You are torturing me now! So I was supposed to spot that you don't use an "O"
The puzzle after correction ported without problems & has a single solution
tarek
SudokuFP (massive)
56 posts
• Page 4 of 4 • 1, 2, 3, 4
Re: SudokuFP (massive)
by tarek » Thu Jan 31, 2019 10:36 pm
You are torturing me now! So I was supposed to spot that you don't use an "O"
The puzzle after correction ported without problems & has a single solution
tarek
-
tarek - Posts: 3762
- Joined: 05 January 2006
Re: SudokuFP (massive)
by Mathimagics » Fri Feb 01, 2019 1:24 am
tarek wrote:You just need to look at r5c5 to get your answer
Indeed!
enxio27 also pointed this out earlier on in the thread.
I have to get more sleep, I'm getting CBF (chronic brain fade)
-
Mathimagics - 2017 Supporter
- Posts: 1926
- Joined: 27 May 2015
- Location: Canberra
Re: SudokuFP (massive)
by m_b_metcalf » Fri Feb 01, 2019 7:24 am
tarek wrote:In keeping with Easy & not minimal theme These 16x16 puzzles (subject to verification) need the X, XNC & NC+ to solve. If that is the case then they will be useful for testing constraints, minimization & even P&P solving
- Code: Select all
.3....8.......C.468..FB.CE5...19.9....E..G.........F5......A.......D2......CA....2....7..F....3...4..C.D..A...F.......A..B....4..C.......2....9.5F7..D36B4.....2.........1....A....8E.......F......63......4..........4..5......2.E..A.F9C.......4....C..8....7.
.3....8.......C.468..FB.CE5...19.9....E..G.........F5......A.......D2......CA....2....7..F....3...4..C.D..A...F.......A..B....4..C.......2....9.5F7..D36B4.....2.........1....A....8E......7F......63.................4..5......2.E..A.F9C.......4....C..8....7.
Here's the second one reduced to 60 clues and with a difficulty rating around SE 9.
Regards,
Mike
- Code: Select all
.3....8.......C.468...B..E5...19.9....E.............5......A.......D.......CA....2....7..F....3...4..C.D..A...F.......A..B....4..C.......2....9.5F7..D36B......2.........1....A....8E.......F......63......4..........4..5......2.E..A.F9C.......4....C..8......
P.S. Sorry about the 'O'!
-
m_b_metcalf - 2017 Supporter
- Posts: 13626
- Joined: 15 May 2006
- Location: Berlin
Re: SudokuFP (massive)
by hkociemba1 » Fri Feb 01, 2019 10:20 am
And this is a further reduced version of the first one with only 55 givens. I assume it is extremely hard - my solver needs about 15 min to find the solution (Mike's 62 givens version which has SE~9 solves in 0.5 s). Trying to further reduce the puzzle takes too long so it still might not be minimal.
- Code: Select all
+-------------+-------------+-------------+-------------+
| . 3 . . | . . 8 . | . . . . | . . . . |
| 4 6 8 . | . . 11 . | . 14 5 . | . . 1 9 |
| . 9 . . | . . 14 . | . . . . | . . . . |
| . . . 15 | 5 . . . | . . . . | . . . . |
+-------------+-------------+-------------+-------------+
| . . . 13 | 2 . . . | . . . . | 10 . . . |
| . 2 . . | . . 7 . | . 15 . . | . . 3 . |
| . . 4 . | . 12 . 13 | . . 10 . | . . 15 . |
| . . . . | . . 10 . | . 11 . . | . . 4 . |
+-------------+-------------+-------------+-------------+
| . . . . | . . . . | . . . . | . . 9 . |
| 5 15 7 . | . 13 . . | 11 . . . | . . . 2 |
| . . . . | . . . . | . 1 . . | . . 10 . |
| . . . 8 | 14 . . . | . . . . | 15 . . . |
+-------------+-------------+-------------+-------------+
| . . . 6 | 3 . . . | . . . 4 | . . . . |
| . . . . | . . 4 . | . 5 . . | . . . . |
| 2 . 14 . | . 10 . 15 | 9 . . . | . . . . |
| . 4 . . | . . 12 . | . 8 . . | . . 7 . |
+-------------+-------------+-------------+-------------+
-
hkociemba1 - Posts: 60
- Joined: 08 August 2018
Re: SudokuFP (massive)
by m_b_metcalf » Fri Feb 01, 2019 12:59 pm
hkociemba1 wrote:And this is a further reduced version of the first one with only 55 givens. I assume it is extremely hard - my solver needs about 15 min to find the solution (Mike's 62 givens version which has SE~9 solves in 0.5 s). Trying to further reduce the puzzle takes too long so it still might not be minimal.
My program solves 2 cells using pointing, but the remainder of the puzzle kills my solver stone dead.
Regards,
Mike
-
m_b_metcalf - 2017 Supporter
- Posts: 13626
- Joined: 15 May 2006
- Location: Berlin
Re: SudokuFP (massive)
by Mathimagics » Fri Feb 01, 2019 4:17 pm
.
My general-purpose high-speed solver is called dSolver. This is dobrichev's fsss2 (Fast Simple Sudoku Solver v2) extended by me so that it can operate in any Sudoku variant mode (eg SudokuP, SudokuW, SudokuX, SudokuJ in any combination), and optionally in FP mode (which of course includes NC, NC+).
The keywords here are Fast and Simple. Simple because it is essentially a singles-only solver with DFS capability (for T&E when guessing is required). Fast because it takes advantage of the 128-bit register capabilities that most modern CPU's have. (fsss2 is the solver that is used in Mladen's GridChecker tool). (Note: "simple" does not mean limited puzzle-solving capability!)
These solvers are particularly good at puzzle reduction, minimalising, etc. Thus, if I had a 16x16 version of dSolver, I could most likely do reductions/minimal testing rapidly on any of the sample puzzles above by Mike, tarek, hkociemba1, etc.
Now, as it happens, I would very much like to have a dSolver16 for my personal puzzle generator system, and indeed I have already made a start on this. I won't bore you with the details, but the 9x9 model can be extended to 16x16 support without too much fuss.
So, a working dSolver16 should only take a few days to complete, and I will hopefully be back here soon with minimal versions of your puzzle samples.
PS: I didn't mention XNC support, which will require some additional tinkering with dSolver (grumble, grumble). But I am willing to introduce this heretical variant mode for the benefit of the FP tribe, seeing as how they seem to have all wandered off the orthogonal-adjacency reservation!
My general-purpose high-speed solver is called dSolver. This is dobrichev's fsss2 (Fast Simple Sudoku Solver v2) extended by me so that it can operate in any Sudoku variant mode (eg SudokuP, SudokuW, SudokuX, SudokuJ in any combination), and optionally in FP mode (which of course includes NC, NC+).
The keywords here are Fast and Simple. Simple because it is essentially a singles-only solver with DFS capability (for T&E when guessing is required). Fast because it takes advantage of the 128-bit register capabilities that most modern CPU's have. (fsss2 is the solver that is used in Mladen's GridChecker tool). (Note: "simple" does not mean limited puzzle-solving capability!)
These solvers are particularly good at puzzle reduction, minimalising, etc. Thus, if I had a 16x16 version of dSolver, I could most likely do reductions/minimal testing rapidly on any of the sample puzzles above by Mike, tarek, hkociemba1, etc.
Now, as it happens, I would very much like to have a dSolver16 for my personal puzzle generator system, and indeed I have already made a start on this. I won't bore you with the details, but the 9x9 model can be extended to 16x16 support without too much fuss.
So, a working dSolver16 should only take a few days to complete, and I will hopefully be back here soon with minimal versions of your puzzle samples.
PS: I didn't mention XNC support, which will require some additional tinkering with dSolver (grumble, grumble). But I am willing to introduce this heretical variant mode for the benefit of the FP tribe, seeing as how they seem to have all wandered off the orthogonal-adjacency reservation!
-
Mathimagics - 2017 Supporter
- Posts: 1926
- Joined: 27 May 2015
- Location: Canberra
Re: SudokuFP (massive)
by tarek » Fri Feb 01, 2019 7:28 pm
Regarding the 2 easy puzzles I've posted. I'm posting the solver solutions when the constraints are disabled.
The puzzles should have each a unique solution with X:On XNC:On FP(1,N):On NC:On
This should help with making sure that there are no bugs in my code or anybody's code for comparison
The puzzles should have each a unique solution with X:On XNC:On FP(1,N):On NC:On
Puzzles with single solution: Show
This should help with making sure that there are no bugs in my code or anybody's code for comparison
Disable X then XNC then FP(1,N): Show
-
tarek - Posts: 3762
- Joined: 05 January 2006
Re: SudokuFP (massive)
by m_b_metcalf » Sat Feb 02, 2019 9:03 am
tarek,
I can confirm that 'X:On XNC:Off FP(1,N):On NC:On' results in two solutions in each case. The other combinations would need significant excavation to carry out, so I pass.
Regards,
Mike
I can confirm that 'X:On XNC:Off FP(1,N):On NC:On' results in two solutions in each case. The other combinations would need significant excavation to carry out, so I pass.
Regards,
Mike
-
m_b_metcalf - 2017 Supporter
- Posts: 13626
- Joined: 15 May 2006
- Location: Berlin
Re: SudokuFP (massive)
by hkociemba1 » Sat Feb 02, 2019 10:03 am
I can confirm X:Off XNC:On FP(1,9):On NC:On with 13/14 solutions and X:On XNC:Off FP(1,N):On NC:On with 2/2 solutions.
For X:On XNC:On FP(1,N):Off NC:On I would have to do some program modifications.
For X:On XNC:On FP(1,N):Off NC:On I would have to do some program modifications.
-
hkociemba1 - Posts: 60
- Joined: 08 August 2018
Re: SudokuFP (massive)
by tarek » Sat Feb 02, 2019 10:54 am
Thanks Mike & hkociemba1 …
This is really helpful. These puzzles could also serve as a benchmark for these solving techniques if anybody is interested in adding XNC and NC+ to their arsenal.
tarek
This is really helpful. These puzzles could also serve as a benchmark for these solving techniques if anybody is interested in adding XNC and NC+ to their arsenal.
tarek
-
tarek - Posts: 3762
- Joined: 05 January 2006
Re: SudokuFP (massive)
by tarek » Sat Feb 02, 2019 5:55 pm
m_b_metcalf wrote:The other combinations would need significant excavation to carry out
hkociemba1 wrote:have to do some program modifications.
I understand that. I for instance linked NC with FP(1,N) and NC with XNC. If I disable NC then that would disable FP(1,N) and XNC. To allow XNC without NC for instance will require some modifications
tarek
-
tarek - Posts: 3762
- Joined: 05 January 2006
56 posts
• Page 4 of 4 • 1, 2, 3, 4
