Construction puzzle ( 9 queens for both sides )

Sort:
Avatar of watcha

a) Is it legal for both sides to have 9 queens ?

b) If it is legal:

  1. what is the maximum total number of pieces ( ie. knights, bishops, rooks and queens ) that can be on board in such a position ;
  2. what is the shortest possible way to reach such a position ;
  3. construct a game in which both sides have 9 queens but the position is balanced and playable for both sides ( ie. respected engines would value it between -0.5 and +0.5 ).
Avatar of Remellion

It is legal, and the minimum required captures for pawns to pass are 8, so there can be 2 kings, 18 queens and 4 other pieces at most.

Construction of a proof game is absolutely tedious though.



Avatar of watcha
Remellion írta:

It is legal, and the minimum required captures for pawns to pass are 8, so there can be 2 kings, 18 queens and 4 other pieces at most.

Construction of a proof game is absolutely tedious though.

You did it in 57 moves! Good job! Is your position equal? When I first succeeded it took 70 moves and the position was actually mate in 1 :).

This is my first succesful attempt:



Avatar of Remellion

Oddly enough, my version has the unique 57...Qfd1# mate in 1 despite 3 queens watching that square. No other move suffices.

Very likely I could lower the 57 moves by shifting the kings to safety first, e.g. 24. Kf2(-Kg3-g5-Kg4). Move efficiency comes with having a plan from move 1, and my plan didn't consider the kings and checks messing up with the speed.

Arithmetically, you need at least 11 ply per pair of same-coloured pawns. So that's 88 ply already. At least 8 pieces need to suicide themselves, which needs around ~12 ply (guesstimate, and is what my construction had.) Probably another 6 ply to account for the kings running. So I (estimate) a lower bound of 108 ply = 54 moves to be necessary. Which implies that my construction can be optimised further via the king moves, and not much else.

As for the equal position thing, After 81...Qgg6 I wonder how even it is. Looks fairly even.

Avatar of Remellion

Here, with nothing but K + harems on the board, 53 moves (105 ply). Evidently my arithmetic was wrong. And there is still slightly more room for optimisation, particularly since white needed 2 more ply than black to finish.

Avatar of BigDoggProblem

Ed Collins' site has the 18 Q task done in 48.0 moves.

http://www.edcollins.com/chess/queens.htm

Avatar of watcha
Remellion írta:

Here, with nothing but K + harems on the board, 53 moves (105 ply). Evidently my arithmetic was wrong. And there is still slightly more room for optimisation, particularly since white needed 2 more ply than black to finish.

One thing is absolutely certain: you have to queen all pawns. Since it takes at least 5 moves to queen a pawn this means that 80 plies is the absolute minimum. The only question is how many extra moves you have to make beyond 80 plies and what sort of moves.

Your solution has the following extra moves:

white piece move 5
  king move 4
  queen move 4
  +pawn move 0
black piece move 5
  king move 2
  queen move 5
  +pawn move 0
  total extra moves 25
  minimum pawn moves 80
  total 105
Avatar of watcha
BigDoggProblem írta:

Ed Collins' site has the 18 Q task done in 48.0 moves.

http://www.edcollins.com/chess/queens.htm

Ed Collins' second solution ( which visually looks the best ) has the following moves:

white piece move 4
  king move 2
  queen move 2
  +pawn move 0
black piece move 4
  king move 3
  queen move 1
  +pawn move 0
  total extra moves 16
  minimum pawn moves 80
  total 96