Talk:N-queens minimum and knights and bishops: Difference between revisions
Talk:N-queens minimum and knights and bishops (view source)
Revision as of 18:16, 13 May 2022
, 1 year ago→Simple solution for bishops: more optimisations
(→Simple solution for bishops: Thanks to Pete!) |
(→Simple solution for bishops: more optimisations) |
||
| (One intermediate revision by the same user not shown) | |||
Line 107:
::::Do you think it's worth even bothering with bishops when we know from general reasoning that the answer will be 'n' for an 'n x n' board and its trivial to construct an example?
:::::Well, yeah, of course it is, though I'll grant you not very exciting. Simply only ever place bishops in the middle column, either by testing for that or even better yet changing the column loop. --[[User:Petelomax|Pete Lomax]] ([[User talk:Petelomax|talk]]) 00:39, 13 May 2022 (UTC)
::::Incidentally, I'm working on a GLPK wrapper for Wren as I'd like to have something in the LP/MIP solver area which can at least solve 'small' problems relatively quickly. --[[User:PureFox|PureFox]] ([[User talk:PureFox|talk]]) 17:45, 12 May 2022 (UTC)
Line 114 ⟶ 115:
:First queen on a 10x10 now limited to 9 places (was 28), first knight for n>=3 now limited to 2 places (was 3).
:Time for all solutions to 10x10 dropped from 12 mins to 3 mins (almost all being 10N examining 21,801,024 positions). --[[User:Petelomax|Pete Lomax]] ([[User talk:Petelomax|talk]]) 14:12, 11 May 2022 (UTC)
::Found a cheeky little trick (aka a circumstantial optimisation) which made the tricky 10N only examine 8,163,658 positions, and the total time dropped to 1 min 20s, plus I resurrected cheat mode since that now drops it to 8.3s all in up to 10x0. --[[User:Petelomax|Pete Lomax]] ([[User talk:Petelomax|talk]]) 18:16, 13 May 2022 (UTC)
==Some Observations==
| |||