Eigenvalues of perturbed matrix. Rouché's theorem.

In summary, to show that the perturbed matrix has exactly m eigenvalues (with algebraic multiplicity) inside the given region, you can use Rouché's theorem with the above choice of f and g.
  • #1
SrEstroncio
62
0

Homework Statement



Let [itex] \lambda_0 \in \mathbb{C} [/itex] be an eingenvalue of the [itex] n \times n [/itex] matrix [itex] A [/itex] with algebraic multiplicity [itex] m [/itex], that is, is an [itex]m[/itex]-nth zero of [itex] \det{A-\lambda I} [/itex]. Consider the perturbed matrix [itex] A+ \epsilon B [/itex], where [itex] |\epsilon | \ll 1 [/itex] and [itex] B [/itex] is any [itex] n \times n [/itex] matrix.

Show that given [itex] \delta \gt 0 [/itex], [itex] \alpha \gt 0 [/itex] exists so that, for [itex] | \epsilon | \lt \alpha [/itex], the matrix [itex] A + \epsilon B [/itex] has exactly [itex] m [/itex] eigenvalues (with algebraic multiplicity) inside [itex] | z - \lambda | \lt \delta [/itex]

Homework Equations



Rouché's theorem states that if [itex] f [/itex] is holomorphic in a region and [itex] |g(z)| \lt |f(z)| [/itex] on a curve (suitable for integration) inside the open region, then [itex] f [/itex] and [itex] f+ g [/itex] have exactly the same amount of zeros (with multiplicity) inside the curve.

The Attempt at a Solution



I first expanding the [itex] \det [/itex] function in power series and tried applying the integral which counts the number of zeros inside a region but I got stuck with several terms I couldn't get rid of. The TA told me to apply Rouché's theorem, but I can't figure out a way to exact the "sum" out of the determinant.
Any ideas? Any help would be appreciated.
 
Physics news on Phys.org
  • #2
SrEstroncio said:

Homework Statement



Let [itex] \lambda_0 \in \mathbb{C} [/itex] be an eingenvalue of the [itex] n \times n [/itex] matrix [itex] A [/itex] with algebraic multiplicity [itex] m [/itex], that is, is an [itex]m[/itex]-nth zero of [itex] \det{A-\lambda I} [/itex]. Consider the perturbed matrix [itex] A+ \epsilon B [/itex], where [itex] |\epsilon | \ll 1 [/itex] and [itex] B [/itex] is any [itex] n \times n [/itex] matrix.

Show that given [itex] \delta \gt 0 [/itex], [itex] \alpha \gt 0 [/itex] exists so that, for [itex] | \epsilon | \lt \alpha [/itex], the matrix [itex] A + \epsilon B [/itex] has exactly [itex] m [/itex] eigenvalues (with algebraic multiplicity) inside [itex] | z - \lambda | \lt \delta [/itex]

Should that not be [itex]|z - \lambda_0| < \delta[/itex]?

Homework Equations



Rouché's theorem states that if [itex] f [/itex] is holomorphic in a region and [itex] |g(z)| \lt |f(z)| [/itex] on a curve (suitable for integration) inside the open region, then [itex] f [/itex] and [itex] f+ g [/itex] have exactly the same amount of zeros (with multiplicity) inside the curve.

The Attempt at a Solution



I first expanding the [itex] \det [/itex] function in power series and tried applying the integral which counts the number of zeros inside a region but I got stuck with several terms I couldn't get rid of. The TA told me to apply Rouché's theorem, but I can't figure out a way to exact the "sum" out of the determinant.
Any ideas? Any help would be appreciated.

I think you want to take
[tex]f(z) = \det (A - zI)[/tex]
and
[tex]g(z,\epsilon) = \det (A + \epsilon B - zI) - f(z)[/tex]
and then look at the curve [itex]|z - \lambda_0| = \delta[/itex]. If you can show that, for all [itex]\delta > 0[/itex], there exists [itex]\alpha > 0[/itex] such that for all [itex]|\epsilon| < \alpha[/itex], [itex]|g(z,\epsilon)| < |f(z)|[/itex] on that curve, then Rouché's theorem will give you the result.
 
  • Like
Likes 1 person

1. What are eigenvalues of a matrix?

The eigenvalues of a matrix are the values that satisfy the characteristic equation of the matrix, which is formed by equating the determinant of the matrix minus a scalar value to zero. These values represent the scaling factors of the corresponding eigenvectors of the matrix.

2. How does perturbation affect the eigenvalues of a matrix?

Perturbation refers to the small changes or disturbances in the elements of a matrix. These changes can cause the eigenvalues of the matrix to shift or change. This change in eigenvalues can be analyzed using perturbation theory.

3. What is Rouché's theorem?

Rouché's theorem is a mathematical theorem that states that if a small perturbation is applied to a matrix, the eigenvalues of the perturbed matrix will be close to the eigenvalues of the original matrix. It is a useful tool in analyzing the effects of perturbation on the eigenvalues of a matrix.

4. How is Rouché's theorem used in perturbation analysis?

Rouché's theorem is used in perturbation analysis to estimate the changes in the eigenvalues of a matrix when small perturbations are applied. It provides a way to predict the behavior of the eigenvalues and how they will shift under perturbation.

5. Can Rouché's theorem be applied to any type of matrix?

Yes, Rouché's theorem can be applied to any square matrix, regardless of its size or properties. However, it is most commonly used for perturbation analysis of real symmetric matrices and complex Hermitian matrices.

Similar threads

  • Calculus and Beyond Homework Help
Replies
11
Views
2K
  • Calculus and Beyond Homework Help
Replies
5
Views
1K
  • Calculus and Beyond Homework Help
Replies
7
Views
3K
  • Linear and Abstract Algebra
Replies
1
Views
1K
  • Programming and Computer Science
Replies
1
Views
2K
  • Quantum Physics
Replies
2
Views
965
  • Calculus and Beyond Homework Help
Replies
5
Views
2K
  • Calculus and Beyond Homework Help
Replies
3
Views
2K
  • Calculus and Beyond Homework Help
Replies
29
Views
3K
  • Linear and Abstract Algebra
Replies
15
Views
4K
Back
Top