Sudoku polynomial algorithm?

algorithm, complexity-theory, np-complete

Solution

Because the generalized Sudoku problem (with n2 × n2 grids) is NP-hard, if there were a known polynomial-time algorithm for solving Sudoku puzzles, it would prove P = NP. That would be a huge deal.

Something to keep in mind is that Sudoku puzzles as we know them are all 9 × 9 grids. As a result, when trying to measure the time complexity of solving a Sudoku puzzle, it actually isn't a good idea to use big-O notation because big-O notation talks about how an algorithm scales in the long term and all we care about is solving Sudoku puzzles of a fixed size. Restated differently, if you wanted to talk about a polynomial-time for solving Sudoku puzzles, you'd have to answer the question "polynomial in what variable?" If you're always solving a 9 × 9 Sudoku grid, the answer to that question isn't really clear.

Problem

I have a project to do for a complexity and problem solving course, and I've decided to base the project on Sudoku. From the research I've done, Sudoku is an NP-Complete problem (which is required for the project), and I've found a few ways of creating algorithms for it. I'm planning on doing a brute force solving method, and I need to do two other methods. I've found some ways, such as solving it as an Exact Cover problem, and I've found a paper that describes Sudoku as a SAT problem. But my question is this: Is there a proven polynomial solution for Sudoku? My teacher seems to think there was a "clever" solution by a "senior" gentleman about 5 years ago, but that's all he can remember. Does anybody know what this solution is, or what any other polynomial solution is? I'd appreciate any information or tips. Thanks!

Original source