[Paper Review] Analysis of a time-dependent problem of mixed migration and population dynamics

François Castella, Philippe Chartier+1 more|arXiv (Cornell University)|Dec 7, 2015

Mathematical and Theoretical Epidemiology and Ecology Models5 references3 citations

TL;DR

This paper develops a mathematical framework for analyzing a time-dependent prey-predator model with fast spatial migration and slow demographic interactions, using a generalized center manifold theorem combined with averaging techniques. The key result is that the system reduces to an autonomous Lotka-Volterra system with averaged coefficients, and higher-order analysis reveals explicit destabilization of limit cycles due to fast oscillations.

ABSTRACT

In this work, we consider a system of differential equations modeling the dynamics of some populations of preys and predators, moving in space according to rapidly oscillating time-dependent transport terms, and interacting with each other through a Lotka-Volterra term. These two contributions naturally induce two separated time-scales in the problem. A generalized center manifold theorem is derived to handle the situation where the linear terms are depending on the fast time in a periodic way. The resulting equations are then amenable to averaging methods. As a product of these combined techniques, one obtains an autonomous differential system in reduced dimension whose dynamics can be analyzed in a much simpler way as compared to original equations. Strikingly enough, this system is of Lotka-Volterra form with modified coefficients. Besides, a higher order perturbation analysis allows to show that the oscillations on the original model destabilize the cycles of the averaged Volterra system in a way that can be explicitely computed.

Motivation & Objective

• To analyze a time-dependent system of differential equations modeling prey-predator dynamics with two distinct time-scales: fast migration and slow demographic interactions.
• To address the challenge of periodic, rapidly oscillating migration terms that depend on fast time and break standard averaging assumptions.
• To derive a reduced, autonomous dynamical system that captures the essential long-term behavior of the original highly oscillatory model.
• To quantify the impact of fast oscillations on the stability of limit cycles in the averaged system through higher-order perturbation analysis.
• To validate the theoretical framework with a numerical example involving two spatial sites, demonstrating convergence and error scaling.

Proposed method

• Formulates a system of ODEs with a small parameter ε representing the ratio of migration to demographic time-scales.
• Applies a generalized center manifold theorem to construct a slow manifold that depends periodically on the fast time variable θ = t/ε.
• Derives an explicit recursive formula to approximate the center manifold up to arbitrary order in ε using solutions to a partial differential equation.
• Performs a periodic change of variables to eliminate fast oscillations and applies averaging techniques to derive a simplified, autonomous system.
• Uses numerical integration (RK4) and implicit Euler schemes to validate the approximation, comparing exact and approximate solutions.
• Employs error analysis on log-log scales to confirm the O(ε) convergence rate of the first-order center manifold approximation.

Experimental results

Research questions

• RQ1 How can a system with fast, time-periodic migration and slow population dynamics be reduced to a simpler, autonomous system?
• RQ2 What is the structure of the center manifold when the linear terms depend periodically on the fast time variable?
• RQ3 How do fast oscillations in migration rates affect the stability of limit cycles in the averaged Lotka-Volterra system?
• RQ4 Can the center manifold be computed explicitly up to arbitrary order in ε using recursive relations?
• RQ5 What is the convergence rate of the first-order center manifold approximation in terms of the small parameter ε?

Key findings

• The reduced system is of Lotka-Volterra form with coefficients that are spatial and temporal averages of the original system's parameters.
• The center manifold depends periodically on the fast time variable and can be approximated to arbitrary order using a recursive solution of a PDE.
• Higher-order perturbation analysis shows that fast oscillations destabilize the limit cycles of the averaged system, with the destabilization effect explicitly computable.
• Numerical validation confirms that the first-order center manifold approximation yields an error of order O(ε), with slopes of 1.0386 for preys and 1.0080 for predators on log-log error plots.
• When initial conditions are chosen on the center manifold, the error decays exponentially in t/ε and stabilizes at O(ε), confirming the asymptotic validity of the approximation.
• The method successfully captures the dynamics of a two-site model, with numerical results matching the theoretical predictions for both solution trajectories and error behavior.