| Ruud Site Admin
| Joined: 17 Sep 2005 | Posts: 708 | : | Location: Netherlands | Items |
|
Posted: Sun Oct 16, 2005 12:28 pm Post subject: Colouring: Minimum number of candidate cells |
|
|
I'm trying to figure out what the minimum number of candidate cells for a digit would be, to avoid useless colouring.
Here's what I've figured out so far:
When a digit is bound to 9 cells in the grid, that digit is completely solved.
There is no situation with 8 bound digits, because the last one would be a single, and therefore bound.
With 7 bound digits, the 4 candidates will form an X-wing shape, because there are only 2 rows and 2 columns that allow the remaining digits. Colouring would be useless.
With 6 bound digits, there can be between 6 and 9 candidates for the remaining 3 digits. With less than 6 candidates, at least one of the digits will be a hidden single. However, I can see no scenario where colouring would be able to eliminate candidates.
So this is my provisionary minimum:
1. At least 4 unbound digits
2. At least 9 candidate cells (4 pairs + 1 extra)
I would appreciate an expert opinion. |
|
| Lummox JR
| Joined: 07 Sep 2005 | Posts: 202 | : | | Items |
|
Posted: Mon Oct 17, 2005 12:03 am Post subject: |
|
|
I believe you may be correct. Coloring with 3 unbound digits is certainly impossible, but with 4 you can find a workable pattern. It did also require 9 cells for me to build a pattern.
Code: | . . .|. . .
a . .|. . A
. A .|a . .
-----------
A * .|A . .
. A .|. . a
. . .|. . . |
With 8 cells you can only end up with singles or pointing pairs, or a completely consistent set of choices with nothing else to eliminate. With fewer the same problems all apply. |
|