CCRL Discussion Board

Analogous to, but on a much smaller scale than, the KNNKP challenge ... Ignoring the 50-move rule, wtm can win 136,304,160 positions: with the 50-move rule, wtm should only win 136,275,692 of these. So 28,468 of these positions are affected by the 50m-rule. Only 6,434 of these have DTZ > 50: for the...

Yes, sz is right to talk about 'game state' rather than 'FEN' as the latter is merely one representation of the former. Nalimov's EGTs, btw, do take into account 'en passant capture' but not castling rights. I think I have a different attitude about EGTs to some other contributors, and I had it befo...

Ignoring the 50-move rule, wtm can win 40,546,725 positions: with the 50-move rule, wtm should only win 29,861,757 of these. So 10,684,968 positions are affected by the 50m-rule. Only 1,966,997 of these have DTZ > 50: for these, let's assume you focus on reducing DTZ. That leaves 8,717,971 positions...

Hope I haven't got any finger-trouble here ... wtm lost 88,147,287,576 ... 38.53% of positions: maxDTC/Z = 517 (24 positions), average DTC/Z = 49.59 (!) wtm drawn 49,153,557,962 ... 21.49% wtm won 91,451,411,502 ... 39.98%: maxDTC/Z = 32 (1 position), average DTC/Z = 1.13 btm lost 12,758,649,652 ......

My first goal as a test-case and benchmark for what perfomance to expect will be to use the 7-men disk-storage scheme and in-RAM processing code to generate a 6-men tablebase for KQ*Q*KN!N!, not making use of the exchange symmetry of like pieces. Here N! means an invulnerable Knight, and Q* means a...

I would like to understand more clearly what people have in mind when they say 'make the move-counter part of the index', just for the avoidance of misunderstanding. Do they mean, in effect 'extend the FEN to include the move-counter'? I suspect so. This would appear to multiply the number of positi...

ernest,

Marc had/has a gtbgen code which certainly could produce DTZk EGTs (and purportedly DTCk and DTMk EGTs though I don't know how it did that).
John Tamplin used it to produce DTZ50 EGTs for all 3- to 5-man and 6-man P-less endgames.
g

I noted that, where DTZ >= any DTZ that can occur in a subgame, dr = dzr = dz. For KRBNKQN, we have maxDTZ = 17 (for KBNKQN), 40 (KRNKQN), 26 (KRBKQN), 29 (KRBNKQ) and 01 (KRBNKN): successor-maxDTZ = 40. Thus, for dz >= 40, we know dr = drz = dz directly. The most important effect is that, with posi...

... thanks mb - good to see you online again - g

I think I have explained as clearly as is possible: - how the ER EGT to the DTR/DTZR metric is created: like other EGTs, it does not need to recognise move-count information, - how the ER EGT should be used, given that minimaxing-paths to current-DTR-targets can be hidden by longer paths to lower DT...

a reply primarily to hgm ... notice that I am not lazily recycling the "DTR is no good!" thread title, which I consider inappropriate and poorly informed as well as over-theatrical :-) DTR is not so much based on 'philosophy' but on the simple observation that, in the presence of a k-move ...

Notnale has hit the - nail - squarely on the head. Actual chess players show fallibility in many different ways. While the EZ50 EGT to the DTZ50 metric is necessary and sufficient to win from a position that should be won under the 50-move rule, it is specific to a 50-move rule, one which I don't re...

Did Marc Bourzutschky ever publish any stats for the Konoval-MB KRBNKQN DTC EGT? If so, does anyone have them?
g

The 'ZRd' part of the strategies discussed above only applies when we are considering a move to a position with the same DTR as the current position. If the DTR of the considered position is higher, we should not be considering the position anyway. If the DTR of the considered position is lower, tha...

hgm's association of DTM with the L_1 norm and of DTR with the L_infinity norm is insightful - but unfortunately, I don't think it's particularly helpful, although it would be very elegant if it were. More simply, we can say that DTM and DTR are 'about' the whole winning line, not just the current p...

As John Nunn and others have pointed out, the raw EGT statistics can be most misleading. There are, e.g., a lot of positions where the defender will simply capture a piece that is 'en prise' or where some 'simple tactic' is available. The endgame only becomes a 'typical position' if it is a deeper w...

... to kronsteen's remark that 'stm loses is a full-point mzug'. Not in fact: more commonly, the position is drawn if a null-move is played. It is true that there are no full-point (i.e. loss-win-loss) zugs in KQRKQR: as was long suspected, there are none in 6-man P-less chess (althought this is 'on...

hgm also described a scenario in which the defender dangled an unachievable DTR-objective in front of the attacker. As sz points out, it is not even necessary for the defender to do this. I didn't notice hgm's scenario first time round as hgm talked about the defender making a 'sub-optimal move'. We...

I am wondering if hgm's proposed approach to 7-man EGTs will generate correct EGTs for less than 7 men. If so, its correctness may be verified against existing Nalimov DTM EGTs which are [reasonably] assumed to be correct. I'm wondering how generic and portable the concepts are: the more they are, t...

... see post under the thread prematurely entitled 'DTR is no good'... g

As szyzygy correctly pointed out (for which I am most grateful), the DTR == {(DTR, DTZR)} EGT cannot be used as easily as I anticipated, even though DTR and DTZR are well-defined and can be generated for all positions. The difficulty is that the EGT may offer a route to a sharper DTR-goal which is, ...

I have lists of all known 3- to 6-man zugs on my machine, but haven't finished reviewing and summarising them for general release yet. The work was done (with me turning the handle) with Eiko Bleicher's Java program running on his copy of the Nalimov DTM EGTs. Try these for KQRKQR - '0 n' means 'los...

This board was started by Kirill Kryukov, 'KK'. Quite good initials for the Chess Endgame, I think :-). The EGTs produced so far have not considered castling rights. Indeed, no game has retained any castling rights as far as the 49th move, but there have been castlings on the 48th. So the motivation...

http://kirill-kryukov.com/chess/discussion-board/viewtopic.php?f=6&t=3083 ... KK's thread on an 'EGTB Test Suite' is perhaps what KK had in mind. KK, even though obviously everything is a draw, focuses on topics like: - the number of physically different positions: what is a 'biggest figure' (wt...

g