[Paper Review] Quasilinear eigenvalues

Julián Fernández Bonder, Juan Pablo Pinasco+1 more|arXiv (Cornell University)|Feb 26, 2014

Advanced Mathematical Modeling in Engineering36 references7 citations

TL;DR

This paper extends eigenvalue theory for the p-Laplacian to a general class of quasilinear elliptic operators with monotone, homogeneous structure. It establishes homogenization results for nonlinear eigenvalues, proving convergence of eigenvalues and eigenfunctions under periodic oscillations, with uniform bounds on nodal domain measures independent of the oscillation scale.

ABSTRACT

In this work, we review and extend some well known results for the eigenvalues of the Dirichlet $p-$Laplace operator to a more general class of monotone quasilinear elliptic operators. As an application we obtain some homogenization results for nonlinear eigenvalues.

Motivation & Objective

• To generalize known results on p-Laplacian eigenvalues to a broader class of quasilinear, monotone elliptic operators.
• To analyze the variational spectrum and spectral properties of these generalized operators.
• To establish homogenization results for nonlinear eigenvalues under periodic oscillations of coefficients and weights.
• To prove uniform lower bounds on the measure of nodal domains of eigenfunctions, independent of the homogenization parameter.
• To demonstrate convergence of eigenvalues and eigenfunctions in the homogenization limit.

Proposed method

• Formalize the eigenvalue problem for quasilinear operators via the divergence form $-\operatorname{div}(a(x,\nabla u)) = \lambda \rho(x)|u|^{p-2}u$ with homogeneous, monotone $a(x,\xi)$.
• Use variational methods and critical point theory to define the spectrum and eigenfunctions.
• Apply concentration-compactness and monotonicity techniques to analyze the behavior of eigenfunctions under periodic oscillations.
• Establish uniform lower bounds on nodal domain measures using comparison with the p-Laplacian and known estimates on $\mu_1(\mathcal{N})$.
• Use the Krasnoselskii genus and symmetric critical point theory to characterize eigenvalues via min-max principles.
• Prove convergence of eigenvalues $\lambda_k^\varepsilon$ to the limit eigenvalue $\lambda_k$ by combining upper and lower estimates via nodal domain control.

Experimental results

Research questions

• RQ1 How do the spectral properties of the p-Laplacian extend to more general quasilinear, monotone operators with homogeneous structure?
• RQ2 What is the behavior of eigenvalues and eigenfunctions under periodic homogenization of the coefficients and weights?
• RQ3 Can uniform lower bounds on the measure of nodal domains be established independently of the homogenization parameter $\varepsilon$?
• RQ4 Does the $k$-th eigenvalue of the oscillating problem converge to the $k$-th eigenvalue of the limit problem?
• RQ5 What is the relationship between the variational spectrum and the actual spectrum of the generalized eigenvalue problem?

Key findings

• The $k$-th eigenvalue $\lambda_k^\varepsilon$ of the oscillating problem converges to the $k$-th eigenvalue $\lambda_k$ of the limit problem as $\varepsilon \to 0$.
• Each nodal domain of the eigenfunction $u_k^\varepsilon$ has measure bounded below by a positive constant $C(k)$ independent of $\varepsilon$, ensuring no concentration.
• The first eigenvalue on each nodal domain satisfies $\lambda_{1,i}^\varepsilon \to \lambda_{1,i} = \lambda_k$ as $\varepsilon \to 0$, with $\lambda_{1,i}$ equal to the $k$-th eigenvalue in the limit.
• The upper bound $\lambda_k^\varepsilon \leq \max_i \lambda_{1,i}^\varepsilon$ holds, and since $\lambda_{1,i}^\varepsilon \to \lambda_k$, it follows that $\limsup_{\varepsilon \to 0} \lambda_k^\varepsilon \leq \lambda_k$.
• The lower bound $\lambda_k \leq \liminf_{\varepsilon \to 0} \lambda_k^\varepsilon$ is established via nodal domain measure control and the fact that eigenvalues cannot drop below $\lambda_k$.
• The combination of upper and lower bounds yields $\lim_{\varepsilon \to 0} \lambda_k^\varepsilon = \lambda_k$, proving convergence of the $k$-th eigenvalue.