Sponsor
Portland State University. Department of Mathematics and Statistics
First Advisor
Jeffrey Ovall
Term of Graduation
Spring 2024
Date of Publication
5-22-2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy (Ph.D.) in Mathematical Sciences
Department
Mathematics and Statistics
Language
English
DOI
10.15760/etd.3749
Physical Description
1 online resource (x, 157 pages)
Abstract
We develop computational tools for exploring eigenvector localization for a class of selfadjoint, elliptic eigenvalue problems regardless of the cause for localization. The user inputs a desired region R (not necessarily connected), a tolerance for the amount localization in R, and the desired energy range [a,b]. The tool outputs eigenvectors concentrated within the tolerance inside R and within [a, b]. We develop ample theory that justifies our algorithm, which involves a complex, compact perturbation of the operator L, Ls = L+isχR, for some (small) s > 0. Our central idea can be summarized as follows: if (λ,ψ) is an eigenpair of L with ψ highly localized in R, then there will be an eigenpair (μ,φ) of the shifted operator Ls such that μ is near λ+is and φ is near ψ i.e. the eigenpair (μ,φ) of Ls is close to the eigenpair (λ + is, ψ) of L + is. The algorithm finds eigenpairs (μ, φ) of Ls with I(μ) near s and R(μ) in or near [a, b]. In practice, the algorithm proves robust, immediately eliminating over 90% of unwanted eigenmodes, and many examples are provided. A post processing feature is included, which consists of (a few) inverse iterations that further eliminate unwanted eigenvectors and even identify computationally difficult cases.
Although most of the theory is developed with one kind of operator in mind, we show that it applies directly to the magnetic Laplacian, Hˆ (A) := (−i∇ − A(x))2 as well. We additionally provide a method for a priori prediction of where eigen- i functions may localize for the magnetic Laplacian. Essentially, if an eigenvector achieves a global maximum (in modulus) at x0 ∈ Ω, then A behaves similarly to a conservative vector field in a neighborhood around x0 that depends on the eigenvalue. We provide numerical examples to show the eigenmodes localize where |curlA| is small, at least for low energies, as expected based on our theory.
Rights
© 2024 Robyn Ashley Markee Reid
In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).
Persistent Identifier
https://archives.pdx.edu/ds/psu/42216
Recommended Citation
Reid, Robyn Ashley Markee, "Computational Tools for Exploring Eigenvector Localization" (2024). Dissertations and Theses. Paper 6617.
https://doi.org/10.15760/etd.3749
Comments
This material is based upon work supported by the National Science Foundation under Grant No. DMS-2208056 and Grant No. DMS-2136228.