Lately, I’ve just started playing this game called Sudoku and found it quite interesting. As usual, the first thing that comes to my mind when I have a rough idea of how a game works is to code an algorithm to solve the puzzle using a computer.

There are a few approaches to this problem I can think of so far. The easiest and slowest of all is the brute-force iteration method. Just sub in a number, check if it contradicts any rules, and move to the next box, with a whole decision making tree of what to do upon alternative situations.

The second approach is by elimination. Fill up all the blanks with all possible but legal numbers. Through the checking of rules in a circular method from the centre, alternatives will be eliminated and most puzzles can be solved.

The third approach, the hardest and most interesting, is by inspection. By “visually” inspecting how set numbers are placed in a gird, reference to the rules of the game, some solutions are obvious and can be easily solved. The greatest challenge is to let a linear piece of code “read” the grids and say, “this looks like the only possible solution here”.

I shall think a little more about the data representation schema before starting work.

Python sounds like a rather good language to use...