got a working version now without the counting rules. And instead of 50 I use 64 moves rule that is pawn moves and captures will reset it to 0.
I will try to implement next the bare king counting rules. It is interesting to see an engine not to bare the opponent if the number of moves to mate
is still not possible.

Code: Select all

Computer Makruk Chess Game
TC 40/5min, 2013.03.12
                        Score     Bilis_v0.2           NebiyuChess_1.43     Sjaak 524            Fairy-Max 4.8S      
----------------------------------------------------------------------------------------------------------------------
 1: Bilis_v0.2        45.0 / 60   XXXXXXXXXXXXXXXXXXXX 11==011011===010=11= 11111=11=10=1110==1= =11111=11=11=111111=  (+36 -6 =18)
 2: NebiyuChess_1.43   7.5 / 20   00==100100===101=00= XXXXXXXXXXXXXXXXXXXX .................... ....................  (+4 -9 =7)
 3: Sjaak 524          5.0 / 20   00000=00=01=0001==0= .................... XXXXXXXXXXXXXXXXXXXX ....................  (+2 -12 =6)
 4: Fairy-Max 4.8S     2.5 / 20   =00000=00=00=000000= .................... .................... XXXXXXXXXXXXXXXXXXXX  (+0 -15 =5)
----------------------------------------------------------------------------------------------------------------------
60 games: +19 =18 -23

[hgm]

Indeed, I wondered about that. The counting rules did not say anything about counting while there still are Pawns.

I am also working on an upgrade of Pair-o-Max, equiping it with true move sorting and killer heuristic. (As well as in-tree repeat detection and update of the game phase.) I could make a dedicated Makruk version that is shy to bare the opponent or waste all Pawns.

Makruk is played by millions of people, and having an engine that plays it well is a wothwhile endeavor in itself.