7-men Syzygy attempt

Discussion of chess software programming and technical issues.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Arpad Rusz
Posts: 270
Joined: Sat Apr 17, 2010 12:34 pm
Location: Budapest

Re: 7-men Syzygy attempt

Post by Arpad Rusz » Thu May 24, 2018 8:48 pm

syzygy wrote:
Thu May 24, 2018 4:22 pm
The end of the story is that to get KPPPvKPP, you need all other 4v3 tables first. So that's what being worked on.
Yes, I know how it works, I myself have written a tablebase generator for a chess variant. :)

Of course, is it desirable to have the complete set, and I plan to get it someday. But for now, I feel perfectly safe to download only the tablebases with material difference less or equal than five.

noobpwnftw
Posts: 340
Joined: Sun Nov 08, 2015 10:10 pm

Re: 7-men Syzygy attempt

Post by noobpwnftw » Thu May 24, 2018 9:02 pm

Progress update: both machines are steadily building pawnful tables, recently, upload is on hold because I'm moving data to another server with larger storage, should resume in a few days.
The new server will be able to hold much more data for the years to come, and much faster.

Dann Corbit
Posts: 9983
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: 7-men Syzygy attempt

Post by Dann Corbit » Thu May 24, 2018 10:41 pm

noobpwnftw wrote:
Thu May 24, 2018 9:02 pm
Progress update: both machines are steadily building pawnful tables, recently, upload is on hold because I'm moving data to another server with larger storage, should resume in a few days.
The new server will be able to hold much more data for the years to come, and much faster.
I kneel in genuflection before you and Ronald for your incredibly useful service to mankind. Or, at least, useful service to a bunch of geeky chess fanatics.

Besides chess itself, the outcomes are also mathematically interesting.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Ozymandias
Posts: 1102
Joined: Sun Oct 25, 2009 12:30 am

Re: 7-men Syzygy attempt

Post by Ozymandias » Fri May 25, 2018 4:58 am

Dann Corbit wrote:
Thu May 24, 2018 10:41 pm
useful service to a bunch of geeky chess fanatics
That sounds about right. :wink:

User avatar
Nordlandia
Posts: 2406
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: 7-men Syzygy attempt

Post by Nordlandia » Fri May 25, 2018 6:17 am

Sesse wrote:
Thu May 24, 2018 8:31 pm
Nordlandia wrote:
Thu May 24, 2018 7:57 pm
Houdini's intelligent probing algorithm is comparable to how an floodgate operate. Water is tablebases, gate is egtb probe depth (intelligent probing). At 0 the gate is completely open, while at 5 the gate is half closed which means the tablebase flow is restricted by 50%.
I am unable to make sense of that analogy in any reasonable way.
I made this diagram: Image

Sesse
Posts: 180
Joined: Mon Apr 30, 2018 9:51 pm
Contact:

Re: 7-men Syzygy attempt

Post by Sesse » Fri May 25, 2018 7:25 am

I still don't understand a thing. Are you sure it really works this way?

User avatar
Nordlandia
Posts: 2406
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: 7-men Syzygy attempt

Post by Nordlandia » Fri May 25, 2018 8:43 am

Only Houdart know for sure. Intelligent probing like this is good solution for 7-men syzygy as egtb flow is enormous, intelligent algorithm based on the diagram is maybe a way to approach 7-men. Intelligent probing algorithm should in theory limit speed wasted on egtb probe.

From Houdini manual.

EGTB Probe Depth
The frequency of the EGTB probing can be configured with the EGTB Probe Depth
option.

EGTB probes are relatively slow compared to a normal evaluation by the engine.
Suppose you have a K+Q+P against K+N ending. Even without consulting the table
bases Houdini knows that this ending is easily won for the K+Q+P side. Consulting the
EGTB for this position would reduce Houdini's playing strength, as it could easily have
searched and evaluated many additional positions instead of making the rather useless
EGTB probe.
Houdini contains a lot of end game knowledge in its native evaluation function - a lot
more than most other chess engines. For example, even in a non-obvious ending like
KBP v KB with bishops of the same color, Houdini knows quite well which positions are
won and which are drawn. This means that except for difficult-to-evaluate endings like
KQP v KQ or KRP v KR the EGTB do not really increase Houdini's playing strength.
To handle this situation in an intelligent way, Houdini will vary the frequency at which
the EGTB are probed depending on the actual end game that is occurring. In difficultto-
evaluate endings Houdini will probe the EGTB files much earlier than in endings that
are easy to evaluate.
The frequency of the probing is influenced by the EGTB Probe Depth parameter. You
can decrease this parameter to force Houdini to use the EGTB more frequently, or
increase it to have Houdini use the EGTB less frequently. The default value of 1 is
appropriate for systems with the table bases on SSD. If the table bases are on a
slower, mechanical hard disk you should consider increasing this value to reduce the
frequency of probing.

syzygy
Posts: 4450
Joined: Tue Feb 28, 2012 10:56 pm

Re: 7-men Syzygy attempt

Post by syzygy » Fri May 25, 2018 1:35 pm

Nordlandia wrote:
Fri May 25, 2018 8:43 am
From Houdini manual.
Which corresponds to what I said and thankfully is silent on sluice gates.

syzygy
Posts: 4450
Joined: Tue Feb 28, 2012 10:56 pm

Re: 7-men Syzygy attempt

Post by syzygy » Sat May 26, 2018 12:49 am

Arpad Rusz wrote:
Thu May 24, 2018 11:42 am
I have created a heat map for the 4v3 tablebases.
I just noticed that RPP is missing ;-)

Arpad Rusz
Posts: 270
Joined: Sat Apr 17, 2010 12:34 pm
Location: Budapest

Re: 7-men Syzygy attempt

Post by Arpad Rusz » Sat May 26, 2018 7:40 pm

Thanks, now it is fixed. I have added the RPP line.
I will be great to finally have the RPP/RP tablebase, and RPP/QP will be important to avoid some fortresses.

Post Reply