Some doubts going back reading this post of Blue
http://forum.enjoysudoku.com/tridagon-t45572-30.html"champagne"
Asssuming that a tridagon pattern in the solution is the following
Code: Select all
..x x..
.x. .x.
x.. ..x
x.. x--
.x. -x-
..x --.
where the last digit on the pattern in the last box is anywhere but not in the last cell
"blue"
Nothing to add to this,
this is by far the most common final status in mith's file
I counted these too:
Code: Select all
..x x..
.x. .x.
x.. ..x
x.. x--
.x. -.-
..x --x
mith's list has 2,113,220 puzzles of the first type, over 28026 "starts", and 1,629,623 puzzles of the 2nd type, over 19937 "starts".
FWIW: By big list, has 2,146,660,838 "starts" of the first type (on 1,508,823,488 solution grids), and 821,671,928 of the 2nd type (on 744,278,570 solution grids).
The grid that I listed 24 puzzles/"starts" for, has 17 of first type, and 7 of the 2nd.
==================================================================
I have doubts that the "vicinity" search can switch from the first to the second pattern. One possibility is that mith's file is already done of several seeds