An operator theoretic approach to uniform (anti-)maximum principles

DOI10.1016/J.JDE.2021.11.037MaRDI QIDQ2064306zbMATH OpenOpenAlexFDO

Publication date 5 January 2022

Published in Journal of Differential Equations (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2104.12205

eventual positivity eventually positive resolvents Krein-Rutman theorem maximum principle uniform anti-maximum principle

Mathematics Subject Classification ID

Positive solutions to PDEs (35B09) Maximum principles in context of PDEs (35B50) Banach lattices (46B42) Positive linear operators and order-bounded operators (47B65)

Abstract: Maximum principles and uniform anti-maximum principles are a ubiquitous topic in PDE theory that is closely tied to the Krein--Rutman theorem and kernel estimates for resolvents. We take up a classical idea of Tak'av{c} - to prove (anti-)maximum principles in an abstract operator theoretic framework - and combine it with recent ideas from the theory of eventually positive operator semigroups. This enables us to derive necessary and sufficient conditions for (anti-)maximum principles in a very general setting. Consequently, we are able to either prove or disprove (anti-)maximum principles for a large variety of concrete differential operators. As a bonus, for several operators that are already known to satisfy or to not satisfy anti-maximum principles, our theory gives a very clear and concise explanation of this behaviour.

Review text

Given a general differential operator \(A\) with domain \(\mathrm{dom}(A)\) in a Banach lattice \(E\), the authors study the question whether the operator \(A\) satisfies the maximum principle or the anti-maximum principle in a neighborhood of a simple eigenvalue. That is, given the equation \[ (\lambda-A)u=f \] for real numbers \(\lambda\) in the resolvent set of \(A\) the problem is to answer whether \(f\geq 0\) implies \(u\geq 0\) (maximum principle) or \(u\leq 0\) (anti-maximum principle). Indeed, in this work, the authors actually study a stronger implication, which studies \(\lambda\) such that the solution \(u\) to the above equation actually dominates a positive multiple of the leading eigenfunction of \(A\) (or a negative multiple in case of the anti-maximum principle). The main contribution of this article uses two assumptions given a fixed positive function \(u\in E\), a fixed positive functional \(\phi\in E'\), and an eigenvalue \(\lambda_0\) of \(A\). The first assumption on \(A\), \(u\), and \(\phi\) is the \textit{domination assumption} stating that there are integers \(m_1,m_2\geq 0\) such that the modulus of every element of \(\mathrm{dom}(A^{m_1})\) resp. \(\mathrm{dom}((A')^{m_2})\) is dominated by a positive multiple of \(u\) resp. \(\phi\). The second assumption is a \textit{spectral assumption}, which states that \(\lambda_0\) is a geometrically simple eigenvalue of \(A\) and of \(A'\) and the eigenspace \(\ker(\lambda_0-A)\) resp. \(\ker(\lambda_0-A')\) is spanned by a vector \(v\) resp. \(\psi\) such that \(v\) dominates a positive multiple of \(u\) and \(\psi\) dominates a positive multiple of \(\phi\). Under these assumptions the authors then show that if there is a number \(\mu_0>\lambda_0\) in the resolvent set of \(A\) such that the resolvent \(R(\mu_0,A)=(\mu_0-A)^{-1}\) dominates a positive multiple of the functional \(f\mapsto -\langle \phi,f\rangle u\), then \(R(\mu,A)\) dominates a positive multiple of \(f\mapsto \langle \phi,f\rangle u\) for all \(\mu\) in a right neighborhood of \(\lambda_0\). This gives the \textit{uniform maximum principle}, that is a maximum principle statement independent of the right hand side of the equation. In a similar way, the authors also get a uniform anti-maximum principle in a left-neighborhood of \(\lambda_0\). The authors complement their work with several examples including the Laplacian with different boundary conditions, polyharmonic operators, and a delay differential operator.

Recommendations

Cites work

Cited in

(9)

This page was built for publication: An operator theoretic approach to uniform (anti-)maximum principles

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2064306)