minimum number of clues per band/stack

Everything about Sudoku that doesn't fit in one of the other sections

Re: minimum number of clues per band/stack

Postby Serg » Sat Jun 04, 2016 8:28 pm

Hi, champagne!
champagne wrote:I am running the last test before the start of the 27+6+3 case

I have 27692 ED patterns after magic 40 against 61079 found by Serg with no filter
I confirm 27692 ED patterns of the 27+6+3 case after "40 maximal patterns list" filtering. (I attached my list of filtered patterns to this post.)
Code: Select all
Distrib. Patterns   Filtered

3+6+27     61,079     27,692
4+5+27     90,897     52,267
3+7+27    155,054     77,289
4+6+27    277,435    210,116
5+5+27    169,258    134,467
5+6+27  1,056,939    977,292
Total   1,810,662  1,479,123
You can see filtering efficiency isn't high in this case.

(87.2 KiB) Downloaded 36 times
2017 Supporter
Posts: 511
Joined: 01 June 2010
Location: Russia

Re: minimum number of clues per band/stack

Postby champagne » Sun Jun 05, 2016 3:14 pm

Hi Serg,

Thanks for having cross checked the magic 40 effect.

In fact, as we can see in your table, the effect is significant mainly with a band of 3 clues.

this is not very positive for the search of 17 through a 27 + X + Y scan.

The scan 27 + 6 + 3 is ongoing for 2 weeks after a bug in my first approach has been fixed. I am expecting a 3 to 4 months search. The new process is faster than the previous one in a ratio between i:50 1:100, but this is not enough to go beyond the 27+X+3.

I am looking something new as well for a search on the catalog, but it is to early to tell what can be the outcome.

The process that I am testing took ideas in that thread
2017 Supporter
Posts: 5650
Joined: 02 August 2007
Location: France Brittany


Return to General