Still looking for the engine that can analyze this position

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
MikeB
Posts: 3380
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Still looking for the engine that can analyze this posit

Post by MikeB » Thu Dec 02, 2010 2:51 am

Milos wrote:
MikeB wrote: The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige.
Why it is not possible???
It's quite possible, it's however highly improbable.
It's impossible for both sides - black and white to promote all 16 pawns to Q. Think about how each side would get ALL the pawns to the queenning square. There are not enough non-pawn pieces on the board to facilitate the column manuevering for the pawns to pass each other. It is not "highly improbable" , but 100% impossible.

bob
Posts: 20471
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Still looking for the engine that can analyze this posit

Post by bob » Thu Dec 02, 2010 2:52 am

mhalstern wrote:The fritz12 Gui dosen't understand the illegality of the position and will let any engine play from it. It only considers illegal positions as those with more extra pieces than pawns could have been promoted to.
The position in question certainly should be flagged since one side has 31 queens...

bob
Posts: 20471
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Still looking for the engine that can analyze this posit

Post by bob » Thu Dec 02, 2010 2:55 am

MikeB wrote:
Dann Corbit wrote:[d]qqqqkqqq/qqqqqqqq/qqqqqqqq/qqqqqqqq/QQQQQQQQ/QQQQQQQQ/QQQQQQQQ/QQQQKQQQ w - -

What do you mean it's not a real chess position? I want to see it analyzed.

So far, every engine has crashed instantly except Rybka of all the engines that I tried, but Rybka refused to utter a single syllable after 15 mintues of thought. I think the slimy fish knows the answer, but he's not telling just to spite me.
slightly different twist -


max threads set to 1.
White(1): [d] 3qk3/qqqqqqqq/8/8/8/8/QQQQQQQQ/3QK3 w - - 0 0

Code: Select all

White(1): g
              time surplus   0.00  time limit 30:00 (+0.00) (30:00)
              depth   time  score   variation (1)
                1    13.11  12.68   1. Qaxf7+  (4.4Mnps)
                1->  13.14  12.68   1. Qaxf7+  (4.4Mnps)
                2    13.14  12.68   1. Qaxf7+ Qxf7 2. Qcxh7
                2->  13.17  12.68   1. Qaxf7+ Qxf7 2. Qcxh7
                3    13.19  12.68   1. Qaxf7+ Qxf7 2. Qcxh7
                3->  13.20  12.68   1. Qaxf7+ Qxf7 2. Qcxh7
                4    13.22     +1   1. Qaxf7+!
                4    13.22     +3   1. Qaxf7+!
                4    13.22     +M   1. Qaxf7+!
                4    13.22  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qhxf8#
                4->  13.25  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qhxf8#
                5    13.25  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
                5->  13.26  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
                6    13.28  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
                6->  13.30  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
                7    13.31  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
                7->  13.34  Mat04   1. Qaxf7+ Qxf7 2. Qh8+ Qxh8 3. Qxh8+
                                    Qf8 4. Qgg6#
              time=13.36  mat=0  n=58429045  fh=98%  nps=4.4M
              extensions=566 qchecks=10K reduced=959 pruned=18K
              predicted=0  evals=64K  50move=0  EGTBprobes=0  hits=0
              SMP->  splits=0  aborts=0  data=0/256  elap=13.36

mate in 4 moves.

White(1): Qaxf7+
              time used:  13.36
Black(1):
Crafty has no output for 13 seconds and then annouces mate-in-4. The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige. One oddity whether I use four cores or one core, Crafty takes about the same amount of time with the about same # of nps.
It it likely the position is reachable. One has 6 pieces to sacrifice to open holes in the pawn structure, and both sides can then push pawns through those holes ignoring captures if they want. I'd have to think pretty long on this to decide if it is really a reachable position or not, and I don't intend to ever include this kind of legality test in Crafty since it is computationally intractable...

But clearly positions with > 9 queens on one side, or > 10 n/b/r on one side can't be reached, ever, and Crafty does detect those.

bob
Posts: 20471
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Still looking for the engine that can analyze this posit

Post by bob » Thu Dec 02, 2010 2:57 am

Graham Banks wrote:
Milos wrote:
MikeB wrote: The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige.
Why it is not possible???
It's quite possible, it's however highly improbable.
Wouldn't the last move have had to have been a pawn promoting to a queen, in which case the square from which it had moved would be empty?
Why? one can first create 8 new queens, then shuffle them at will to reach that position...

bob
Posts: 20471
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Still looking for the engine that can analyze this posit

Post by bob » Thu Dec 02, 2010 2:59 am

MikeB wrote:
Milos wrote:
MikeB wrote: The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige.
Why it is not possible???
It's quite possible, it's however highly improbable.
It's impossible for both sides - black and white to promote all 16 pawns to Q. Think about how each side would get ALL the pawns to the queenning square. There are not enough non-pawn pieces on the board to facilitate the column manuevering for the pawns to pass each other. It is not "highly improbable" , but 100% impossible.
Here's an idea. White sacrifices 4 pieces to double black pawns on the a, c, e and g files. Black sacrifices 4 pieces to double white pawns on the b, d, f and h files. Now you just push the pawns up the ranks and promote them one at a time.

I don't see why it is impossible...

User avatar
Graham Banks
Posts: 32990
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Still looking for the engine that can analyze this posit

Post by Graham Banks » Thu Dec 02, 2010 3:09 am

bob wrote:
Graham Banks wrote:
Milos wrote:
MikeB wrote: The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige.
Why it is not possible???
It's quite possible, it's however highly improbable.
Wouldn't the last move have had to have been a pawn promoting to a queen, in which case the square from which it had moved would be empty?
Why? one can first create 8 new queens, then shuffle them at will to reach that position...
Suggest what the last move could have been. I'm looking at the position where there are no empty squares on the board.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

Milos
Posts: 3387
Joined: Wed Nov 25, 2009 12:47 am

Re: Still looking for the engine that can analyze this posit

Post by Milos » Thu Dec 02, 2010 3:15 am

Graham Banks wrote:Suggest what the last move could have been. I'm looking at the position where there are no empty squares on the board.
We are not talking about that position. That one is clearly impossible (and illegal).
We are talking about this one:
[d] 3qk3/qqqqqqqq/8/8/8/8/QQQQQQQQ/3QK3 w - - 0 0
which is perfectly legal and possible.

User avatar
Graham Banks
Posts: 32990
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Still looking for the engine that can analyze this posit

Post by Graham Banks » Thu Dec 02, 2010 3:22 am

Milos wrote:
Graham Banks wrote:Suggest what the last move could have been. I'm looking at the position where there are no empty squares on the board.
We are not talking about that position. That one is clearly impossible (and illegal).
We are talking about this one:
[d] 3qk3/qqqqqqqq/8/8/8/8/QQQQQQQQ/3QK3 w - - 0 0
which is perfectly legal and possible.
Okay. My bad. 8-)
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

MikeB
Posts: 3380
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Still looking for the engine that can analyze this posit

Post by MikeB » Thu Dec 02, 2010 5:09 am

bob wrote:
MikeB wrote:
Milos wrote:
MikeB wrote: The position is not possibe from a real game of chess - but Crafty sees the 8+1 queens on both sides and is happy to oblige.
Why it is not possible???
It's quite possible, it's however highly improbable.
It's impossible for both sides - black and white to promote all 16 pawns to Q. Think about how each side would get ALL the pawns to the queenning square. There are not enough non-pawn pieces on the board to facilitate the column manuevering for the pawns to pass each other. It is not "highly improbable" , but 100% impossible.
Here's an idea. White sacrifices 4 pieces to double black pawns on the a, c, e and g files. Black sacrifices 4 pieces to double white pawns on the b, d, f and h files. Now you just push the pawns up the ranks and promote them one at a time.

I don't see why it is impossible...
Good point _ didn't think of that.

MikeB
Posts: 3380
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Still looking for the engine that can analyze this posit

Post by MikeB » Thu Dec 02, 2010 5:16 am

Just for kicks , I had crafty compute the starting chess position , substituting Qs for Ps. First ply took 65 minutes to compute. I know that position is impossible for.sure!. ;)

Post Reply