Question about eigenvalues when a sequence of matrix converges

407 Views Asked by At

Let ${A_n}$ be a sequence of $p \times p$ symmetric positive semi-definite matrix and $A_n$ converges to a matrix $A$, that is every elements of $A_n$ converges to corresponding element of $A$.

Then can I say anything about the relationship between the eigenvaues of $A_n$ and $A$? For example, If the norm of $A_n-A$ is small enough, can I make the eigenvalues of $A_n$ close enough to that of $A$?

1

There are 1 best solutions below

0
On BEST ANSWER

Since you can see $A_n$ as pertubated matrices to the exact $A$, that is diagonalizable, you can use the Bauer Fike Theorem.

It states, that the difference in eigenvalues is bounded by the condition of your eigenvector-basis and the norm of your pertubation.

Since your matrix is normal, you can set this condition number to 1.