Most common chess variant?

[Isaac]

I'm a huge fan of the crazyhouse variant, have you considered it?

[Ferdy]

I'm a huge fan of the crazyhouse variant, have you considered it?

Try Imortal. I plan to convert it to bitboard for easier attack calculation, but have not find the right time yet.

[Evert]

I would have thought it made it easier that the second step would always have to be the same as the first one, in a locust.

That probably depends on the implementation. Locusts are tricky in SjaakII anyway (I still haven't decided how to handle them in general) because the move generator is set-based rather than direction-based.
Steppers are the exception here, but for those I don't have multi-step...

[Isaac]

I'm a huge fan of the crazyhouse variant, have you considered it?

Try Imortal. I plan to convert it to bitboard for easier attack calculation, but have not find the right time yet.

I remember I tried it agains tjchess and it was significantly weaker.
One interesting but sad thing about crazyhouse programs is that an average player can beat them every once in a while.

[Ferdy]

I'm a huge fan of the crazyhouse variant, have you considered it?

Try Imortal. I plan to convert it to bitboard for easier attack calculation, but have not find the right time yet.

I remember I tried it agains tjchess and it was significantly weaker.
One interesting but sad thing about crazyhouse programs is that an average player can beat them every once in a while.

Ola had done a nice tournament before with a fair opening distribution, TC 40/20 minutes.

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB 

                              Sunse TJche Imort Sjeng Feuer Pulsa KKFCh 
 1. Sunsetter                 ##### 01101 01111 11101 11111 11111 11111 
                              ##### 11111 00111 01111 11111 11011 111=1   86%  51.5 (1585.0, 1293.3) 

 2. TJchess 1.1-x64           10010 ##### 10011 01111 11110 11111 11111 
                              00000 ##### 11001 11110 11111 01111 11111   73%  44.0 (1660.0, 978.0) 

 3. Imortal v1.0              10000 01100 ##### 11111 11011 11111 11111 
                              11000 00110 ##### 11011 11101 11111 11111   73%  44.0 (1660.0, 960.5) 

 4. Sjeng 11.2                00010 10000 00000 ##### 11001 11111 01111 
                              10000 00001 00100 ##### 01110 01001 11111   45%  27.0 (1830.0, 522.5) 

 5. Feuerstein                00000 00001 00100 00110 ##### 10010 11111 
                              00000 00000 00010 10001 ##### 11111 11111   40%  24.0 (1860.0, 387.0) 

 6. Pulsar2009-9b             00000 00000 00000 00000 01101 ##### 10111 
                              00100 10000 00000 10110 00000 ##### 11101   27%  16.0 (1940.0, 276.5) 

 7. KKFChess v2.6.6           00000 00000 00000 10000 00000 01000 ##### 
                              000=0 00000 00000 00000 00000 00010 #####    6%   3.5 (2065.0,  84.8)

Link:
http://www.talkchess.com/forum/viewtopi ... crazyhouse

Where can we best test the engine to play against humans?

[stegemma]

[...]For immediate reference, mine are

Code: Select all

1	 23
2	 529
3	 12012
4	 273026
5	 6223994
6	 142078049
7	 3299342874

Thanks, now it's ok up to depth 5...

[Evert]

[...]For immediate reference, mine are

Code: Select all

1	 23
2	 529
3	 12012
4	 273026
5	 6223994
6	 142078049
7	 3299342874

Thanks, now it's ok up to depth 5...

SjaakII can do divided perft if you need it. I would guess there is a problem with promotions, which are on the sixth rank, to Met (Ferz) and mandatory.

[stegemma]

[...]For immediate reference, mine are

Code: Select all

1	 23
2	 529
3	 12012
4	 273026
5	 6223994
6	 142078049
7	 3299342874

Thanks, now it's ok up to depth 5...

SjaakII can do divided perft if you need it. I would guess there is a problem with promotions, which are on the sixth rank, to Met (Ferz) and mandatory.

At depth 6 the pawns can reach the promotion row but only at ply 8 they can move again, as Met. I'm investigating about check detection... maybe because a promoted pawn already on rows 6, with black king moved up some row.

[stegemma]

In fact there were an error in my makrukQueenSquares table, used for check testing. Now i get only 7 moves of error:

Code: Select all

debug: perft 6 Nodes: 142078056, Time: 26395 ms, Nodes/s: 5382560
release: perft 6 Nodes: 142078056, Time: 5423 ms, Nodes/s: 26194331

do to my limited knowledge of this game, it would be hard to debug... but i try to compare with your divide.

Thanks for the information.

[syzygy]

Do you really need a different search for suicide chess? I never really looked at it because I don't find it particularly interesting, but I figured that making the piece values negative would do it?

That worked in the first few days after suicide chess was added to FICS, but nowadays it will make the engine lose miserably.

In suicide chess mobility is all important, so the more pieces the better unless tactics tells you otherwise.

What works extremely well in suicide chess to cover tactics is proof number search. But it's not trivial to combine this with alpha-beta.

Anyway, suicide chess is about to be solved as a win for white (only 1...b6 is not yet known to lose to 1.e3!).