Is this obvious? If a puzzle has a solution, then it is solvable, right? Not at all! Look again at the
Mini Sudoku Definitions: Just because a puzzle has a solution doesnβt mean that it has exactly one solution (i.e., it is solvable). But even if we donβt think this needs a proof, letβs be paranoid again and use this as an excuse to focus on the logical structure of the proof.
Notice that we are proving that the claim is true no matter what mini sudoku puzzle we start with. We might start with this puzzle:
Clearly there is only one way to complete the puzzle: The top row has only one open square, so we can only put one digit in it, and then columns 2 and 3 only have one open square each, so we can fill those uniquely.
Looking at a single example can often be helpful when crafting a proof, but proving a general statement with an example is NEVER a correct proof.
While there are only around 152.58 billion mini sudoku puzzles (and most of those are not valid, fewer have solutions, and even fewer have exactly one empty square), we donβt really want to check all possible puzzles. So instead, we fix an arbitrary valid mini sudoku puzzle. We assume that it has a solution and has exactly three open squares. From this, we prove that there is only one possible solution.
Proof.
Let be an arbitrary mini sudoku puzzle. Assume has exactly three empty squares and that is a solution.
Since is arbitrary, we donβt know how the three empty squares are arranged. They could all be in different rows, or two could be in the same row, or all three could be in the same row.
If all the empty squares are in different rows, then in each row, there is exactly one empty square. The other three squares are filled with three different digits (since the puzzle is valid), so there is only one choice to fill the empty square. This number must be the number used in the solution
Now consider the case where two of the empty squares are in the same row, and the third square is in a different row. The third empty squareβs row has three different digits, so there is only one choice for the last square, and it must agree with Once this is filled in, the other two empty squares must be in two different columns, each of which has three filled-in digits. In each of these columns, the three filled-in digits are different, so there is only one choice for the empty square. So again, any solution must be exactly
Finally, if all three empty squares are in the same row, then they are all in different columns. So using the same argument as we did when the empty squares were in different rows, but using columns instead, we see that is the only solution.
We have considered all possible cases, and in each case, is the only solution, so is solvable.