asymptotic behavior of a matrix multiplying with a vector

58 Views Asked by At

I want to find the asympototic behaviors of $||M^nv||$ with some rate. [Just claiming that it's divergent is not enough - it requires finding some parallel bounds.]

We are given that M is 2x2 with characteristic polynomial $t^2-6t+9$.

(1) How does the asymptotic behavior of $||M^nv||$ have anything to do with diagonalizability of a matrix?

(2) I only saw that it simulates asymptotically to $3^nv$, but I was told that there is another situation involving $n-1$ somewhere. Is this the situation when it's diagonalizable or not? What is the other situation and how is it derived?

This is not a homework question and I don't really know which topic this problem belongs to. Any help with two questions above will be appreciated!

1

There are 1 best solutions below

2
On BEST ANSWER

You can have two slightly different cases, as you suspected.

  • In the diagonalizable case, in some basis you get $M=\pmatrix{3&0\\0&3}$ so that indeed $M^nv=3^nv$.

  • In the non-diagonalizable case, there is a basis where you get the Jacobi normal form $M=\pmatrix{3&1\\0&3}$ so that $M^n=3^n\pmatrix{1&\frac{n}{3}\\0&1}$. So indeed, the norm (in the original basis) of the matrix-power-vector product can have components that behave like $n3^{n-1}$.