How to prove $(I-e^{At})^{-1}$ only contains positive element?

62 Views Asked by At

Given a symmetric $N\times N$ matrix $A$, with eigenvalues $-x_1,-x_2,-x_3,\dots,-x_N$ and $x_1,x_2,\dots,x_N >0$.

$A$ is known as a negative-definite matrix.

We can diagonalize $A$ as $A = W*D*W^{-1}$

$D$ is a diagonal matrix with $A$'s eigenvalues on the diagonal position and all other elements are $0$. $W$ is composed by $A$'s eigenvector corresponding to eigenvalues of $A$.

how can I prove that $(I-e^{At})^{-1}$ only contains positive elements? if $I$ is a $N\times N$ identity matrix, $t>0$.

The current progress I have is

$$e^{At} = W* (e^{Dt}) *W^{-1}$$

$$I-e^{At} = W* (e^{I-Dt}) *W^{-1}$$

$e^{At}$ only contains positive elements

How to prove that $(I-e^{At})^{-1}$ only contains positive elements? Thank you

1

There are 1 best solutions below

0
On BEST ANSWER

For $t>0$ we can use the geometric series formula:

$$(I-e^{At})^{-1}=\sum_{n=0}^\infty \left ( e^{At} \right )^n.$$

where by convention $B^0=I$ for a square matrix $B$. Here we have used that $\| e^{At} \| < 1$ for $t>0$ in the Euclidean operator norm. This follows because the eigenvalues of $e^{At}$ are $e^{\lambda t}$ for eigenvalues $\lambda$ of $A$, and because the Euclidean operator norm is the largest singular value, which for a symmetric matrix is the eigenvalue of largest magnitude.

So it follows from what you've already shown (that $e^{At}$ has only positive elements).

The geometric series formula for matrices is proven the same way as the geometric series formula for numbers: prove that $\sum_{n=0}^N A^n = (I-A^{N+1})(I-A)^{-1}$ and then send $N \to \infty$.