Given a matrix, how do you find entries of M^n?

101 Views Asked by At

If I am given a matrix $M=\begin{pmatrix} 13 & 1 \\ -16 & 5 \end{pmatrix}$

I need to find formulas for the entries of$ M^n$, where $n$ is a positive integer. I first calculated all the eigenvalues which are 9.So the eigenvector is: $$ \left[ \begin{matrix} -1/4\\ 1\\ \end{matrix} \right] $$.

I know that $M = PDP^{-1}$, so $ M^n = PD^nP^{-1}$.

So, $D=\begin{pmatrix} 9^n & 1 \\ 0 & 9^n \end{pmatrix}$

and $P=\begin{pmatrix} -1 & -1/4 \\ 4 & 0 \end{pmatrix}$

so is $P^{-1}$ $=\begin{pmatrix} 0 & 1/4 \\ -4 & -1 \end{pmatrix}$

so $M^n$ = \begin{pmatrix} -1 & -1/4 \\ 4 & 0 \end{pmatrix}\begin{pmatrix} 9^n & 1 \\ 0 & 9^n \end{pmatrix}\begin{pmatrix} 0 & 1/4 \\ -4 & -1 \end{pmatrix}

which gives \begin{pmatrix} 9^n+4 & 1 \\ -16 & 9^n-4\end{pmatrix}

But it is wrong. Does anyone know what I am doing wrong?

1

There are 1 best solutions below

0
On BEST ANSWER

The characteristic equation is

$ | \lambda I - M | = (\lambda - 13)(\lambda - 5) + 16 = \lambda^2 - 18 \lambda + 81 =(\lambda - 9)^2$

It roots are $9$ with algebraic multiplicity of $2$.

Now we find the eigenvectors

Let $v = [v_1 , v_2]^T $ then we have to solve the system

$ (\lambda I - M) v = 0 $

which is

$ \begin{bmatrix} -4 && -1 \\ 16 && 4 \end{bmatrix} \begin{bmatrix} x_1 \\ x_2 \end{bmatrix} = \begin{bmatrix} 0 \\ 0 \end{bmatrix}$

The solutions is

$\begin{bmatrix} v_1 \\ v_2 \end{bmatrix} = t \begin{bmatrix} -\frac{1}{4} \\ 1 \end{bmatrix} $

Since there is only one eigenvector, we have to generate a generalized eigenvector as follows

$ (M - \lambda I) u = v $

so that

$ \begin{bmatrix} 4 && 1 \\ -16 && -4 \end{bmatrix} \begin{bmatrix} u_1 \\u_2 \end{bmatrix} = \begin{bmatrix} -\frac{1}{4} \\ 1 \end{bmatrix} $

The system is consistent, and has the solution

$ \begin{bmatrix} u_1 \\ u_2 \end{bmatrix} = \begin{bmatrix} \frac{1}{4}(-\frac{1}{4} - t) \\ t \end{bmatrix} $

Put $t = 1$ you get

$ u = \begin{bmatrix} -\dfrac{5}{16} \\ 1 \end{bmatrix} $

So now

$ M [ v, u] = [\lambda v , v + \lambda u ] = \begin{bmatrix} v, u \end{bmatrix} \begin{bmatrix} \lambda && 1 \\ 0 && \lambda \end{bmatrix} $

so that we can take

$P = [v, u] = \begin{bmatrix} -\frac{1}{4} && -\frac{5}{16} \\ 1 && 1 \end{bmatrix} $

then

$M = P J P^{-1}$ where

$J = \begin{bmatrix} \lambda && 1 \\ 0 && \lambda \end{bmatrix} = \begin{bmatrix} 9 && 1 \\ 0 && 9 \end{bmatrix} $

Note that $P$ is not unique, we could have chosen

$ v = \begin{bmatrix} -1 \\ 4 \end{bmatrix} $

and by setting $t = 0$ results in

$ u = \begin{bmatrix} -\frac{1}{4} \\ 0 \end{bmatrix} $

so that

$ P = \begin{bmatrix} -1 && -\frac{1}{4} \\ 4 && 0 \end{bmatrix} $

Now,

$M^n = P J^n P^{-1} $

You can show by induction, or otherwise, that

$J^n = \begin{bmatrix} 9^n && n (9)^{n-1} \\ 0 && 9^n \end{bmatrix}$

and from this you can compute $M^n$

$M^n = \begin{bmatrix} -\frac{1}{4} && -\frac{5}{16} \\ 1 && 1 \end{bmatrix} \begin{bmatrix} 9^n && n (9)^{n-1} \\ 0 && 9^n \end{bmatrix} \begin{bmatrix} 16 && 5 \\ -16 && -4 \end{bmatrix} $

Multiplying out the first two matrices

$M^n = \begin{bmatrix} - \dfrac{9^n}{4} && -\frac{9^{n-1}}{16} (4 n + 45 ) \\ 9^n && 9^{n-1} ( n + 9 ) \end{bmatrix} \begin{bmatrix} 16 && 5 \\ -16 && -4 \end{bmatrix} $

and this becomes

$M^n = \begin{bmatrix} 9^{n-1} ( 4n + 9 ) && 9^{n-1} n \\ -16 n (9)^{n-1} && 9^{n-1} (9 - 4n) \end{bmatrix} = 9^{n-1} \begin{bmatrix} 4 n + 9 && n \\ -16 n && 9 - 4n \end{bmatrix} $

I checked $M^2 $ by direct calculation and using the above expression, and they match.