As proposed in this thread,
http://forum.enjoysudoku.com/high-density-files-for-solution-grids-and-18-clues-puzzles-t42669-60.html
I think that a good competitor for this is a puzzle shown as
The rank 1- 5 472 730 538 of the solution grid
A binary field of 81 bit for the active cells of the puzzle.
I made a test on “loki”, the first puzzle known with a tridagon pattern
- Code: Select all
57....9..........8.1.........168..4......28.9..2.9416.....2.....6.9.82.4...41.6..
576843921493261578218759436951687342647132859832594167184326795365978214729415683 solution grid
- Code: Select all
123456789457189236698372145241893657589627314736514928314968572862735491975241863 min lexical
1.3.5......71.9...69.37......1..3..75.96.73.....51.......96..........4....5...86. min lexical loki
To get the min lexical in this case, we had first to transpose the grid, then
465 978 312 rows reordering
978 645 132 columns reordering
And this solution grid has no auto morph. (expected to be 1/10000 of the solution grids)
The solution grid has the rank 3 636 310 645.
Using this 64 characters equivalence
".123456789abcdefghiklmnopqrstuvwxyzABCDEFGHIKLMNOPQRSTUVWXYZ&"
The bit field becomes
mx5sxBL13p.84Z
so, in text mode, the canonical “loki” could be
3636310645mx5sxBL13p.84Z
Easy to decompress (some micro seconds per puzzle),
Not too hard to compress (28 milliseconds in my test)
with a sorting sequence per solution grid.
and for sure, 2 puzzles having a different canonical morph are different.
note: in my current code, the auto morphs issue is not covered
I’ll make a test on the 17 clues data base to see what gives the .zip compressor on such a file
but it would be interesting to test the new data base of puzzles having the tridagon pattern.
I am missing the link to download the current status of this data base;