Karlin Theory On Growth and Mixing Extended to Linear Differential Equations

TL;DR

This paper extends Karlin's theorem to linear differential equations with site-specific growth and mixing, showing that increased mixing reduces the spectral abscissa, with applications to ecological and biological models.

Contribution

It generalizes Karlin's growth-mixing theorem to differential equations, providing an analytic solution and broadening its applicability.

Findings

Spectral abscissa decreases with increased mixing parameter m.

Provides an analytic solution to growth and mixing models.

Applicable to ecological and tissue growth models.

Abstract

Karlin's (1982) Theorem 5.2 shows that linear systems alternating between growth and mixing phases have lower asymptotic growth with greater mixing. Here this result is extended to linear differential equations that combine site-specific growth or decay rates, and mixing between sites, showing that the spectral abscissa of a matrix D + m A decreases with m, where D does-not-equal c I is a real diagonal matrix, A is an irreducible matrix with non-negative off-diagonal elements (an ML- or essentially non-negative matrix), and m >= 0. The result is based on the inequality: u' A v < r(A), where u and v are the left and right Perron vectors of the matrix D + A, and r(A) is the spectral abscissa and Perron root of A. The result gives an analytic solution to prior work that relied on two-site or numerical simulation of models of growth and mixing, such as source and sink ecological models, or multiple tissue compartment models of microbe growth. The result has applications to the Lyapunov stability of perturbations in nonlinear systems.