jook

I've been playing with visual studio, :) .
It can solve easy ones so far. I need to add some more rules so that it can tackle the tougher ones.

magnetboy

can it solve ones with more than one solution?

jook

I was aiming for the ones with one solution. I've heard of those with more solutions (at the sudoku programmers forum). My program doesn't brute force the solution though, so I expect it would only find different solutions by applying its rules in a different order. There are plenty of brute force programs out on the web. I'll probably add a brute force option to mine, for when it gets stuck. I've still got some solving rules to implement though. The source will go up on planetsourcecode when I'm finished, and I'll put a link here to the compiled program for anyone who wants a copy.
I've tested it with sudoku's from The Times book of sudoku's and it solved all the ones I tried.

edit

hmm the point of the pizzles in the newspaper is that there is only one solution. I'd guess those with more are mistakes, and are found by brute forcing them.

jook

four rules now:

1)if a square definately contains a number then the other squares in the same row, column and block cannot be that number
2)if there is only one square in a row, column or block that can be a number then it is that number
3)if there is a square in a row that can only appear in a certain block then remove it as a possibility for the squares in the row that are not in that block. repeat for column/block.
4)if there is a number that can only appear in either of two squares in a row then look to see if there is another number that can also appear in only those two squares. If so then remove the other numbers as possibilities for those two squares. repeat for columns and blocks. repeat for groups of 3,4,5 numbers/squares.

rule 4 took yonks to program correctly, but it's done now. So on to rule 5...