The New Sudoku Players' Forum

[urhegyi]

Can you check the generation of JSX with the Wednesday Layout in the Jigsaw list?
Run a taskmanager and when processor use falls to zero the program often terminates unexpectedly and shows nothing.
Try generating with option "show the analysis of the generated sudoku jigsaw" enabled or disabled and compare. Try also with fiendish/diabolic as difficulty and also maximum/exact difficulty.
Solutions are possible I can generate them with other programs but not with JigsawExplainer.
Examples:

Code: Select all

....7......7.56..273.................6......3..1.9.63.1.67.3.9...4..1.........9.. 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX
.5.8.2.1..........7...6...1.........9..5.4..3.........1...4...8..........1.6.8.7. 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX


Rating:

Code: Select all

....7......7.56..273.................6......3..1.9.63.1.67.3.9...4..1.........9.. 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX ED=2.6/1.2/1.2
.5.8.2.1..........7...6...1.........9..5.4..3.........1...4...8..........1.6.8.7. 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX ED=6.6/1.7/1.7


I tried it with the version before from Jigsawexplainer with the high processor and memory use and it finds a solution:

Code: Select all

........5.4.......1........7.2.........92.5..47...5....163...4...5..........4.... 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX


rating:

Code: Select all

........5.4.......1........7.2.........92.5..47...5....163...4...5..........4.... 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX ED=2.9/1.2/1.2


Are there changes in the generator in the newest version? If so, what exactly are the differences in generating approach between the 2 versions?

[1to9only]

Using the 1 Feb JigsawExplainer, I have generated Wednesdays, 2 JSs and 1 JSX in the last 10 minutes or so. So the program does work.
As to the hang, I dont know the cause. I'll consider it when I come across it. Work on JigsawExplainer is rather time limited.

[urhegyi]

Using the 1 Feb JigsawExplainer, I have generated Wednesdays, 2 JSs and 1 JSX in the last 10 minutes or so. So the program does work.
As to the hang, I dont know the cause. I'll consider it when I come across it. Work on JigsawExplainer is rather time limited.

JS causes no problems it's only with JSX!
The first I generated works, the second hangs!
I click on close. Reopen another generator dialog , the first generates correctly, the second hangs again!
When it hangs and you click on stop, the label isn't synchronized and doesn't change to "generate" anymore!
When I dont close now but just change the state of the analyze or not checkbox and click on stop again(which is an unsynchronized generate) it restarts correctly and produces a solution and change its label to the correct generate state.
But the next time it hangs again and I have to change the checkbox again and this helps anymore, so I have to close.
Update: Changed the Wednesday layout to a similar layout, JS works fine, JSX hangs!
The layout I use now to test and causes the same problem is:

Code: Select all

...4..52.952.6.4.........1.7........2.1.....5.6....159...8......1..5.94.....2.... 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX


Rating:

Code: Select all

...4..52.952.6.4.........1.7........2.1.....5.6....159...8......1..5.94.....2.... 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX ED=2.6/1.2/1.2

Can't generate any JSX.
Try now with the version before 1 feb. Will see what it does.
Update: Yes, it finds a solution:

Code: Select all

7.....8.......2....6.........8...........825................1.9.......23.475..6.. 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX

Rating:

Code: Select all

7.....8.......2....6.........8...........825................1.9.......23.475..6.. 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX ED=2.9/1.5/1.5

And one more generated by SiSeSuSo:

Code: Select all

.5.236.4...3...5...65...31.............................84...16...6...9...4.591.3. 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX

Rating:

Code: Select all

.5.236.4...3...5...65...31.............................84...16...6...9...4.591.3. 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX ED=4.4/1.5/1.5

One with xyz-wing to prove generating fiendish JSX with this layout is possible!

Code: Select all

Analysis results 
Difficulty rating: 4,4
This Sudoku Jigsaw can be solved using the following logical methods:
57 x Hidden Single
 1 x Direct Hidden Pair
 1 x Naked Single
 7 x Cage Pointing
 4 x Claiming
 1 x Generalized Intersection
 1 x Hidden Pair
 1 x XYZ-Wing


[urhegyi]

2 simple JSX layouts which give JigsawExplainer a headache when trying to generate!

Code: Select all

........5.4.......1........7.2.........92.5..47...5....163...4...5..........4.... 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX ED=2.9/1.2/1.2
7.....8.......2....6.........8...........825................1.9.......23.475..6.. 122233333112223343111223444611525444615555574666585774666988777969988877999998887 JSX ED=2.9/1.5/1.5

[1to9only]

I am aware of several bugs in JigsawExplainer, but as far as I can tell none of them has been reported here!

I am also reasonably satisfied that the 1 Feb JigsawExplainer generates j19 (Wednesday) JSXs without issues - I shall not spend more time on this!
Simply click 'Generate' and wait. Clicking 'Stop' multiple times doesnot help, as the 2nd click is on 'Generate'!!

If not interested in high-rated jigsaws, turn off the Dynamic and Nested chains (the last 3) in the Techniques dialog - this will speed up jigsaw generation.

j19 (Wednesday) JSX, ED=1.5/1.5/1.5 (singles only):

Code: Select all

8...........................6..23..957.....849..68..3...........................7 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX


[urhegyi]

I did some tests with UR 4.5 and BUG 5.7 disabled on a Wednesday grid and JSX. With this condition it works fine. So from now on in case of JSX I will do this even on single non overlapping grids.
I added the orthogonal symmetry, before I allowed only non-symmetrical grids, which takes longer to generate.

Code: Select all

1..352..9..........8.....3............9...3............4.....9..........2..843..6 111222223111122233441152333444555633744456663774555666777859966778889999788888999 JSX


Code: Select all

Analysis results 
Difficulty rating: 3,4
This Sudoku Jigsaw can be solved using the following logical methods:
63 x Hidden Single
 1 x Direct Claiming
 1 x Direct Hidden Pair
 8 x Cage Pointing
 5 x Claiming
 7 x Generalized Intersection
 1 x Naked Pair
 1 x X-Wing
 4 x Hidden Pair


[urhegyi]

Using the 1 Feb JigsawExplainer, I have generated Wednesdays, 2 JSs and 1 JSX in the last 10 minutes or so. So the program does work.
As to the hang, I dont know the cause. I'll consider it when I come across it. Work on JigsawExplainer is rather time limited.

I think I have found the exact case when this problem is reproducable.
I just generated a few JSX with the fiendish option, with "Full" as allowed symmetry type selected. It works fine.
Now I disable all symmetry types except "None". When now trying to generate the processor use goes to zero all the time.
I'm using this layout:

Code: Select all

...7.2.......8.............4...7...1.5.139.4.3...2...9.............6.......2.7... 111222233111123338114422238444552338475555538477655888476668899477769999776666999


I remember that when I had this isue with the "Wednesday" layout I was also generating with only "none" symmetrical types enabled, when I enable other symmetries it works as you described.

[1to9only]

I have changed the jigsaw generation code several times since the last release - it now runs slower!
It also appears that some earlier reported bugs have been sqashed in my current version.

JSX - Layout

Code: Select all

111222233111123338114422238444552338475555538477655888476668899477769999776666999


JSX - Fiendish - Full

Code: Select all

...4.7......8.6.......1....19.....28..3...9..58.....34....4.......2.8......5.1... ED=2.8
.9.6.2.5.5.......8.........1.......9.........9.......6.........4.......2.2.5.3.1. ED=2.9

The 2nd jigsaw looks interesting!

. 9 . 6 . 2 . 5 .
5 . . . . . . . 8
. . . . . . . . .
1 . . . . . . . 9
. . . . . . . . .
9 . . . . . . . 6
. . . . . . . . .
4 . . . . . . . 2
. 2 . 5 . 3 . 1 . ED=2.9



JSX - Fiendish - None

Code: Select all

.4........6...............81.............3..4...16..5.7............9......8...... ED=20.0 ED=9.1/7.7/2.6
.9.....4............8...........3.............................6........82...4..57 ED=20.0 ED=8.5/1.5/1.5
9.............29.6...........3.........................8.....5.........73....4... ED=20.0 ED=9.2/9.2/2.6
...7........................2.5.......6............8...9...3..........4....2...8. ED=20.0 ED=8.0/1.5/1.5
..2....1......5................6........5...........7...359..............67.....4 ED=20.0 ED=7.7/1.2/1.2
...3.........7.....6.5...8....................9.7.......1..............9.4.....5. ED=2.9



The ED=20.0 are Diabolical rated jigsaws.

Edit: For the ED=20.0 grids, I've added the Diabolical ratings.

[Hajime]

I have changed the jigsaw generation code several times since the last release - it now runs slower!
It also appears that some earlier reported bugs have been sqashed in my current version.

Hélas...
The first I try has multiple solutions:

Code: Select all

.4........6...............81.............3..4...16..5.7............9......8...... ED=20.0

Code: Select all

#1//B4,JS
247385691961874523536912478125749386879523164394168752782631945453296817618457239
111222233111123338114422238444552338475555538477655888476668899477769999776666999

#1//B4,JS
947328615562814397319675428153249786275983164824167953786451239431596872698732541
111222233111123338114422238444552338475555538477655888476668899477769999776666999


[1to9only]

The first I try has multiple solutions:

They are all JSXs.

[urhegyi]

About your multiple solutions: yours are JS while the examples are JSX with diagonals!
So as I understand your code changes when the rating is unknown SE 20.0 it has the freeze effect and don't display anything? The question now is when I don't ask to rate just generate it isn't showing anything too.

[Hajime]

The first I try has multiple solutions:

They are all JSXs.

Oopsy daicy. I was wrong. I can now confirm that has a unique solution AND I or SiSeSuSo cannot solve it by far (without T&E)

[urhegyi]

About your multiple solutions: yours are JS while the examples are JSX with diagonals!
So as I understand your code changes when the rating is unknown SE 20.0 it has the freeze effect and don't display anything? The question now is when I don't ask to rate just generate it isn't showing anything too.

edit:
To compare the version before 1 feb was able to find a non symmetrical JSX, so they exist(but generating with the older version with full processor use and none symmetrical case takes also much longer then symmetrical cases). The latest version has the display problem every time when SE 20.0 has occured it dont want to continue!
result of older version:

Code: Select all

.47.8........49........5......7...........8..9...2.7...3...................6..... 111222233111123338114422238444552338475555538477655888476668899477769999776666999 JSX


rating:

Code: Select all

.47.8........49........5......7...........8..9...2.7...3...................6..... 111222233111123338114422238444552338475555538477655888476668899477769999776666999 ED=2.9/1.2/1.2

and the one you posted above:

Code: Select all

...3.........7.....6.5...8....................9.7.......1..............9.4.....5. 111222233111123338114422238444552338475555538477655888476668899477769999776666999 ED=2.9/2.8/2.6


This layout can also be used for JSW and JSWX! (the one I posted with 10 clues in an other thread)

[1to9only]

Just a note to say that from looking at the code:
You can select as many symmetry types as you wish, when you click on Generate, the programs randomly selects one type, and produces a jigsaw with that symmetry type.

[urhegyi]

A brandnew JSWX layout with a lot of possibilities:

Code: Select all

#1/WX/B4,JS
.6.....8....6.8.......................73.14.......................9.6....2.....6.
111122222113344422133544442133555442133356667899555667899995667889996677888887777


Code: Select all

Eliminated candidates per Method and per Sudoku

Method   \  Sudoku |   SER |     1
                   |-------|------
Not counted elims  |     0 |   137
Naked Singles      |   0.1 |    41
Hidden Singles     |   0.2 |   115
Naked Single  [1]  |   2.5 |     4
Naked Quad    [4]  |     5 |     3
Hidden Quad   [5]  |   5.4 |     7
Hidden Triple [6]  |     4 |    13
Locked Singles[2]  |   2.8 |    87
Locked Singles[3]  |   2.9 |    12
Locked Singles[4]  |     3 |     4
Turbot-fish   [4]  |   4.2 |     4
                   |-------|------
Eliminated Cand's  |   427 |   427
Sum(SER * Cand's)  | 449.1 | 449.1

Initial Candidates :   427
Maximum SER rating :   5.4 - Approach
Labour rating      : 449.1 - Experimental rating