About Warm-ups Problems News

Problem 6

Posted 04/02/2015 and updated 06/19/2015
Refresh the webpage if formulas are not shown correctly.
----------------------
Previous    Next
----------------------

This problem is derived from the previous problem, Problem 5. In this problem, we consider the set of all the binary square matrices of size `n\times n`, denoted by `\mathcal B_{n,n}`. Suppose `M\in\mathcal B_{n,n}` and `P\in \mathcal B_{2,2}`, we define `mch(P,M)` as the number of occurences of `P` as blocks in `M`. If `mch(P,M)=0`, we say `M` avoids `P`.

For example, `P=[[1,1],[1,1]]\in \mathcal B_{2,2}` and `M=[[0,0,1,1],[1,0,1,1],[1,1,1,1],[1,1,1,0]]\in \mathcal B_{4,4}`.

We see `mch(P,M)=4`. Note that occurences could be overlapped. In this example, there are two occurences in the third and fourth column.

`a(n)` is the number of elements in `\mathcal B_{n,n}` avoiding `P=[[0,1],[1,1]]`.

For example, `a(1)=2`, `a(2)=15`.

Find `a(n)`.



A more general question is as follows,

`m(n,k)` is the number of elements in `\mathcal B_{n,n}` having exact `k` occurences of `P`, where `P` could be any matrix in `\mathcal B_{2,2}`. It's easy to see `m(n,0)=a(n)`.

Find `m(n,k)` or the generating function of `m(n,k)`.

----------------------

Updated 06/19/2015 by Brian Miceli

One possible idea is to develop some approach like two-dimensional Goulden-Jackson cluster method.

Back to top