Hi everyone,
Hopefully I can get help here with bigger sudokus than 9x9. I tried my luck with a 32x32 sudoku in a newspaper, but it stumped me.
I made a copy in Microsoft Excel to allow adding multiple pencilmarks, but this is as far as I got. (See the attachment.)
[Pencilmark explanation: - green cells contain values I filled in
- small texts are pencilmarks. For example, 2/A/D means the cell can only have those values
- lowercase pencilmarks: h means horizontally linked, v means vertically linked, b means box linked. For example, if a row can only contain three '2'-s, they are all marked as 2h. Combinations can exist, for example: 4vbh.]
I am not looking for the final solution (the solution is accessible at the end of the newspaper anyway), instead I want to know what the next step is and why. And in case I would get stuck again later, a step-by-step solution would be helpful for reference, but that may be too much to ask for.
Alternatively, if someone manages to solve it, I would glady hear about solving techniques that make solving giant sudokus easier.
NOTE: If necessary, I can manually re-type the puzzle with a different format to input it easier into solving tools. Just please clarify what the format should look like.
Stuck on 32x32 sudoku
10 posts
• Page 1 of 1
Stuck on 32x32 sudoku
by Submachine » Fri Mar 26, 2021 4:54 pm
- Attachments
-
32x32 Sudoku.xlsx- Partially solved sudoku with chaotic pencilmarks
- (19.39 KiB) Downloaded 162 times
- Submachine
- Posts: 4
- Joined: 26 March 2021
Re: Stuck on 32x32 sudoku
by 1to9only » Fri Mar 26, 2021 6:50 pm
- Code: Select all
G Q . . E B . . . K . 7 . . . I O . 3 . 2 . . T . . . 0 9 4 . .
F D . 1 . O . . . . 9 . E . 8 . . R . P S . . U G 7 3 Q . . K N
P . U . . . . . 4 . . . . L 7 . 8 H . E . G . N J 9 . . R . . .
4 . O E . . F 8 I Q 2 U J S . . 9 B 5 . D . V C H 6 T . . . 1 7
3 R C . . 2 Q 6 G . . A H . . 0 M 4 U 7 . . 8 5 . . . . . . . D
7 . . T . K U H . S V C O . F . . . . . 3 6 J . L . E . Q . . 8
. . 8 . J . T . . E . F . M 1 V . . . C K O . . . . U 4 2 H . A
H B V 9 . L . . M . 3 8 . 2 U Q S 1 F . 7 E . . . D I N . . . .
. . Q U A F K . . 9 . R . . 5 . . E B V 1 3 . D M . G 2 H . 4 6
C . . . . 9 L Q 3 1 . . R . B M P G . 5 A T . . . . 6 . . . O F
R G D H . 4 O . . L N . . . 2 . . 8 . . . J . E 0 V 7 . . 3 . .
A J 9 . . . . M 2 . . . I . E S 4 U . . . C F . D . . . . . N 1
2 O 5 6 . I S P V C . 0 G 3 D 7 . . K R N 4 . . E 1 L . M B 8 J
E T . F 2 D 1 . . G . P . . 9 . 3 . . N . 0 . . B . C . . . . R
B V K . N E 0 . T . 8 . 1 4 H C L 6 . . . . . . Q I J . D . . .
1 . . 4 . J B G E . . I T F . O H . S . . . 6 K P 3 . . . . 0 C
L P . . . . 2 K 5 8 . . . I . H E . 4 S U . . 3 V R F J G . . .
Q . . O . . H . . . . . . . 3 G J 9 . 8 . D . 4 . A . K . 1 M .
D . . . . . . E . . R . 2 . 4 . . . . . G . T . . U 9 L . . C .
V 1 R C . G 4 I . . M E 5 P . . F . 6 U O . S 8 7 . H . J D . K
U . . . . . D A . J Q . . . K F G 7 . L . . . 6 T . . . . S E 9
K 9 F . . U V N 6 . I 4 . . A . . . . . . M C . . G . . 8 R . 3
T 5 G . . 3 . . . . F L Q . S N D M . O . . K A I 0 4 . . . . H
M E . J F 8 . S A . G K U 1 L . R 3 . . 9 . 7 . O Q D C 4 N . .
. . . D K 0 7 . . . E M . 9 I 1 V S R . . B . F . . P . 6 8 5 L
0 . 3 V I 5 . . . . P 9 7 . . . 1 C N . 6 . E . . J . T . F . 4
6 . . M Q V C 4 . R T G . . O . U D E 3 P 7 H . N 8 1 . S . . 0
8 . 1 . . . . . Q O . . 4 . 6 D 7 . . G I . . 2 9 E M . . C H V
S F E G 1 A 9 B 7 4 . V . H P U . . 8 I 5 N D M 3 C 0 6 T . R O
. . . N . . 8 J L . S . . . Q 2 . . H . . . . 1 . . V 7 . U . E
9 K . . D S E U 8 . . 6 N . J . . 0 . 4 . V . . . L Q . 1 . 7 .
. 7 L Q 3 H . . B . . 1 S C V E 6 . . . T 8 O . . . R D . . P G
I've extracted the current state of the 32x32 sudoku, including solved cells, but excluding pencilmarks.
I'm assuming 0-9,A-V map to the numbers 1-32.
Grid in 1-32 format:
- Code: Select all
17 27 . . 15 12 . . . 21 . 8 . . . 19 25 . 4 . 3 . . 30 . . . 1 10 5 . .
16 14 . 2 . 25 . . . . 10 . 15 . 9 . . 28 . 26 29 . . 31 17 8 4 27 . . 21 24
26 . 31 . . . . . 5 . . . . 22 8 . 9 18 . 15 . 17 . 24 20 10 . . 28 . . .
5 . 25 15 . . 16 9 19 27 3 31 20 29 . . 10 12 6 . 14 . 32 13 18 7 30 . . . 2 8
4 28 13 . . 3 27 7 17 . . 11 18 . . 1 23 5 31 8 . . 9 6 . . . . . . . 14
8 . . 30 . 21 31 18 . 29 32 13 25 . 16 . . . . . 4 7 20 . 22 . 15 . 27 . . 9
. . 9 . 20 . 30 . . 15 . 16 . 23 2 32 . . . 13 21 25 . . . . 31 5 3 18 . 11
18 12 32 10 . 22 . . 23 . 4 9 . 3 31 27 29 2 16 . 8 15 . . . 14 19 24 . . . .
. . 27 31 11 16 21 . . 10 . 28 . . 6 . . 15 12 32 2 4 . 14 23 . 17 3 18 . 5 7
13 . . . . 10 22 27 4 2 . . 28 . 12 23 26 17 . 6 11 30 . . . . 7 . . . 25 16
28 17 14 18 . 5 25 . . 22 24 . . . 3 . . 9 . . . 20 . 15 1 32 8 . . 4 . .
11 20 10 . . . . 23 3 . . . 19 . 15 29 5 31 . . . 13 16 . 14 . . . . . 24 2
3 25 6 7 . 19 29 26 32 13 . 1 17 4 14 8 . . 21 28 24 5 . . 15 2 22 . 23 12 9 20
15 30 . 16 3 14 2 . . 17 . 26 . . 10 . 4 . . 24 . 1 . . 12 . 13 . . . . 28
12 32 21 . 24 15 1 . 30 . 9 . 2 5 18 13 22 7 . . . . . . 27 19 20 . 14 . . .
2 . . 5 . 20 12 17 15 . . 19 30 16 . 25 18 . 29 . . . 7 21 26 4 . . . . 1 13
22 26 . . . . 3 21 6 9 . . . 19 . 18 15 . 5 29 31 . . 4 32 28 16 20 17 . . .
27 . . 25 . . 18 . . . . . . . 4 17 20 10 . 9 . 14 . 5 . 11 . 21 . 2 23 .
14 . . . . . . 15 . . 28 . 3 . 5 . . . . . 17 . 30 . . 31 10 22 . . 13 .
32 2 28 13 . 17 5 19 . . 23 15 6 26 . . 16 . 7 31 25 . 29 9 8 . 18 . 20 14 . 21
31 . . . . . 14 11 . 20 27 . . . 21 16 17 8 . 22 . . . 7 30 . . . . 29 15 10
21 10 16 . . 31 32 24 7 . 19 5 . . 11 . . . . . . 23 13 . . 17 . . 9 28 . 4
30 6 17 . . 4 . . . . 16 22 27 . 29 24 14 23 . 25 . . 21 11 19 1 5 . . . . 18
23 15 . 20 16 9 . 29 11 . 17 21 31 2 22 . 28 4 . . 10 . 8 . 25 27 14 13 5 24 . .
. . . 14 21 1 8 . . . 15 23 . 10 19 2 32 29 28 . . 12 . 16 . . 26 . 7 9 6 22
1 . 4 32 19 6 . . . . 26 10 8 . . . 2 13 24 . 7 . 15 . . 20 . 30 . 16 . 5
7 . . 23 27 32 13 5 . 28 30 17 . . 25 . 31 14 15 4 26 8 18 . 24 9 2 . 29 . . 1
9 . 2 . . . . . 27 25 . . 5 . 7 14 8 . . 17 19 . . 3 10 15 23 . . 13 18 32
29 16 15 17 2 11 10 12 8 5 . 32 . 18 26 31 . . 9 19 6 24 14 23 4 13 1 7 30 . 28 25
. . . 24 . . 9 20 22 . 29 . . . 27 3 . . 18 . . . . 2 . . 32 8 . 31 . 15
10 21 . . 14 29 15 31 9 . . 7 24 . 20 . . 1 . 5 . 32 . . . 22 27 . 2 . 8 .
. 8 22 27 4 18 . . 12 . . 2 29 13 32 15 7 . . . 30 9 25 . . . 28 14 . . 26 17
Now I just need a 32x32 sudoku solver!
[Edit] Fixed missing dots (empty cells) at end of a couple of lines. Thx to creint for pointing this out.
Last edited by 1to9only on Fri Mar 26, 2021 11:23 pm, edited 1 time in total.
-
1to9only - Posts: 4177
- Joined: 04 April 2018
Re: Stuck on 32x32 sudoku
by creint » Fri Mar 26, 2021 10:57 pm
1to9only: you were missing one last . in your format, 31 items on one line.
It needs some rounds of locked singles, if you don't use the naked singles you may need many rounds of sets.
Easiest advise is just to solve it with a solver, it saves a lot of time.
It is not about needing more techniques to solve those, because most of them just use the basics. It's more about looking in the right places to save time. Looking randomly is very ineffective.
Scan the whole grid once for singles + locked singles. Not applying your result. Yes this takes a lot of time and it feels like you don't make progress. Finding naked singles without pencilmarks is hard, but if you don't create pencilmarks than you still need them in this puzzle for sets if you don't use them.
If find a single you write down your excluded pencilmark locations.
Make sure you never miss anything or else you need to start scanning the whole puzzle again for something if your affected constraints dry up.
Apply your results and repeat but this time look only in your affected constraints.
You only need to scan affected cells for naked singles.
And scan affected digits in constraints for hidden single/locked singles.
Example: -1r1c1 -> scan r1c1, 1r1, 1c1, 1b1; you don't need to scan other digits in row 1.
r1c1 only 1 pencilmark left in cell?
1r1 only 1 pencilmark left or is the group pencilmarks of digit 1 contained in another constraint?
You can skip delays scanning for locked singles but you should resume scanning where you left off, maybe in reverse gives faster results.
On larger puzzles the chance of finding something at random decreases.
Example layout
Step 1
-1r1c1
-2r1c2
Scanlist 1
[constraint, hidden single, naked single, locked single]
r1c1, x, x, x
r1c2, x, x
1r1
1c1
1b1
2r1
2c1
2b1
You could keep seperate tables to index pencilmarks, but more bookkeeping can give errors.
For this puzzle you need ~160+ steps. 2500 initial pencilmarks / 160 = ~15 pencilmarks each step, 450 initial empty cells / 160 = ~3 filled cell each step.
I could generate one that only takes ~6 steps where the chance of finding something is much higher.
Let me know if this method works for manual solving.
- Code: Select all
12 32 21 . 24 15 1 . 30 . 9 . 2 5 18 13 22 7 . . . . . . 27 19 20 . 14 . .
- Code: Select all
27 . . 25 . . 18 . . . . . . . 4 17 20 10 . 9 . 14 . 5 . 11 . 21 . 2 23
It needs some rounds of locked singles, if you don't use the naked singles you may need many rounds of sets.
Easiest advise is just to solve it with a solver, it saves a lot of time.
It is not about needing more techniques to solve those, because most of them just use the basics. It's more about looking in the right places to save time. Looking randomly is very ineffective.
Scan the whole grid once for singles + locked singles. Not applying your result. Yes this takes a lot of time and it feels like you don't make progress. Finding naked singles without pencilmarks is hard, but if you don't create pencilmarks than you still need them in this puzzle for sets if you don't use them.
If find a single you write down your excluded pencilmark locations.
Make sure you never miss anything or else you need to start scanning the whole puzzle again for something if your affected constraints dry up.
Apply your results and repeat but this time look only in your affected constraints.
You only need to scan affected cells for naked singles.
And scan affected digits in constraints for hidden single/locked singles.
Example: -1r1c1 -> scan r1c1, 1r1, 1c1, 1b1; you don't need to scan other digits in row 1.
r1c1 only 1 pencilmark left in cell?
1r1 only 1 pencilmark left or is the group pencilmarks of digit 1 contained in another constraint?
You can skip delays scanning for locked singles but you should resume scanning where you left off, maybe in reverse gives faster results.
On larger puzzles the chance of finding something at random decreases.
Example layout
Step 1
-1r1c1
-2r1c2
Scanlist 1
[constraint, hidden single, naked single, locked single]
r1c1, x, x, x
r1c2, x, x
1r1
1c1
1b1
2r1
2c1
2b1
You could keep seperate tables to index pencilmarks, but more bookkeeping can give errors.
For this puzzle you need ~160+ steps. 2500 initial pencilmarks / 160 = ~15 pencilmarks each step, 450 initial empty cells / 160 = ~3 filled cell each step.
I could generate one that only takes ~6 steps where the chance of finding something is much higher.
Let me know if this method works for manual solving.
- creint
- Posts: 397
- Joined: 20 January 2018
Re: Stuck on 32x32 sudoku
by Submachine » Sat Mar 27, 2021 10:17 pm
Thanks for the responses! I must have been really tired to not notice that single. Especially ironic if you see my pencilmark for that cell.
I already scanned the entire grid for naked singles, that was the only one I missed.
I would like to use a solver, but most online tools I found were for 9x9 grids. And I do use pencilmarks, though I may be thinking of something different. Most of the time when I finish all basic techniques I know of and don't find anything, I am left with looking randomly, which - as you said - is indeed very ineffective.
Today I started creating an even bigger grid, where each cell is a 6x6 grid containing all possible values. In this bigger grid, I can keep track of ruled out values easier, but it will be a lot of time to make. (And it will all be manual, so errors can happen, as already proved by the missed single.)
I already scanned the entire grid for naked singles, that was the only one I missed.
I would like to use a solver, but most online tools I found were for 9x9 grids. And I do use pencilmarks, though I may be thinking of something different. Most of the time when I finish all basic techniques I know of and don't find anything, I am left with looking randomly, which - as you said - is indeed very ineffective.
Today I started creating an even bigger grid, where each cell is a 6x6 grid containing all possible values. In this bigger grid, I can keep track of ruled out values easier, but it will be a lot of time to make. (And it will all be manual, so errors can happen, as already proved by the missed single.)
- Submachine
- Posts: 4
- Joined: 26 March 2021
Re: Stuck on 32x32 sudoku
by creint » Sun Mar 28, 2021 9:13 pm
Submachine wrote:Today I started creating an even bigger grid, where each cell is a 6x6 grid containing all possible values. In this bigger grid, I can keep track of ruled out values easier, but it will be a lot of time to make. (And it will all be manual, so errors can happen, as already proved by the missed single.)
You want to create manually a large grid instead of solving one?
My solver can do all those things. If you cannot paste it then you need to do manual designing, cells+constraint. Probably complex for other users.
- creint
- Posts: 397
- Joined: 20 January 2018
Re: Stuck on 32x32 sudoku
by 1to9only » Mon Mar 29, 2021 2:43 pm
See: http://forum.enjoysudoku.com/giant-sudoku-s-16x16-25x25-36x36-100x100-t6578.html
Near the end of the thread, the 25x25 and 36x36 sudokus posted are solvable with singles only.
The last post has a link to a github repo that contains some pascal source code.
Near the end of the thread, the 25x25 and 36x36 sudokus posted are solvable with singles only.
The last post has a link to a github repo that contains some pascal source code.
-
1to9only - Posts: 4177
- Joined: 04 April 2018
Re: Stuck on 32x32 sudoku
by Submachine » Tue Mar 30, 2021 6:29 am
1to9only wrote:See: http://forum.enjoysudoku.com/giant-sudoku-s-16x16-25x25-36x36-100x100-t6578.html
Near the end of the thread, the 25x25 and 36x36 sudokus posted are solvable with singles only.
The last post has a link to a github repo that contains some pascal source code.
Thank you, though I am still trying to solve my current puzzle first.
And the size of this puzzle is 32x32, which does not fit the description.creint wrote:You want to create manually a large grid instead of solving one? My solver can do all those things.
No no, I still intend to solve it. But you said I need more pencilmarks to solve this puzzle, and creating a larger grid is my only way to keep track of 2-to-32 pencilmarks per cell.
There may be a confusion, because I did not receive a solver (you said your solver can do it, but I don't have it). But I would be happy to receive one. I just don't know much about github and pascal source codes, so I would prefer something with a user-friendly interface.
Should I create a new post in a different subtopic to look for a solver?
- Submachine
- Posts: 4
- Joined: 26 March 2021
Re: Stuck on 32x32 sudoku
by 1to9only » Tue Mar 30, 2021 10:23 am
For the partially solved 32x32 sudoku in post #1:
Solution Path:: Show
-
1to9only - Posts: 4177
- Joined: 04 April 2018
Re: Stuck on 32x32 sudoku
by Submachine » Sat May 15, 2021 1:11 pm
I managed to make a breakthrough earlier with this hint from creint:
I created another grid for pencilmarks only, and it made solving easier. I managed to fill in 47 cells just with the help of an extra grid for pencilmarks, and I will not miss a naked single again.
Now I came back because I ran out of ideas again, but wow, thank you 1to9only for creating the play-by-play. I already found the next move with it, and extended my solving method to not miss this type of clue again.
if you don't create pencilmarks than you still need them in this puzzle for sets
I created another grid for pencilmarks only, and it made solving easier. I managed to fill in 47 cells just with the help of an extra grid for pencilmarks, and I will not miss a naked single again.
Now I came back because I ran out of ideas again, but wow, thank you 1to9only for creating the play-by-play.
I already found the next move with it, and extended my solving method to not miss this type of clue again.What I missed: Show
- Submachine
- Posts: 4
- Joined: 26 March 2021
10 posts
• Page 1 of 1
