The New Sudoku Players' Forum

[Pat]

The whole thing completed in about 8 hours.

with over 40,000 17-clue puzzles now,
could we run the MegaClue test
on the new puzzles only?

or would we have to re-run it on the entire set of 17s?

( the old puzzle-numbers are no longer valid;
the good news is that the new puzzle-numbers will remain valid )

~ Pat

[gfroyle]

could we run the MegaClue test
on the new puzzles only?

Do you mean the counts of "numbers of solutions of the puzzles with one clue deleted"?

For all of the 17-clue puzzles, I've calculated the numbers of solutions of each of the 17 puzzles obtained by deleting one clue in turn, and incorporated them into my database.

So if this is what you mean, then I can easily calculate any particular statistic you would like from these figures...

I have not yet put a proper front end on so that these stats are accessible easily, but here are some examples..

Examples:

Most mega of clues: Clue # 3 from Puzzle #15586 leading to 11339281 solutions

[Clue #3 means the third clue in the MIN LEX version of puzzle #15586, not the downloadable "sudoku17" file]

[NOTE: At the moment the only way you can easily pull this out of the DB is to get the PDF from http://people.csse.uwa.edu.au/gordon/sudoku17pdf.php?id=15586 which is not really suitable for programmers]


Least mega of clues: 2 solutions obtained in lots of ways

We can play lots of min-max games

(1) Max max megaclue: which puzzles have overall maximum solutions among their 17 1-clue deletions

Code: Select all

+-------+----------+
| id    | maxmax   |
+-------+----------+
| 15586 | 11339281 |
| 21355 | 10789663 |
| 33611 |  9370174 |
| 33864 |  9239041 |
| 33920 |  8587343 |
|  8104 |  8159472 |
| 32773 |  8159472 |
| 39852 |  7856752 |
| 15578 |  7547833 |
| 19861 |  7487776 |
+-------+----------+

(2) Min max megaclue: Which puzzles have the minimum value of the maximum solutions over all 17 1-clue deletions

Code: Select all

+-------+--------+
| id    | minmax |
+-------+--------+
| 20515 |   8503 |
| 28773 |  14437 |
| 22111 |  15656 |
| 22731 |  16741 |
| 31403 |  17171 |
| 10832 |  18938 |
| 22734 |  19005 |
| 17025 |  19155 |
|  1158 |  19450 |
| 32506 |  20040 |
+-------+--------+


(In other words, 1-clue deletions of puzzle #20515 have AT MOST 8503 solutions, no matter which clue you remove)

(3) Max min megaclue : which puzzles have the maximum value of the minimum number of solutions over all 17 1-clue deletions

Code: Select all

+-------+--------+
| id    | maxmin |
+-------+--------+
|  3673 |  81469 |
| 12919 |  43983 |
| 20858 |  37293 |
| 33503 |  31584 |
| 36412 |  29545 |
| 29019 |  25522 |
|  3832 |  25248 |
| 14078 |  23961 |
| 28992 |  21953 |
| 29018 |  21586 |
+-------+--------+

In other words, 1-clue deletions of puzzle #3673 have AT LEAST 81469 solutions no matter which clue is deleted..



Can do averages, standard deviations etc etc, but not now - I'm off to bed.

Cheers

Gordon

[Pat]

thanks, gfroyle, that's great

as a spinoff from the MegaClue topic,
the large Unavoidable Sets
produced great progress in:
Greatest divergance possible in dual solution puzzle?

[coloin]

If one removes a single clue from a minimal 9x9 sudoku, the minimum number of solutions is 2 ... but what is the maximum?.......

We seem to have this sorted for the 17 clues.....how about for a minimal 25 clue puzzle ?

Redirected from another thread here is my initial offering

Code: Select all

+---+---+---+
|.2.|...|...|
|...|...|.6.|
|687|..9|.34|
+---+---+---+
|..3|9..|621|
|.7.|..2|...|
|...|5.8|..7|
+---+---+---+
|..2|.1.|..3|
|...|..3|8..|
|.4.|...|..6|
+---+---+---+  omitting the 1 at r7c5 gives 6040 grid solutions

Can anyone do better...?

C

[JPF]

quick search : 25 clues minimal

Code: Select all

 *-----------*
 |.2.|...|...|
 |...|...|.6.|
 |6.7|..9|.34|
 |---+---+---|
 |..3|9..|621|
 |.7.|..2|...|
 |...|5.8|..7|
 |---+---+---|
 |..2|.1.|..3|
 |...|..3|8..|
 |.48|...|..6|
 *-----------*   omitting the 1 at r7c5 gives 10855 solutions



but one can do much better.
Why 25 clues ?

JPF

[JPF]

still 25 clues, minimal :

Code: Select all

 *-----------*
 |.2.|...|...|
 |.3.|...|.6.|
 |687|..9|.34|
 |---+---+---|
 |8..|9..|62.|
 |.7.|..2|...|
 |...|5.8|..7|
 |---+---+---|
 |..2|.1.|..3|
 |...|..3|8..|
 |.4.|...|..6|
 *-----------*    omitting the 1 at r7c5 gives 12845 solutions.


JPF

[coloin]

why 25 clues ?

It was in response to mauricio's post here
Which is some way ahead.
C

[coloin]

My attempt was made on a random selection of puzzles.....but I can see how you improved it.

I have made similar changes to mauricio's puzzle.....which shows we do have much furthur to go.

I was going to leave it for a bit, but I couldnt resist the latest "nice trick"
Here is the complete 25 clue minimal puzzle

Code: Select all

+---+---+---+
|.3.|...|...|
|...|..3|.4.|
|.24|..5|37.|
+---+---+---+
|...|..2|7..|
|.18|..7|.6.|
|...|1.6|.5.|
+---+---+---+
|.6.|...|487|
|.8.|...|.3.|
|1..|9..|...|
+---+---+---+

omitting the 9 @r9c4 gives 114528 grid solutions

C

[gsf]

My attempt was made on a random selection of puzzles.....but I can see how you improved it.

these options to my solver will stream through a collection of puzzles and list puzzles with
megaclue count > the previous listed puzzle

Code: Select all

-qFN -f'%v # %(required(max))x' -e 'max(required(max))'


[JPF]

Code: Select all

 *-----------*
 |.3.|...|...|
 |...|..3|.4.|
 |.24|..5|37.|
 |---+---+---|
 |...|..2|7..|
 |.18|..7|.6.|
 |...|1.6|.5.|
 |---+---+---|
 |.6.|..1|4.7|
 |.8.|...|.3.|
 |1..|9..|...|
 *-----------*  omitting the 9 @r9c4 gives 138142 grid solutions



etc...

JPF

[coloin]

We are obviously using the same technique !

Code: Select all

155323      .3............3.4..24..537......27...18..7.6....1.6.5..6......7.8....43.14.......
145563      .3............3.4..24..537......27...18..7.6....1.6.5..6...1..7.8.....3.14.......
138142      .3............3.4..24..537......27...18..7.6....1.6.5..6...14.7.8.....3.1........
131767      ...........1..3.4..24..537......27...18..7.6....1.6.5..6....48778.....3..........
127763      .3............3.4..24..537......27...18..7.6....1.6.5..6......7.8....13.14.......

I wonder if this furthurs the max clues exercise....

C

[Mauricio]

I wonder if those 24 clue puzzles can be part of a high clue sudoku (>36). I think that due to the fact that it all started with this sudoku

Code: Select all

Ravel's 38 -14 clues +1 clue. Minimal.
+-------+-------+-------+
| . . . | . . . | . . . |
| . . . | . . 3 | . 4 . |
| . 2 4 | . . 5 | 3 7 . |
+-------+-------+-------+
| . . . | . . 2 | 7 . . |
| . 1 8 | . . 7 | . 6 . |
| 2 . . | 1 . 6 | . 5 . |
+-------+-------+-------+
| . 6 . | . . . | 4 8 . |
| . 8 . | . . . | . 3 . |
| 1 . . | 9 . . | . . 6 |
+-------+-------+-------+
25 clue minimal puzzle - 9 @ r9c4 = 32213 grid solutions


In the remaining 24 clue subpuzzle, you can add 14 clues minimally. Can this be possible too with the given subpuzzles?

[Havard]

In the remaining 24 clue subpuzzle, you can add 14 clues minimally. Can this be possible too with the given subpuzzles?

A -1/+14? I would not get my hopes up. Even though you can bail out any time the puzzle reaches a solution before all 14 are added, I think the searchspace would be far to big.

Havard

[coloin]

the search space would be far to big

I initially was of this opinon

Except......
The search options are reduced by
Never insert a 9 clue
Never insert more than 6 of any of the other clues [1-8]

Having said that I cant make a single minimal puzzle from any of these subpuzzles, so I think its not going to work.

Minimality is [as usual] the difficult issue and therefore all the clues need to have high solution counts, these individual counts all progressivly reducing with each additional clue.

C

[Mauricio]

I think the searchspace would be far to big.

I think that too, I was just thinking that it could be possible that those subpuzzles may be part of a larger minimal puzzle. Obviously trying to actually find those puzzles exhaustively is not an optimal way to find minimal puzzles with high number of clues.

In the meantime I started with this subpuzzle coloin posted:

Code: Select all

. 3 .|. . .|. . .
. . .|. . 3|. 4 .
. 2 4|. . 5|3 7 .
-----+-----+-----
. . .|. . 2|7 . .
. 1 8|. . 7|. 6 .
. . .|1 . 6|. 5 .
-----+-----+-----
. 6 .|. . .|. . 7
. 8 .|. . .|4 3 .
1 4 .|. . .|. . .
coloin says it has 155323 grid solutions, I have not checked that fact. 24 clues


Doing a nested 1off/2on I arrived to:

Code: Select all

Minimal, coloin subpuzzle + 10 clues. 34 clues.
. 3 .|4 . .|5 . .
. . 1|. 7 3|. 4 .
. 2 4|. 1 5|3 7 .
-----+-----+-----
. . .|. 8 2|7 . .
. 1 8|. . 7|. 6 .
2 . .|1 . 6|. 5 .
-----+-----+-----
. 6 5|. . .|. . 7
. 8 2|. . .|4 3 .
1 4 .|. . .|6 . .

My resources to do that were not optimal, and I think more clues are possible.

Addendum: I found this, but I am stuck at 35 clues at the moment

Code: Select all

. 3 .|. . .|5 . .
. . .|2 . 3|. 4 .
. 2 4|. . 5|3 7 .
-----+-----+-----
. . .|3 8 2|7 . .
. 1 8|. 5 7|2 6 .
. . .|1 . 6|. 5 .
-----+-----+-----
. 6 .|. . .|1 . 7
. 8 7|6 . .|4 3 .
1 4 .|. . .|6 2 .