The New Sudoku Players' Forum

[1to9only]

you mean g1r79c8 = g2r13c2 = g1r6c9,g2r4c1

yes. well spotted. i'll correct the mistake.

also, jigsaws layouts used are this:

Code: Select all

111112223144412223144452223144455333666555333666557779688857979688877999688877999
999778886999778886979758886977755666333555666333554441322254441322214441322211111


[urhegyi]

I've corrected the jigsaw definition, it now solves:

Code: Select all

3...1.....9.8...........489.7.5.....1...2............5..7........3........8..2... 111112223144412223144452223144455333666555333666557779688857979688877999688877999
...6..2........7........8..7............4...2.....4.8.943...........7.6.....2...3 999778886999778886979758886977755666333555666333554441322254441322214441322211111


Code: Select all

3...1.....9.8...........489.7.5.....1...2............5..7........3........8..2... ED=3.2/1.2/1.2
...6..2........7........8..7............4...2.....4.8.943...........7.6.....2...3 ED=2.6/1.2/1.2

385416927294875136621357489976543812169728543832194765417269358753681294548932671
358619247294536718671293854789462135837145692526374981943781526412857369165928473


Yes the rating 3.2 confirms the need of an x-wing. I found one in grid 2 when solving manually yesterday.(R13C16 on 3's in columns)

[1to9only]

I've created another pre-release SudokuJigsawExplainer: here.
To solve/rate a twodoku:

Code: Select all

java.exe -Xrs -Xmx500m -cp SudokuJigsawExplainer.jar diuf.sudoku.test.twodoku --input=puzzle.txt [options]


I'll do a slightly longer post a bit later: here to explain the options.

[urhegyi]

Next level of difficulty. How to translate this to a definition of the 2 grids?

20070818_s2_THINKER_u84_c32_0.png (5.03 KiB) Viewed 765 times

solution:

Hidden Text: Show

20070818_s2_THINKER_u84_c32_0-s.png (13.72 KiB) Viewed 755 times

[1to9only]

These are nice overlapped jigsaws, but SudokuExplainer is for 9x9 sudokus which I've tweaked to solve other variants of 9x9 sudokus including overlapped ones. But these jigsaws are no longer 9x9, having cells that stick out!

An approximation for JigsawExplainer could be this - solving the jigsaw is not guaranteed:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. 111223333111122223142223333145555666444555666444475866999975868999777888997787788
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... 887787799888777999868579999668574444666555444666555541333322241322221111333322111


Edit: Corrected jigsaw definition. See further below: here.

I've not tried these in JigsawExplainer.

This latest twodoku is NOT like the previous twodokus posted (aka sensei-1) with 1 nonet overlap. It is a sensei-2 with 2 nonets overlap. So the 'twodoku' solver will not solve this.

[urhegyi]

This is a better example:

738b4710b912c8fc897cbb32fb039245d7882145.png (62.94 KiB) Viewed 748 times

[Hajime]

Next level of difficulty. How to translate this to a definition of the 2 grids?

20070818_s2_THINKER_u84_c32_0.png

solution:

Hidden Text: Show

20070818_s2_THINKER_u84_c32_0-s.png

Because the two grids must have 9 * 9 digits each, you can determine (proof?) that r9c3 and r7c4 must be equal and change the borders of the Jigsaw a bit to match the border of the grid. Also for r1c7 and r3c6 of grid 2. Now SiSeSuSo must be able to solve this....
And "no" , SiSeSuSo chokes on it

[Hajime]

This is a better example:

738b4710b912c8fc897cbb32fb039245d7882145.png

Can you add the puzzle in code-lines?

[1to9only]

This is a better example:

738b4710b912c8fc897cbb32fb039245d7882145.png

Can you add the puzzle in code-lines?

Code: Select all

4.5...8.1..8...4.228.7........8.9.........78.8..6.7....1.58..37....7.5.8357..8... 111112222311144222331444522333444555633477555663777755688877999688887999666688999
.3789..515.8.3........85.73...1.3....23.........5.9.3...53...94351...9..8.....3.5 999886666999788886999778886557777366555774336555444333225444133222441113222211111


From the number of clues, I think this would be Easy!

[Hajime]

This is a better example:

738b4710b912c8fc897cbb32fb039245d7882145.png

Can you add the puzzle in code-lines?

Code: Select all

4.5...8.1..8...4.228.7........8.9.........78.8..6.7....1.58..37....7.5.8357..8... 111112222311144222331444522333444555633477555663777755688877999688887999666688999
.3789..515.8.3........85.73...1.3....23.........5.9.3...53...94351...9..8.....3.5 999886666999788886999778886557777366555774336555444333225444133222441113222211111


From the number of clues, I think this would be Easy!

Yes, per grid unsolvable, but the puzzle can be solved with Pointing/Claiming only.
So about SER=2.8

Hidden Text: Show

Code: Select all

#2//B4,JS/H16,JS
465923871578136492283714965746859123139245786821697354914582637692371548357468219
111112222311144222331444522333444555633477555663777755688877999688887999666688999
637894251548936127219685473472153689923478516186549732765321894351762948894217365
999886666999788886999778886557777366555774336555444333225444133222441113222211111


[urhegyi]

These are nice overlapped jigsaws, but SudokuExplainer is for 9x9 sudokus which I've tweaked to solve other variants of 9x9 sudokus including overlapped ones. But these jigsaws are no longer 9x9, having cells that stick out!

An approximation for JigsawExplainer could be this - solving the jigsaw is not guaranteed:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. 111223333111122223142223333145555666444555666444475866999975868999777888997777888
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... 888777799888777999868579999668574444666555444666555541333322241322221111333322111


I've not tried these in JigsawExplainer.

This latest twodoku is NOT like the previous twodokus posted (aka sensei-1) with 1 nonet overlap. It is a sensei-2 with 2 nonets overlap. So the 'twodoku' solver will not solve this.

This should be working:
grid1:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. 111223333111122233114222333445526666444555666444555566999778888999777788999787788

grid2:

Code: Select all

.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... 778888999777788999787788999665555444666655444666555444111122333112222333111222333

Can any of you try this? I'm interested if SiSeSuSo and/or JigsawExplainer can solve this.

[urhegyi]

My latest version of SiSeSuSo of November without the latest bugfixes can only solve it with brute force.

[1to9only]

To obtain the jigsaw definition for grid 1:

Code: Select all

111223333
111122223
142223333
145555666
444555666
444475866
999975868 8.
999777888 ..
9977A7B8C ..

   ..7.8....
   .........
   .........
   .........
   .........
   .........


I previously posted cells ABC=788 - this is incorrect. ABC=878 leads to the twodoku being solved.
For grid 2, its version of ABC=878 also.

The corrected jigsaw definitions:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. 111223333111122223142223333145555666444555666444475866999975868999777888997787788
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... 887787799888777999868579999668574444666555444666555541333322241322221111333322111


The rating:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. ED=1.9/1.2/1.2
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... ED=2.6/1.0/1.0


The solution:

Hidden Text: Show

Code: Select all

523647189461785932892173456917326548386459271754261893249518367178934625635892714
518367429934625178892714356647192583251938764783456291465873912126549837379281645


[Hajime]

My latest version of SiSeSuSo of November without the latest bugfixes can only solve it with brute force.

SiSeSuSo_20201217_211039.jpg

Indeed only solvable with bruteforce by SiSeSuSo, but...
there are multiple (more than 10) solutions.
Eg

Code: Select all

#2//B4,JS/E16,JS
597346182426185973382761459961427538873659241154273896249518367718934625635892714
111223333111122233114222333445526666444555666444555566999778888999777788999787788
518367429934625178892714356647192583251938647783456291465873912176249835329581764
778888999777788999787788999665555444666655444666555444111122333112222333111222333

Code: Select all

#2//B4,JS/E16,JS
596147283721385946482763159913426578867259431354671892249518367178934625635892714
111223333111122233114222333445526666444555666444555566999778888999777788999787788
518367249934625178892714356746293581251978634683451792465832917327149865179586423
778888999777788999787788999665555444666655444666555444111122333112222333111222333


[Hajime]

To obtain the jigsaw definition for grid 1:
The corrected jigsaw definitions:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. 111223333111122223142223333145555666444555666444475866999975868999777888997787788
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... 887787799888777999868579999668574444666555444666555541333322241322221111333322111


The rating:

Code: Select all

....4........8......2....5.......5..........1.54....9.24..18.6...8......6...9.71. ED=1.9/1.2/1.2
.18.6...9......1...9.71..56.4....58.2..........3.......6....9......4........8.... ED=2.6/1.0/1.0


Indeed, with the bugfix (not yet in the november version), SiSeSuSo solves this with
Pointing/Claiming only in both grids and no BruteForce.
PS1. Busy implementing (UVW)XYZ method, than a new relaese can be expected.
PS2. Also busy for a nice Xmas puzzle like http://forum.enjoysudoku.com/81-candles-t35567.html