The New Sudoku Players' Forum

[jpg]

http://www.mathstat.dal.ca/~jpg/sudoku/

[tarek]

My solver is javascript, but it struggles with very advanced puzzles requiring techniques above xy wings in terms of speed..........

somebody mentioned elsewhere that the problem is mainly due to language itself being slow as it depends on the browser. do you have the same problem ?

what impact would writing the whole script embedded in the .htm file instead of .js file have in terms of speed ?

[jpg]

Javascript is definitely slower than a compiled language, but it's more than fast enough for Sudoku with the right algorithms (Sudoku Helper can find chains of length 8 or 9 in just a few seconds). Embedding the script in the .htm file will have no effect on the speed.

[tarek]

thanx for that. the algorithms you use are short and concise, which will explain the efficiency. can you give me a tough puzzle (which your solver can solve) with the Time needed to solve it, so that i could see how much slower my solver is (i am assuming of course ).

I can attempt this on my own as your solver is available on the net. I hoped that you may have something already done.

regards,

tarek

[jpg]

Try selecting the puzzle called "1-constrained #6" which I think is the hardest example there. You'll have to time it on your own computer because the time it takes depends on the speed of the computer. You can also just browse through the examples and select ones to try on your program.

J.P.

[tarek]

I tried that one & used my stopwatch. Now I'm not sure that this was the best example becuase my solver used a different technique from yours at the end. I think the methods at the level of chains diverged. I'm sure that your solver is faster on most of the examples but this one my solver did in 2.5 Sec compared to 5 Sec from your solver.

I'll check the rest in due course & if I find any from my collection which you may find interesting, I'll post them to you.

regrds,

tarek

[jpg]

What is the URL for your solver? I'm curious to see what technique it used.

[tarek]

I haven't put on the net yet (I've been saying this for 4 months now ).

The technique which I am talking about is valid but apparantly not many use it. I thought it was some form of Forcing chains but apparantly it isn't. It is not new, but apparantly not many use it.

I was just debating it with another member after I posted a reply in a thread asking for help with a diabolical Sudoku. try that puzzle & comapare it with mine & other techniques used in that thread.

I'm happy to post the solution technique if you still wanted to??

[jpg]

I assume you mean the "Any candidate in A forces x in B" technique? I have seen that one debated before; I omitted that technique (and others that involve mutliple chains with positive assumptions at the heads) because many consider them to cross the T&E line. Although from a formal logic point of view it seems valid so I might reconsider.

[tarek]

I assume you mean the "Any candidate in A forces x in B" technique?

That is the one, & it was the technique needed to solve your puzzle. I'm not sure what to call so I started a thread to see what people would say (or re-say) about it