Help for Syzygy probe?

Discussion of chess software programming and technical issues.

Moderators: hgm, Harvey Williamson, bob

kinderchocolate
Posts: 400
Joined: Mon Nov 01, 2010 5:55 am
Contact:

Help for Syzygy probe?

Post by kinderchocolate » Sun Dec 04, 2016 1:20 pm

My FEN string is: "4k3/3bb3/4b3/8/5p2/8/8/4K3 w - - 0 1". This is clear lost for White.

https://www.dropbox.com/s/hos7vld34495nen/ABCD.png?dl=0

In my session, I had KBBBPvK.rtbw and KBBBPvK.rtbz in the path. The Python package (chess.syzygy) was able to probe it (-2 is the result). However, the Fathom code failed to find anything.

Any idea?

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 9:01 pm
Location: Irvine, CA, USA

Re: Help for Syzygy probe?

Post by Dirt » Sun Dec 04, 2016 5:46 pm

kinderchocolate wrote:My FEN string is: "4k3/3bb3/4b3/8/5p2/8/8/4K3 w - - 0 1". This is clear lost for White.

https://www.dropbox.com/s/hos7vld34495nen/ABCD.png?dl=0

In my session, I had KBBBPvK.rtbw and KBBBPvK.rtbz in the path. The Python package (chess.syzygy) was able to probe it (-2 is the result). However, the Fathom code failed to find anything.

Any idea?
You probably need KQBBBvK.rtbw and KQBBBvK.rtbz. Possibly some others, too.
Deasil is the right way to go.

syzygy
Posts: 4161
Joined: Tue Feb 28, 2012 10:56 pm

Re: Help for Syzygy probe?

Post by syzygy » Sun Dec 04, 2016 7:05 pm

kinderchocolate wrote:However, the Fathom code failed to find anything.
What do you mean by "failed to find anything"?

If it returns a draw, there might be a bug in the Fathom code.

If it returns something like a "probe failed" result, then some table seems to be missing or a path seems to be wrong. (Or it still is a bug in the Fathom code if you have set up everything correctly.)

kinderchocolate
Posts: 400
Joined: Mon Nov 01, 2010 5:55 am
Contact:

Re: Help for Syzygy probe?

Post by kinderchocolate » Mon Dec 05, 2016 4:37 am

syzygy wrote:
kinderchocolate wrote:However, the Fathom code failed to find anything.
What do you mean by "failed to find anything"?

If it returns a draw, there might be a bug in the Fathom code.

If it returns something like a "probe failed" result, then some table seems to be missing or a path seems to be wrong. (Or it still is a bug in the Fathom code if you have set up everything correctly.)
-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 9:01 pm
Location: Irvine, CA, USA

Re: Help for Syzygy probe?

Post by Dirt » Mon Dec 05, 2016 6:01 am

kinderchocolate wrote:
syzygy wrote:
kinderchocolate wrote:However, the Fathom code failed to find anything.
What do you mean by "failed to find anything"?

If it returns a draw, there might be a bug in the Fathom code.

If it returns something like a "probe failed" result, then some table seems to be missing or a path seems to be wrong. (Or it still is a bug in the Fathom code if you have set up everything correctly.)
-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.
Do you have KQBBBvK.rtbw and KQBBBvK.rtbz? Others might be necessary, like KBBBN, but if Fathom is written properly that shouldn't really be necessary.
Deasil is the right way to go.

kinderchocolate
Posts: 400
Joined: Mon Nov 01, 2010 5:55 am
Contact:

Re: Help for Syzygy probe?

Post by kinderchocolate » Mon Dec 05, 2016 6:33 am

Dirt wrote:
kinderchocolate wrote:
syzygy wrote:
kinderchocolate wrote:However, the Fathom code failed to find anything.
What do you mean by "failed to find anything"?

If it returns a draw, there might be a bug in the Fathom code.

If it returns something like a "probe failed" result, then some table seems to be missing or a path seems to be wrong. (Or it still is a bug in the Fathom code if you have set up everything correctly.)
-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.
Do you have KQBBBvK.rtbw and KQBBBvK.rtbz? Others might be necessary, like KBBBN, but if Fathom is written properly that shouldn't really be necessary.
I don't have any other 6-piece file. Do I need other 6-piece files? The Python package was able to give correct information with only KBBBPvK files?

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 9:01 pm
Location: Irvine, CA, USA

Re: Help for Syzygy probe?

Post by Dirt » Mon Dec 05, 2016 6:39 am

kinderchocolate wrote:
Dirt wrote:
kinderchocolate wrote:-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.
Do you have KQBBBvK.rtbw and KQBBBvK.rtbz? Others might be necessary, like KBBBN, but if Fathom is written properly that shouldn't really be necessary.
I don't have any other 6-piece file. Do I need other 6-piece files? The Python package was able to give correct information with only KBBBPvK files?
What's it going to do when the pawn promotes? I suspect it just gives up.
Deasil is the right way to go.

kinderchocolate
Posts: 400
Joined: Mon Nov 01, 2010 5:55 am
Contact:

Re: Help for Syzygy probe?

Post by kinderchocolate » Mon Dec 05, 2016 6:41 am

Dirt wrote:
kinderchocolate wrote:
Dirt wrote:
kinderchocolate wrote:-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.
Do you have KQBBBvK.rtbw and KQBBBvK.rtbz? Others might be necessary, like KBBBN, but if Fathom is written properly that shouldn't really be necessary.
I don't have any other 6-piece file. Do I need other 6-piece files? The Python package was able to give correct information with only KBBBPvK files?
What's it going to do when the pawn promotes? I suspect it just gives up.
Ok. I will try to download the other 6-men TB files. My question, how did the Python package figured out the correct results? It certainly didn't have access to any other 6-men TB files.

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 9:01 pm
Location: Irvine, CA, USA

Re: Help for Syzygy probe?

Post by Dirt » Mon Dec 05, 2016 7:20 am

kinderchocolate wrote:Ok. I will try to download the other 6-men TB files. My question, how did the Python package figured out the correct results? It certainly didn't have access to any other 6-men TB files.
The result is easy to read. How to proceed is more complicated. Partial sets of tablebases have been known to cause problems.
Deasil is the right way to go.

basil00
Posts: 55
Joined: Thu Oct 22, 2015 12:14 am

Re: Help for Syzygy probe?

Post by basil00 » Wed Dec 07, 2016 10:33 am

kinderchocolate wrote:-2 is correct because the position is lost for the player to move. The Python package was correct to return the value. However, when I gave the same path and FEN to Fathom, it failed to return WDL. My suspicion is that there is a bug in the fathom code, but not 100% sure. That's why I posted here.
Works fine for me. Fathom output:

[pgn]
Event ""]
[Site ""]
[Date "??"]
[Round "-"]
[White "Syzygy"]
[Black "Syzygy"]
[Result "0-1"]
[FEN "4k3/3bb3/4b3/8/5p2/8/8/4K3 w - - 0 1"]
[WDL "Loss"]
[DTZ "2"]
[WinningMoves ""]
[DrawingMoves ""]
[LosingMoves "Kd1, Kf1, Kd2, Ke2, Kf2"]

1. Kd1 f3 2. Kc1 f2 3. Kb1 f1=Q+ 4. Kc2 Qd1+ 5. Kb2 Qc1+ 6. Kxc1 Bb4 7. Kd1 Bc4 8. Kc2 Ba4+ 9. Kc1 Kd7 10. Kb1 Ba3 11. Ka1 Ba2 12. Kxa2 Bc1 13. Kb1 Bd2 14. Ka1 Kc6 15. Kb1 Kb5 16. Ka1 Kb4 17. Ka2 Bc1 18. Ka1 Ka3 19. Kb1 Bd2 20. Ka1 Bb5 21. Kb1 Bd3+ 22. Ka1 Bc3# 0-1
[/pgn]

I love the DTZ playing style...

Are you using the Fathom tool or library? If you are using the tool, it will fail unless you have the TB files for all positions in the PGN above. You can use the --test option if you just want to get the TB result for the input position only.

Post Reply