Who wants to find a new 17 given sudoku? And maybe a 16?

Lunatic · Posted: Sun May 24, 2009 6:32 pm Post subject:

EDIT: (in darkred)
A 'batch' of sudokus is just a file of sudoku data generated with Create_New_BatchV7.exe to be processed with one execution of BBSolver. I have no idea on how many 17 sudokus there are in the 'gap' between two puzzles. At the same time, Create_New_BatchV7.exe creates a file (CursorX.txt, where X is an increasing number) to keep in memory the cursor's position between two known sudokus for next time when another batch is created.
_________________
Marc
~~~<><~~~<><~~~<><~~~<><~~~

coloin · Posted: Mon May 25, 2009 7:55 am Post subject:

The " gap" between the puzzles consists of min lex 17-templates and clues iterated over these templates.

If there are fewer templates "between" the puzzles - a search might be quicker in some cases.

Edit......On looking at the minlex 17-puzzles.....the gaps look impossibly large. Also [probably] only a few of the gaps will contain a new puzzle.

C

coloin · Posted: Mon Jun 01, 2009 10:40 pm Post subject:

No sign of letting up, still gsf is getting 20 new 17s a day......hmmm.

I have persevered with fixing 8 clues..........and a new 17 [#48306] came out of a batch of 150 17s.... [with my single core]

Either lucky or something in it !

I have no doubt that making 18s with a knockdown process preferentially gets us into regions near non-remote 17s.

By generating 18s with a {-2+2}, but with 8 clues fixed, it perhaps gives a more random selection of 18s to do a final unfixed {-2+1}.

Next batch completes soon.......EDIT......300 17s, of which none were new....none of the last 400 "new" puzzles were even found.

So the {-2+2} isnt an improvement on getting the remote 18s.

C

humble_programmer · Posted: Fri Jun 12, 2009 5:28 pm Post subject:

gsf · Posted: Sat Jun 13, 2009 2:08 am Post subject:

gsf · Posted: Thu Aug 06, 2009 6:12 am Post subject:

the minlex gaps in the 17s catalog can be organized by common prefixes when the entries are sorted

consecutive entries will contain a common prefix of one or more clues followed by the remaining positions

a gap search fills in the remaining positions to make 17 clues and then checks if the 17 is valid

a complete gap search checks all (minlex) combinations of (17 - sizeof(prefix)) clue past the prefix for valid 17s

once a complete search is done on a gap it never has to be searched again

here is a table of (17-sizeof(prefix)) vs # of gaps
e.g., 5 2077 means there are 2077 gaps that have a common prefix of 12=(17-5) clues
the table does not take into account that a larger gap may contain smaller gaps

ChPicard · Posted: Fri Aug 07, 2009 10:28 pm Post subject: Command line

gsf · Posted: Sat Aug 08, 2009 4:19 am Post subject: Re: Command line

ChPicard · Posted: Sat Aug 08, 2009 9:22 am Post subject: Re: Command line

ChPicard

gsf · Posted: Mon Aug 10, 2009 4:36 am Post subject: Re: Command line

gsf

coloin · Posted: Fri Aug 21, 2009 7:45 pm Post subject:

Search still ongoing I see ....