Does the Lomonosov Tablebase (and others) include extra material positions such as KQQvKQQQ?

Question

Lomonosov Endgame Tablebase says it includes all endgames with 7 pieces. But does that include things like KQQ v KQQQ? At every possible position?

Such things seem highly unlikely, but theoretically possible. I recently watched a Fischer-Petrosion game that went to 4 queens in the endgame that also included a bishop v knight and some pawns. They eventually agreed to a draw.

Other unlikely things: KBBvKNNN, KQRvKQQ, KRvKBBBB. Have any of these been evaluated?

So I guess the general question would be, have all 7-or-less endgames been evaluated where at least one side has 2+ queens, 3+ rooks, 3+ bishops, or 3+ knights?

Answer 1

The Lomonosov Endgame Tablebase supports 7 pieces. That means it has all the positions you describe in your question. Otherwise, it's not a tablebase by definition.

The "ridiculous" positions you mention might not happen on the board but might still happen in the engine search. They're still important because pawns can promote, and a chess engine looks at lots of possibilities -- not just positions that actually appear on the board.

All 7-or-less endgames have been analyzed. However, some tablebase implementations leave out bare king endgames.

Answer 2

The Lomonossov tablebases include all 7-man positions with the exception of 6 pieces vs lone king.

Most material distributions you consider "ridiculous" are in fact required to evaluate "interesting" positions because of possible pawn promotions. This works recursively, so if you want to solve e.g. KPPPvKPP, you need (among others) the table KPPQvKPP first. To solve KPPQvKPP, you need (among others) the table KPPQvKPQ first, &c.

Ultimately, you need KQQQvKQQ and all other 4-man vs 3-man tables to solve the endgame KPPPvKPP. So it is necessary to have all possible material constellations, even those who seem to be uninteresting.