Abstract
Eigenvalue problems are omnipresent in physics. Important examples are the time independent Schrödinger equation in a finite orthogonal basis or the harmonic motion of a molecule around its equilibrium structure. Most important are ordinary eigenvalue problems, which involve the solution of a homogeneous system of linear equations with a Hermitian (or symmetric, if real) matrix. Matrices of small dimension can be diagonalized directly by determining the roots of the characteristic polynomial and solving a homogeneous system of linear equations. The Jacobi method uses successive rotations to diagonalize a matrix with a unitary transformation. A very popular method for not too large symmetric matrices reduces the matrix to tridiagonal form which can be diagonalized efficiently with the QL algorithm. Some special tridiagonal matrices can be diagonalized analytically. Krylov-space algorithms are the choice for matrices of very large dimension. We discuss especially the methods by Arnoldi and Lanczos.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
We do not consider general eigenvalue problems here.
- 2.
This matrix is skew symmetric, hence \(\mathrm{i}T\) is Hermitian and has real eigenvalues \(\mathrm{i}\lambda \).
- 3.
To avoid numerical extinction we choose the sign to be that of \(a_{12}\).
- 4.
We do not consider degenerate eigenvalues explicitly here.
- 5.
For simplicity we do not consider eigenvalues which are different but have the same absolute value.
- 6.
The equivalent QL method uses a lower triangular matrix.
- 7.
This is quite different from the Jacobi method since it is not an orthogonal transformation.
- 8.
Or the equivalent QL algorithm.
- 9.
For the QL method, it is numerically more efficient to start at the upper left corner of the matrix.
Author information
Authors and Affiliations
Corresponding author
Problems
Problems
Problem 10.1 Computer Experiment: Disorder in a Tight-Binding Model
We consider a two-dimensional lattice of interacting particles. Pairs of nearest neighbors have an interaction V and the diagonal energies are chosen from a Gaussian distribution
The wave function of the system is given by a linear combination
where on each particle (i, j) one basis function \(\psi _{ij}\) is located. The nonzero elements of the interaction matrix are given by
The Matrix H is numerically diagonalized and the amplitudes \(C_{ij}\) of the lowest state are shown as circles located at the grid points. As a measure of the degree of localization the quantity
is evaluated. Explore the influence of coupling V and disorder \(\varDelta \).
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this chapter
Cite this chapter
Scherer, P.O.J. (2017). Eigenvalue Problems. In: Computational Physics. Graduate Texts in Physics. Springer, Cham. https://doi.org/10.1007/978-3-319-61088-7_10
Download citation
DOI: https://doi.org/10.1007/978-3-319-61088-7_10
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-61087-0
Online ISBN: 978-3-319-61088-7
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)