dobrichev wrote:marek stefanik wrote:Every puzzle that has exactly one solution is valid.
I agree, but the puzzles that have many solutions are valid too.
yes!
we like puzzles with exactly one answer,
but others are valid too
dobrichev wrote:marek stefanik wrote:Every puzzle that has exactly one solution is valid.
I agree, but the puzzles that have many solutions are valid too.
dxSudoku wrote:Guessing is not logic. It's random. Brute-force is guessing. Logic is based on reason not guessing.
marek stefanik wrote:Isn't the number of CSP-Variables in a braid the same as the number of strong inferences in the corresponding dynamic chain?
I'm not convinced that substituting 'dynamic chains' for 'braids' and 'strong inferences' for 'CSP-variables' would make the sentence false.
marek stefanik wrote:Isn't the number of CSP-Variables in a braid the same as the number of strong inferences in the corresponding dynamic chain?
I'm not convinced that substituting 'dynamic chains' for 'braids' and 'strong inferences' for 'CSP-variables' would make the sentence false.
yzfwsf wrote:marek stefanik wrote:Isn't the number of CSP-Variables in a braid the same as the number of strong inferences in the corresponding dynamic chain?
I'm not convinced that substituting 'dynamic chains' for 'braids' and 'strong inferences' for 'CSP-variables' would make the sentence false.
I think the length of the denis_berthier's chains is equivalent to the recursion depth of BFS.
But that's true for dynamic chains too, isn't it? That's what determines the SER.yzfwsf wrote:I think the length of the denis_berthier's chains is equivalent to the recursion depth of BFS.
I see, so you were saying that braids[length] can be defined as a pattern, not braids in general.denis_berthier wrote:,marek stefanik wrote:Isn't the number of CSP-Variables in a braid the same as the number of strong inferences in the corresponding dynamic chain?
I'm not convinced that substituting 'dynamic chains' for 'braids' and 'strong inferences' for 'CSP-variables' would make the sentence false.
The numbers are the same if there are no inner Subsets.
But SE doesn't count the number of "strong inferences"; it counts the total number of inferences (the "nodes"). That makes a huge difference. As I said, it allows to define braids of a fixed length as a pattern; it doesn't allow to define dynamic chain with a fixed number of nodes as a pattern.
marek stefanik wrote:I see, so you were saying that braids[length] can be defined as a pattern, not braids in general.denis_berthier wrote:,marek stefanik wrote:Isn't the number of CSP-Variables in a braid the same as the number of strong inferences in the corresponding dynamic chain?
I'm not convinced that substituting 'dynamic chains' for 'braids' and 'strong inferences' for 'CSP-variables' would make the sentence false.
The numbers are the same if there are no inner Subsets.
But SE doesn't count the number of "strong inferences"; it counts the total number of inferences (the "nodes"). That makes a huge difference. As I said, it allows to define braids of a fixed length as a pattern; it doesn't allow to define dynamic chain with a fixed number of nodes as a pattern.
That's interesting, because the logic is the same no matter what the length is, and even braids of a given length can look completely different in the actual grid, but I think this depends on the definition of the word 'pattern'.
Pupp wrote:I heard that X-Sudoku has problems over 12 SE.
Pupp wrote: there are 2 types of solvers.
1. Solvers that use logic to solve every step to the end of the puzzle. It's the version that most programmers aspire to create.
2.Brute force solvers. They don't use logic to solve puzzles. They quickly guess the answer and cross check. There's only so many guesses needed to validate a puzzle. It's purpose is to just validate that a puzzle has only one solution. I'm guessing it's used mainly by programs churning out puzzles to toss puzzles with multiple solutions.
denis_berthier wrote:... among the pattern-based solvers, there are two sub kinds:
- solvers based directly on pattern-matching: AFAIK, CSP-Rules is the only one of this kind (not counting the toy solver included in the CLIPS examples);
- solvers based on procedural code and in particular on some form of T&E, BFS or DFS for chains. It's hard to say that such solvers "use logic for every step" (unless you stretch the meaning of "logic" to "anything that isn't false").
Serg wrote:Hi, Denis!denis_berthier wrote:... among the pattern-based solvers, there are two sub kinds:
- solvers based directly on pattern-matching: AFAIK, CSP-Rules is the only one of this kind (not counting the toy solver included in the CLIPS examples);
- solvers based on procedural code and in particular on some form of T&E, BFS or DFS for chains. It's hard to say that such solvers "use logic for every step" (unless you stretch the meaning of "logic" to "anything that isn't false").
Do you really mean that your approach doesn't use any form of "T&E, BFS or DFS" for spotting whips/braids?