Yep.
So far I figured this out:
I’ve found out how a genetic algorithm can solve Sudoku.
It’s simple, each chromosome is a vertex with 2 coordinates. These coordinates can have numbers 1-9.
Each coordinate (x, y) are genes.
A vertex tree can be formed from these vertices.
The vertices can all touch each other(line drawn from one to another until they form a solid state) if the constraints are met, which are defined by the Sudoku rules.
All I have to do is have the GA algorithm generate vertices until a tree can be formed.
This could be achieved with a very simple rule system or a neural net.
The biggest problem with this would be the performance hit of building a tree and iterating through it.
The tree would have to be built and re-iterated at every run, this will have an exponential performance hit.