16_17 clues puzzles search in the catalog

http://pagesperso-orange.fr/gpenet/XY27/cat17.htm

The proof that no Sudoku with 16 clues can be produced has been published in august 2013 by  Gary McGuire .

The corresponding article should still be available at the following location

 

http://arxiv.org/pdf/1201.0749.pdf

 

That process was very costly in computer resources and as of to-day, it seems impossible to extend it to a proof that all 17 clues puzzles are known (or to produce missing ones.)

 

 However, many pieces of that process have a wider scope. I’ll refer to it in several places of that document. The reader should have downloaded locally.

 

Several other ways to achieve the task have been tested, but nothing so far showed up having the required performance to make a full scan of the 17 clues field.

 

Here is a new attempt to do that trying to merge some pieces of logic developed in that forum in several threads and mainly in the followings

 

http://forum.enjoysudoku.com/minimum-number-of-clues-per-band-stack-t32120.html

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

 

These 2 main threads have links to other key threads.

 

The following topics are developed here after :

 

A. The scan logic proposed.

B. Some enumeration data

C. Formatting the catalog in the appropriate way

D. Finding missing UAs  when the primary list is exhausted

E. Building the primary lists of UAs 

F. Choosing the cells priority in the scan of bands 1 2

G. What is done when the primary list of UAs is covered.