8-piece endgame tablebases - first findings and interview!

[Paloma]

I am not sure that you are right and I found the following mate in 549 in the 7 piece lomosonov tablebases
Is there a longer mate with 7 pieces without pawns.

Is Lomosonov back for you? Here comes Lomosonov server is offline.

[Uri Blass]

I am not sure that you are right and I found the following mate in 549 in the 7 piece lomosonov tablebases
Is there a longer mate with 7 pieces without pawns.

Is Lomosonov back for you? Here comes Lomosonov server is offline.

No but the records are already saved

[vb4]

Hi I have asked this question in a different format in this thread and although JP responded it was not exactly on track with my question. In the past when tablebases were being developed they had a stat file that looked like the following format:

wtm: Draws: 330177
wtm: Mate in 19: 16
wtm: Mate in 18: 1910
wtm: Mate in 17: 14296
wtm: Mate in 16: 31598
wtm: Mate in 15: 40497
wtm: Mate in 14: 46044
wtm: Mate in 13: 45174
wtm: Mate in 12: 37421
wtm: Mate in 11: 31838
wtm: Mate in 10: 23115
wtm: Mate in 9: 15909
wtm: Mate in 8: 10972
wtm: Mate in 7: 7254
wtm: Mate in 6: 4819
wtm: Mate in 5: 4242
wtm: Mate in 4: 2343
wtm: Mate in 3: 2017
wtm: Mate in 2: 326
wtm: Mate in 1: 824
wtm: Broken positions: 139093
btm: Lost in 0: 194
btm: Lost in 1: 134
btm: Lost in 2: 293
btm: Lost in 3: 434
btm: Lost in 4: 1346
btm: Lost in 5: 1307
btm: Lost in 6: 2750
btm: Lost in 7: 3839
btm: Lost in 8: 6741
btm: Lost in 9: 10021
btm: Lost in 10: 16567
btm: Lost in 11: 23157
btm: Lost in 12: 32371
btm: Lost in 13: 44574
btm: Lost in 14: 53182
btm: Lost in 15: 62434
btm: Lost in 16: 63229
btm: Lost in 17: 32780
btm: Lost in 18: 4498
btm: Lost in 19: 59
btm: Draws: 513732

Does anyone know if this type of file exists with the newly created 8 piece EGTB's??

Thanks in advance,

Les

[syzygy]

For me interesting is question: how long is longest forced win DTM for n-TB.
Someone may suggest: if 8-TB longest mate took 584 moves, then 9-TB can be ~ 584*2=1168 moves, 10-TB ~ 584*4=2336 moves and so on.
But i don't think so it works like than.

A self-declared "expert" on TBs has indeed suggested that every extra piece (more than) doubles max DTM based on max DTM for 3, 4, 5, 6 and 7 pieces (in the process giving someone else credit for naming this "law" after him). That would mean over a 1000 moves for 8-piece TBs, over 2000 moves for 9-pieces TBs. And I guess mate in about 2^34 moves for 32 pieces.

However, I fully agree with you that it does not work like that. Certainly the doubling has to break down long long long before you reach 32 pieces, and there is no reason why it does not break down already for 8 pieces (and it seems it does, although tables with pawns probably haven't been generated yet). That so-called "law" is simply not based on any sound reasoning. (And I am not sure why the article I linked to speaks of "confidence levels" when fitting a curve to the 5 known max DTM values for n= 3, 4, 5, 6 and 7. That makes no sense whatsoever.)

I don't think any forced win (DTM for any TB) ever exceeds 1000 moves.
Even 594 is a very big number.

But i need a patience ... to get answer.

To know if 1000 moves can be exceeded, e.g. with 12 or 13 pieces on the board, we indeed need to be patient

[syzygy]

With these 8 piece egtb's is there anyone that can point me to the stat files that show all the details like how it was done for the 3-4-5-6-7 piece EGTB's?

I am interested in these details as I have been monitoring all the other egtb files.

Thanks in advance Les

Because until now, every longest mate record came from pawnless endgames, so it is not unreasonable to assume the same holds true for 8 pieces. Of the pawnful 8-piece sets resolved and shared, none came close to challenging that 500+ move record.

The longest 7-piece mate has a pawn:
[fen]1n1k4/6Q1/5KP1/8/7b/1r6/8/8 w - -[/fen]
https://tb7.chessok.com/probe/3/61
However, this position does quickly convert into a very long pawnless mate, so I agree it seems likely that the pawnless 8-pieces tables give a good indication for the length of the longest 8-piece mate.

edit: I now see that Uri already posted the position.

[jp]

Hi I have asked this question in a different format in this thread and although JP responded it was not exactly on track with my question. In the past when tablebases were being developed they had a stat file that looked like the following format:
...
Does anyone know if this type of file exists with the newly created 8 piece EGTB's??

I see what you mean now. I was thrown reading the words "that show all the details like how it was done".

Can you give links to that info for <=7-man TBs?

As for 8-man TBs, given that only a (very) small subset of the total TBs exists, I'd guess such files aren't currently available. The undated info said that MB had completed TBs for 15% of the pawnless positions. I don't know why he thinks that'll include even the longest DTM for the pawnless positions, let alone all positions. Maybe he has some argument for that.