Review of maximal positions

[guyhaw]

Eiko Bleicher has surveyed all the 3-3(p) and 4-2(p) Nalimov DTM EGTs and provided lists of maxDTM positions.
Marc Bourzutschky and Yakov Konoval have surveyed all the 'YK' 3-3 and 4-2 DTC EGTs for maxDTC positions. As there are P-less, they are also maxDTZ positions, and many are maxDTZ50 positions also.
The results are included in the attached spreadsheet, with thanks to EN, EB, YK and MB.
g

[jkominek]

I appreciate the effort it takes to maintain this spreadsheet and be the "minder of chess endgame knowledge." Glad you make it available.

May I ask about the sheet "occurances by endgame force"? What database does it distill, and are you multi-counting? That is, is a combination counted only when a game crosses the threshold into 6-man territory, or will pawn promotions also get counted. e.g. krppkr -> kqrpkr

In the sheet "Nalimove index sizes" I added in Wirth-scheme index sizes and computed the ratio of the two -- with the intention of posting the results -- but haven't completed the task. Your listing for 6-men combinations doesn't seem to follow any "easy algorithmic ordering" and so got stopped by that.

Efficiency ratios, as I have them:
3-men: 0.960
4-men: 0.933
5-men: 0.920

Doesn't mean the compressed files would present the same savings, though.

john

[guyhaw]

Thank you: I am merely the curator of some chess endgame knowledge, and not to be confused with the creators and finders of that knowledge. The data has evolved over many years, so hasn't been that great an effort.
I forget who provided the endgame-occurence information: they should be credited in the spreadsheet but probably aren't. I guess they counted games with specific forces occuring, something I could not do with Chessbase then (or now, as it will not run on my PC). So, a game might contribute to several endgames' counts.
There's a paper by Nalimov, self and Heinz [Nalimov's work, Heinz and I completing the write-up] which gives the contribution from Nalimov's idea to avoid obviously-illegal checks. ICGA Journal (2000), or ACG9 Proceedings (2000) - both available via http://www.tinyurl.com/law6k. The long-run % contribution is one thing, but it varies greatly from endgame to endgame.

I have just x-checked the table's 6-man P-less maxDTC/Z positions generated by Marc Bourzutschky (?) from Yakov Konoval's DTC EGTs. Apart from one instance of finger-trouble, they are all ok. They also imply most of the required maxDTZ50 examples, and I'm hoping to fill in the gaps with John Tamplin - whose endgame-query site http://chess.jaet.org/endings/ is back in multi-metric working form.
g

[Kirill Kryukov]

Thanks Guy! I will add those positions to my page when I have a bit of time!