From: Adam Shelly <adam.shelly@...>
Date: 2005-08-25T07:20:49+09:00
Subject: Re: [QUIZ] Sodoku Solver (#43)

On 8/24/05, Kroeger Simon (ext) <simon.kroeger.ext@siemens.com> wrote:
> 
> > Ok, here's my latest.   My big speedup was figuring out that if I make
> > a bad guess, I just eliminate that possibility and re-solve.   See the
> > startGuessing method..
> 
> hmm ... I use cells with only 2 possibilities left for guessing, so if I
> 
> guess wrong, at least I know the right guess. (of course only if there
> are cells with 2 possibilities - but that happens often) If I interpret
> your guess method right you are doing the same thing. So I do not
> understand where the speedup comes from.
> 

I went back and looked at the differences between the two versions.
After a bad guess, my first version made another guess with the next
possibility on the same cell.
Now it just removes the bad guess and goes back to the logical solver.
I also speeded up the guesser itself - it used to check all the cells
for the one with the minimum number of possibilites, now it just finds
the first with 2 possibilities.

So that's why mine got better, but it was bad before.  
I'd like to directly compare the speed of yours to mine on my machine, 
but I can't get it to run correctly, It reports every board as UNSOLVABLE.
Has anyone else had any luck with it?

>Your 'findPairs' looks promissing... I think I will try something
>similar...

I thought that I could speed it up by deducing more cells before I
started guessing, soI tried adding another step to the logic phase:
If any row or col has all it's instances of a given digit in the same
3x3 box, then that digit can be removed from the other 6 cells in the
box.

But it turned out to make the code slower, I think because it wasn't
very efficient,  and in most cases it only ended up solving 1 or 2
extra cells before having to go to the guesser anyway.

-Adam