SudokuSolver Forum

I really like the numerical rating system that Sudocue uses (based on the perceived difficulty of the various solving methods). I find that the numerical system gives a better differentiation than having 4-6 "categories" of difficulty, which are often rather subjective. Unfortunately, it appears likely that Sudocue will never be fixed, which pretty much leaves me hanging.

I see what appears to be a somewhat similar rating system in SudokuSolver, but I'm having a bit of difficulty interpreting the scores. In addition, it doesn't appear that SudokuSolver can score a batch of puzzles at once and dump the data to a file (which would be my preference).

Hi
I've worked extensively with Richard on SudokuSolver's numerical ratings. Here's a few thoughts.We have focussed 99.9% on rating Assassin level killers. So, there is unlikely to be a great correlation for any other type of sudoku. However, that can change. Richard has/is putting in many new sudoku routines into SS. Your input could be really helpful in the order and weightings for these routines. What order would you put the non-killers routines in and what relative weighting would you give to each? Note that the routine order is totally changeable and any order can be saved/exported/imported etc.

The current version starts from a score of about 0.50. But this could change with the next release.It does have an excellent batch solve capability. Richard churns through thousands of vanilla puzzles in the one sitting. On my version, it dumps the results of a batch solve onto the C drive in a text file called "SudokuScores".

Note - Richard is writing up the help file to catch up with all the recent changes since v3.3.0

Hope this clarifies it a bit
Cheers
Ed

The order and weight that Sudocue uses would be a good starting point. I'm still learning the various vanilla techniques, so I'm not the best person to ask about some of this.


How do I do that? All I really care about is the scores, so a listing of each solving technique used isn't really critical to me at this point.

It would be nice to be able to turn off the solving techniques that don't apply to the puzzle at hand (such as killer techniques for a vanilla puzzle, etc.). I'm guessing they're not used if they're not needed, but still, they get in the way.

On the File Menu there is an option called "Batch Solve" - click on this and you will get the batch solve window. This can be loaded with a series of puzzle definitions - you can use the Add button to create a blank line and type in a description and a puzzle definition. Alternatively, you can use the Load button to open a file of stored puzzles. The format of the file should be a comma separated list with a minimum entry per line of "Puzzle Description" , "Puzzle Definition". You can save the file from this window as well. I have a number of different "libraries" of puzzles that I can load and save - for instance I have 170+ Assassins and variations that Ed and I use for benchmarking our scoring and I also have a file with 1000 of Ruud's top 10,000 that I use to test new solving techniques.

Once you have a list of puzzles in the window, put a tick next to the ones you want to run and - to do a scoring run - click on the Score button. You'll get feedback in the window as each puzzle is completed - it will say Solved, Not Solved or Error - let me know if you get any Errors!! Talk to me very nicely, and I might get around to putting the score into this table as well . . .

When the scoring has finished for all the selected puzzles scores will have been written out to a text file at "C:\SudokuScores.txt". The file lists the enabled techniques and then, for each puzzle, the description, count of techniques used, the "score" and the time taken.

You can turn them off individually in the Solving Options window - are you thinking of something different?

Rgds
Richard