Problem. Let be the number of ways to color each cell in a rectangle either red or blue such that each block contains at least one blue cell. Show that is a multiple of , but not a multiple of .
Origin: New Zealand, Camp Selection 2018, problem 7.
Solution. We are going to use induction. Instead of the rectangle , we consider a rectangle ; and let be the number of ways to color the cells of this rectangle with the condition specified in the problem setting. So, we are interested in . In addition to , we introduce as the number of ways to color the cells of with the specified condition and such that the rightmost column (both cells in this column) is red; and as the number of ways to color the cells of with the specified condition and such that the rightmost column has at least one blue cell. Obviously, we have .
We can directly check that and .
Now, consider what happens when we add an additional column to the rectangle on the right, thus going from to . For , the rightmost column must be red, so there is exactly one possibility for coloring this additional column; and for the rightmost rectangle to have a blue cell, the second from the right column must contain at least one blue cell. Therefore, .
For , we have possibilities for coloring the rightmost column (both cells blue, or top blue and bottom red, or top red and bottom blue); and the condition for the rightmost rectangle is satisfied no matter how the second from the right column is colored. Therefore, .
Since , we obtain . Now, since , we obtain , or .
As a side note, this recurrence relation is similar to the one for the Fibonacci sequence, and it is possible to deduce from it the exact formula for :
I am not going to get into details of how to obtain this formula, since this formula doesn't really help to find the maximum power of that is a multiple of; but the technique is the same as for the Fibonacci sequence, which is widely available on the internet.
Back to the original problem, we introduce two additional sequences and such that and (in the definition, we are not assuming that and are integer numbers, we would need to prove it).
We can directly check that and .
Now, we have , hence . Also, we have , hence .
From the two obtained recurrence relations on and and the fact that and are integer numbers, it follows by induction that and are ineteger numbers for any .
Now, let us consider and modulo . We see that and . Then, . Therefore, by induction, for any . Then, . From this, proceeding by induction, we can deduce that .
Now, this immediately gives us that , and thus is a multiple of , but not a multiple of , and we are done.
As a final note, if we consider the general problem for the rectangle , the obtained results give us the maximum power of that is a multiple of, for any except when (because when for , we don't know the maximum power of that is a multiple of). So, the problem would be more difficult for the rectangle than it is for the rectangle . The readers can think about this remaining part of the general problem.