Creating EGTBs.

[Carey]

I have a couple friends who are trying to write their own EGTB generator, but they are having some problems.

Yes, they could use somebody elses code, but they'd rather work out the details themselves.

Since I've never done an EGTB myself, I gave them a link to Muller's code, along with a scan of Ken Thompson's classic paper from the ICCAJ.

Saying they are confused by Muller's code is a bit of an understatement.

And although Ken Thompson's description seemed pretty easy to me when I read it, they are having some trouble getting it to work right.

The depth to mate seems to be much higher than what Ken reports or what they see elsewhere.


Does anybody have a good, clear description of how to generate the table bases, along with reliable numbers they can easily compare with?


It'd probably help if they'd post in here themselves, but they don't seem to be too eager to ask in here. (Probably afraid they'd look a little foolish, since almost everybody else in here is way above our level.)

I've tried to help as much as I can, but I've never done an EGTB at all and have little interest in doing one (and personally I consider it to be cheating to actually use them in a game.)

But I can understand the challenge of writing it and getting it to work, which is why I'm asking for help that I can pass along to them.

Any links, descriptions, pointers, suggestions, etc. will be appreciated and passed along to them.

Carey

[hgm]

The CCRL forum has a section devoted to tablebases. There is not much posting there, but the people that do post are often very knowledgeable. Perhaps you should also post your request there.

It is true that the code of my tablebase builder was not created for didactical purposes...

[rjgibert]

IIRC, Thompson's EGTB is not DTM (=Distance To Mate), but rather DTC (Distance To Conversion), which is the distance to a winning capture or mate. This would explain the discrepancy.