The New Sudoku Players' Forum

[janiv]

*** *69 *3*
**1 *** ***
*6* *** **2

4*5 **3 ***
*** *** *5*
*16 *** **7

*** *9* ***
*** *** *71
1** *** ***

Thanks,
Janiv Ratson.

[paulf2127]

You've made a mistake there mate. Not enough clues to validate.

[angusj]

You've made a mistake there mate. Not enough clues to validate.

I counted 17 which is the very minimum for a unique solution. However, it looks like trial-and-error is required to find a solution if there is one.

[janiv]

myriddle.goop.co.il/riddles/sudoku.asp?id=55

[Animator]

This grid has multiple solutions.

[janiv]

How do U know that ?

[scrose]

Put the grid into a solver. For example, put the grid into rubylips' solver and click the Evaluate button. It will determine there are multiple solutions.

Update: rubylips' command-line solver determined there are at least 16990 solutions, all of them requiring trial-and-error. (I ran out of memory on the 16991st solution.)

[tso]

myriddle.goop.co.il/riddles/sudoku.asp?id=55

I checked a few more puzzles on this site. Several had multiple solutions. One had only 16 clues. On the other hand, I don't read Hebrew -- maybe there are other constraints on the solutions. Is it possible that these are Sudoku variants?

[Pat]

I checked a few more puzzles on this site. Several had multiple solutions.

I don't read Hebrew -- maybe there are other constraints on the solutions. Is it possible that these are Sudoku variants?

my SuDoku correspondent in Israel reports:

it seems that myriddle.goop.co.il/riddles/sudokulist.asp?user=&userType=&sort=2 allows any Registered Member to contribute a puzzle;
this may explain the poor quality of the puzzles.

they're supposed to be regular SuDoku - not variants.

for their Tough puzzles (the 3-star rating), they do say that "guessing is required" - which is not the same as admitting multiple answers, is it?

- Pat

[Sue De Coq]

scrose,

Thanks for the bug report - I've fixed the problem for the next release. FYI, my 3GHz PC found half-a-million solutions in 100s, at which point my patience gave out before my computing resources.

[Pat]

scrose,

Thanks for the bug report - I've fixed the problem for the next release

and for newer members such as i,
perhaps you could mention which software it is you're discussing here?

thanks!

- Pat

[Sue De Coq]

Pat,

perhaps you could mention which software it is you're discussing here?

It's the solver mentioned (and linked to) earlier in the thread by scrose:

Update: rubylips' command-line solver determined there are at least 16990 solutions, all of them requiring trial-and-error. (I ran out of memory on the 16991st solution.)

It's hosted at act365.com/sudoku and the source code is available under GPL at sourceforge.net/projects/sudoku. It's nowhere near as pretty as other solvers out there but it's fairly fast and seems to solve all puzzles. I've been off the project for a while but I'm back on board now so expect to see some progress ...

BTW, rubylips is my real name - Sue De Coq is just a silly nom de forum posting I sometimes use.

[Lardarse]

I don't even want to say what that name implies...

[Sue De Coq]

It's something to do with chickens.

[gsf]

Update: rubylips' command-line solver determined there are at least 16990 solutions, all of them requiring trial-and-error. (I ran out of memory on the 16991st solution.)

The command line solver at http://www.research.att.com/~gsf/sudoku/
reports 50044975 puzzles in 161.486451 seconds.
49762991 up to isomorphism (that took another 2.5 hours to compute.)