The New Sudoku Players' Forum

[champagne]

Hi all,

Blue made huge steps in the search of "17 clues puzzles" in a solution grid.

See http://forum.enjoysudoku.com/bands-and-low-clue-puzzles-t30218-120.html

Waiting for the code (likely as others), I worked on my side on some issues of the process explained by blue.

One preliminary step is the generation of the UAs required to feed the elimination of puzzles tested.

I applied a process suggested by mladen Dobrichev and got surprisingly good results that I share here.
Blue's generation could perform better, but this will be seen in due time.

As in a former old thread, completely obsolete, I prepared a description of the process here.

The code is included in a zip file code.

The zip file contains a source directory "src" and a "project definition" directory.
The code has been prepared and compile using Microsoft visual studio 2013

The UAs generation for one of the 300 000 000 entries to consider in this case on my i7 is somewhere in between 1 ms and 6 ms depending on the number of band 3

I run the code (as attached) with one lexically minimal solution grid

123456789
457189623
986327415

235874196
614293857
879615342

392741568
561938274
748562931

Collecting uas with at most 4 digits, a very efficient set of uas as explained in the attached document, it took less than 2 milliseconds to produce

95 UAs located in bands 1+2, (excluding uas located in one band only)
167 ua2s
55 ua3

Just analyzing hits of cells assigned in bands 1+2 on this set of uas more than 999/1000 of blue's primary puzzles are requiring deeper investigations.

[champagne]

I just lock this post for future description of variations around blue's solution
This should not come before blue's release of the code

After flash remarks from mladen Dobrichev through pm, I made small changes in the first post and in the attached text.