Global compact attractors and their tripartition under persistence

Similar documents
Global attractors, stability, and population persistence

Persistence theory applied to Keen s model a link between mathematical biology and mathematical economics

Permanence Implies the Existence of Interior Periodic Solutions for FDEs

FIXED POINT OF CONTRACTION AND EXPONENTIAL ATTRACTORS. Y. Takei and A. Yagi 1. Received February 22, 2006; revised April 6, 2006

Apparent paradoxes in disease models with horizontal and vertical transmission

(x k ) sequence in F, lim x k = x x F. If F : R n R is a function, level sets and sublevel sets of F are any sets of the form (respectively);

UNIFORM WEAK IMPLIES UNIFORM STRONG PERSISTENCE FOR NON-AUTONOMOUS SEMIFLOWS

Locally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem

arxiv: v1 [math.ds] 11 Feb 2011

An introduction to Mathematical Theory of Control

Epidemics in Two Competing Species

Mathematical Analysis of Epidemiological Models: Introduction

A Note on Some Properties of Local Random Attractors

NOTES ON VECTOR-VALUED INTEGRATION MATH 581, SPRING 2017

Introduction and Preliminaries

Mathematics for Economists

Math 5052 Measure Theory and Functional Analysis II Homework Assignment 7

Geometry and topology of continuous best and near best approximations

LECTURE 15: COMPLETENESS AND CONVEXITY

6. Age structure. for a, t IR +, subject to the boundary condition. (6.3) p(0; t) = and to the initial condition

Mathematical Epidemiology Lecture 1. Matylda Jabłońska-Sabuka

Lecture Notes Introduction to Ergodic Theory

Research Article An Impulse Model for Computer Viruses

Convexity in R n. The following lemma will be needed in a while. Lemma 1 Let x E, u R n. If τ I(x, u), τ 0, define. f(x + τu) f(x). τ.

A note on the monotonicity of matrix Riccati equations

Stochastic modelling of epidemic spread

GLOBAL ATTRACTIVITY IN A CLASS OF NONMONOTONE REACTION-DIFFUSION EQUATIONS WITH TIME DELAY

SYNCHRONIZATION OF NONAUTONOMOUS DYNAMICAL SYSTEMS

Banach Spaces V: A Closer Look at the w- and the w -Topologies

THE SEMIGROUP βs APPLICATIONS TO RAMSEY THEORY

Nonlinear Dynamical Systems Eighth Class

Lecture 4. Chapter 4: Lyapunov Stability. Eugenio Schuster. Mechanical Engineering and Mechanics Lehigh University.

Input to state Stability

Global Attractors in PDE

Chain transitivity, attractivity and strong repellors for semidynamical systems

Modeling and Global Stability Analysis of Ebola Models

macroeconomics M. R. Grasselli Research in Options, December 10, 2012

Notes for Functional Analysis

We will begin our study of topology from a set-theoretic point of view. As the subject

macroeconomics M. R. Grasselli University of Technology Sydney, February 21, 2013

Nonlinear Dynamical Systems Lecture - 01

4 Countability axioms

LYAPUNOV EXPONENTS AND PERSISTENCE IN DISCRETE DYNAMICAL SYSTEMS

Measure Theory on Topological Spaces. Course: Prof. Tony Dorlas 2010 Typset: Cathal Ormond

A NONLOCAL REACTION-DIFFUSION POPULATION MODEL WITH STAGE STRUCTURE

Notes for Functional Analysis

Part III. 10 Topological Space Basics. Topological Spaces

Fragmentability and σ-fragmentability

ASYMPTOTICALLY STABLE EQUILIBRIA FOR MONOTONE SEMIFLOWS. M.W. Hirsch. (Communicated by Aim Sciences)

1 Stochastic Dynamic Programming

2 (Bonus). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

Applied Analysis (APPM 5440): Final exam 1:30pm 4:00pm, Dec. 14, Closed books.

The death of an epidemic

7 Complete metric spaces and function spaces

Global Stability for Mixed Monotone Systems

On the stabilizing effect of specialist predators on founder-controlled communities

Stochastic Viral Dynamics with Beddington-DeAngelis Functional Response

Introduction to SEIR Models

LYAPUNOV STABILITY OF CLOSED SETS IN IMPULSIVE SEMIDYNAMICAL SYSTEMS

Exponential stability of families of linear delay systems

Research Article A Delayed Epidemic Model with Pulse Vaccination

PROBLEMS. (b) (Polarization Identity) Show that in any inner product space

BERNOULLI ACTIONS AND INFINITE ENTROPY

A REMARK ON THE GLOBAL DYNAMICS OF COMPETITIVE SYSTEMS ON ORDERED BANACH SPACES

Integration on Measure Spaces

Competitive Exclusion in a Discrete-time, Size-structured Chemostat Model

MATH 202B - Problem Set 5

Part V. 17 Introduction: What are measures and why measurable sets. Lebesgue Integration Theory

Passivity-based Stabilization of Non-Compact Sets

Takens embedding theorem for infinite-dimensional dynamical systems

Introduction: What one must do to analyze any model Prove the positivity and boundedness of the solutions Determine the disease free equilibrium

INTRODUCTION TO FURSTENBERG S 2 3 CONJECTURE

ME 680- Spring Representation and Stability Concepts

Lecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University

Introduction to Empirical Processes and Semiparametric Inference Lecture 08: Stochastic Convergence

Locally Lipschitzian Guiding Function Method for ODEs.

GLOBAL STABILITY OF SIR MODELS WITH NONLINEAR INCIDENCE AND DISCONTINUOUS TREATMENT

18.175: Lecture 2 Extension theorems, random variables, distributions

Probability and Measure

Chapter 2 Convex Analysis

The effect of population dispersal on the spread of a disease

We denote the space of distributions on Ω by D ( Ω) 2.

CHAPTER V DUAL SPACES

Convergence of Feller Processes

Homework Assignment #5 Due Wednesday, March 3rd.

Global Stability of SEIRS Models in Epidemiology

macroeconomics M. R. Grasselli UMass - Amherst, April 29, 2013

This chapter contains a very bare summary of some basic facts from topology.

PERSISTENCE AND GLOBAL STABILITY FOR A CLASS OF DISCRETE TIME STRUCTURED POPULATION MODELS. Hal L. Smith and Horst R. Thieme

II - REAL ANALYSIS. This property gives us a way to extend the notion of content to finite unions of rectangles: we define

1 Stat 605. Homework I. Due Feb. 1, 2011

A SHORT INTRODUCTION TO BANACH LATTICES AND

3 (Due ). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

Weak convergence and Brownian Motion. (telegram style notes) P.J.C. Spreij

2 Statement of the problem and assumptions

PREPUBLICACIONES DEL DEPARTAMENTO DE MATEMÁTICA APLICADA UNIVERSIDAD COMPLUTENSE DE MADRID MA-UCM

General Convexity, General Concavity, Fixed Points, Geometry, and Min-Max Points

EXPOSITORY NOTES ON DISTRIBUTION THEORY, FALL 2018

LMI Methods in Optimal and Robust Control

Topology. Xiaolong Han. Department of Mathematics, California State University, Northridge, CA 91330, USA address:

Transcription:

Global compact attractors and their tripartition under persistence Horst R. Thieme (joint work with Hal L. Smith) School of Mathematical and Statistical Science Arizona State University GCOE, September 2009 H. R. Thieme (ASU) GCOE 1 / 37

Overview Dynamical systems, semiflows dynamics of populations (human, animal, plant) populations persistence (survival of the population) (global) compact attractors How do persistence and compact attractors interact? The existence of a global compact attractor facilitates persistence. Persistence divides the global compact attractor into three parts: extinction attractor, persistence attractor, transient set. Dynamical Systems and Population Persistence American Mathematical Society, Graduate Series, 2011/12 H. R. Thieme (ASU) GCOE 2 / 37

Semiflows and their state spaces The temporal development of a natural or artificial system can conveniently be modeled by a semiflow. A semiflow consists of a state space, X, a time-set, J, and a map, Φ. The state space X comprehends all possible states of the system: the amounts or densities of the system parts; if structure, their structural distribution. epidemiological system: the amounts or densities of susceptible and infective and possibly exposed and removed individuals. For spatial spread, spatial distributions age-structure: age-distributions H. R. Thieme (ASU) GCOE 3 / 37

Time set time-set J: R + = [0, ) or Z + = N {0} = {0,1,...}. Definition A subset J [0, ) is called a time-set if it has the following properties: 1 0 J and 1 J. 2 If s,t J, then s + t J. 3 If s,t J, and s < t, then t s J. J is a time-set iff Ĵ = J ( J) is a subgroup of (R,+) containing Z. If J is closed, then J = [0, ) or J = {m/n;m Z + } where n N. Depending on the model, the time unit can be a year, month, or day. H. R. Thieme (ASU) GCOE 4 / 37

Semiflow map Φ : J X X. Often Φ itself is called the semiflow. If x X is the initial state of the system (at time 0), then Φ(t,x) is the state at time t. Φ(0,x) = x, x X. Further, semiflows are characterized by the semiflow property: Φ(t + r,x) = Φ(t,Φ(r,x)), r,t J, x X. Write Φ t (x) = Φ(t,x), Φ t : X X, Φ t Φ r = Φ t+r. H. R. Thieme (ASU) GCOE 5 / 37

An endemic model with vaccination and return to susceptibility We consider a population of constant size, proportion of susceptible individuals at time t, S(t), of infected individuals, I(t), of removed individuals, R(t), and of vaccinated individuals, V (t), 1 = S(t) + I(t) + R(t) + V (t). S =µ µs κs σsi + θr + ηv, I =σsi (γ + µ)i, R =γi (θ + µ)r, V =κs (η + µ)v. H. R. Thieme (ASU) GCOE 6 / 37

The solution semiflow I =σ(1 I R V )I (γ + µ)i, R =γi (θ + µ)r, V =κ(1 I R) (κ + η + µ)v. (1) The solutions to this system induce a continuous semiflow on X = {(I,R,V ) R 3 +; I + R + V 1}. (2) Solution semiflow Φ(t,x 0 ) = x(t) = (I(t),R(t),V (t)), x(0) = x 0. H. R. Thieme (ASU) GCOE 7 / 37

Persistence Does the dynamical system persist (remains safely away from extinction) as a whole or at least in parts (which parts?). mathematically formulated by using a persistence function ρ : X R +. (3) For x X, ρ(x) is the abundance of the part of the system that is of particular interest. In our example, host persistence is automatic. For disease persistence, we choose ρ(i,r,v ) = I. We could also choose ρ(i,r,v ) = I + R. H. R. Thieme (ASU) GCOE 8 / 37

Uniform persistence The semiflow Φ is uniformly ρ-persistent if there exists some ǫ > 0 s.t. lim inf t ρ(φ(t,x)) ǫ whenever x X,ρ(x) > 0. (4) Φ uniformly weakly ρ-persistent if (4) holds with lim sup replacing lim inf. OVERALL ASSUMPTIONS X is a metric space. ρ is continuous ρ Φ is continuous. Lotka-Volterra predator-prey model: P = ξp PQ, Q = PQ µq is uniformly weakly, but not uniformly persistent. H. R. Thieme (ASU) GCOE 9 / 37

Persistence with a compact attracting set Theorem Let X be a metric space. Assume there exists a compact set A s.t. Φ(t,x) A as t for every x X with ρ(x) > 0 there are no y A, s,t J: ρ(y) > 0, ρ(φ(s, y)) = 0, and ρ(φ(t + s, y)) > 0. Then Φ is uniformly persistent if it is uniformly weakly persistent. Φ(t,x) A: for every open set U A there exists some r J s.t. Φ(t,x) U for all t J,t r. Hale, Waltman (1989) H. R. Thieme (ASU) GCOE 10 / 37

Compact attractors Definition Let J be a time-set and Φ : J X X be a semiflow. A set K X is said to attract a set M X, if K and Φ t (M) K as t. We also say that M is attracted by K. K is called an attractor of M, if K is invariant and attracts M. In this situation, we also say that M has the attractor K. Φ t (M) K: for every open set U, K U X, there exists some r J s.t. Φ t (M) U for all t J, t r. K forward invariant: Φ t (K) K for all t J, K backward invariant: Φ t (K) K for all t J, K invariant: Φ t (K) = K for all t J. H. R. Thieme (ASU) GCOE 11 / 37

ω-limit set Definition The ω-limit set of a subset M of X is defined as ω(m) = t J Φ ( J t M ), J t = J [t, ). Obviously, ω(m) is a (possibly empty) closed set. Alternative characterization. Lemma An element x in X satisfies x ω(m) if and only if there are sequences (t j ) in J, t j as j, and (x j ) in M such that Φ(t j,x j ) x as j. H. R. Thieme (ASU) GCOE 12 / 37

A result without continuity Let J be a time-set and Φ : J X X a map, M X. Definition (Sell, You, 2002) Φ is called asymptotically compact on M, if, for any sequences (t i ) in J, t i as i, and (x i ) in M, (Φ(t i,x i )) has a convergent subsequence. Theorem Equivalent: 1 Φ is asymptotically compact on M. 2 M is attracted by a non-empty compact set K X. 3 ω(m) is non-empty, compact, and attracts M. If one and then all of these three statements hold, ω(m) K for every compact K X that attracts M. H. R. Thieme (ASU) GCOE 13 / 37

Compact attractors of individual sets Φ is called state-continuous if each map Φ t is continuous. Theorem Let J be a time-set and Φ : J X X a state-continuous semiflow, M X. Then every backward invariant subset of Φ(J M) is contained in ω(m). If M is attracted by a compact set or, equivalently, Φ is asymptotically compact on M, then M has a compact attractor, namely ω(m). ω(m) is the unique compact attractor of M contained in Φ(J M). If C is a subset of X and attracts ω(m), then C attracts M. H. R. Thieme (ASU) GCOE 14 / 37

Compact attractor of classes of sets We assume that Φ is state-continuous. Definition Let C denote a class of subsets of X. A non-empty, compact, invariant set K X is called the compact attractor of C if K attracts all sets in C. Theorem The following are equivalent for a class C of subsets of X. (a) There exists a compact attractor of C. (b) Φ is asymptotically compact on every set M C and M C ω(c) has compact closure in X. (c) There is a compact set in X that attracts every set M C. If (a), (b) or (c) and then all of them hold, the closure of M C ω(c) is the smallest compact attractor of C. H. R. Thieme (ASU) GCOE 15 / 37

Global compact attractor We do not use the notion of a global attractor because there is no agreement in the literature about this term. Definition Instead we use the following terminology for a non-empty compact invariant set A: If C is the class of singleton sets in X and A attracts C, A is called a compact attractor of points. If C is the class of bounded (compact) sets in X and A attracts C, A is called an (actually the) compact attractor of bounded (compact) sets. A is a (the) compact attractor of neighborhoods of compact sets if every compact set in X has a neighborhood that is attracted by A. H. R. Thieme (ASU) GCOE 16 / 37

Global compact attractor in the literature The term global compact attractor has been used in various ways in the literature: compact attractor of points: Ladyzhenskaya, 1991; Matano, Nakamura, 1997 compact attractor of neighborhoods of compact sets: Sell, You, 2002; Magal, Zhao, 2005 compact attractor of bounded sets: Hale 1989; Diekmann, van Gils, Verduyn Lunel, Walter, 1995 exponential attractor (of points) Eden, Foias, Nicolaenko, Temam (1994) Osaki, Tsujikawa, Yagi, Mimura (2002) H. R. Thieme (ASU) GCOE 17 / 37

Applicable concepts for global compact attractors Definition (Hale, 1989) Let Φ : J X X be a state-continuous semiflow. Φ is called point-dissipative (or ultimately bounded) if there exists a bounded subset B of X which attracts all points in X. Φ is called asymptotically smooth if Φ is asymptotically compact on every forward invariant bounded closed set. Φ is called eventually bounded on a set M X if Φ(J r M), J r = J [r, ), is bounded for some r J. If X is a closed subset of R n, Φ is asymptotically smooth. More generally, if Φ r is compact on the metric space X for some r J, Φ is asymptotically smooth. H. R. Thieme (ASU) GCOE 18 / 37

Applicable conditions for global compact attractors Theorem Let Φ : J X X be a state-continuous semiflow. Assume that Φ is point-dissipative and asymptotically smooth. Then there exists a compact attractor of points, namely the closure Ω(X) of x X ω(x). Next result has been inspired by [Magal, Zhao, 2005]. Also [Hale, 1989]. Theorem The following are equivalent for a state-continuous semiflow Φ. (a) Φ is point-dissipative, asymptotically smooth, and eventually bounded on every compact subset K of X. (b) There exists a compact attractor A of neighborhoods of compact sets in X; A attracts every subset of X on which Φ is eventually bounded. H. R. Thieme (ASU) GCOE 19 / 37

Attractors of bounded sets Theorem (Hale, 1989) The following statements are equivalent: 1 There exists a compact attractor of bounded sets (which is unique and contains every bounded backward invariant set). 2 Φ is point-dissipative, asymptotically smooth, and eventually bounded on every bounded set in X. There are attractors of neighborhoods of compact sets that are not attractors of bounded sets. infinite-dimensional examples: Magal, Zhao (2005) planar example: endemic model S =βs + qβi µs SI, I = SI (µ + α)i. β > µ > 0, 0 < qβ < µ + α; state space: S > 0, I > 0. H. R. Thieme (ASU) GCOE 20 / 37

Attractors and stability Theorem Let A be a compact attractor of points in X and A be stable. Then A is the compact attractor of neighborhoods of compact sets. Assume Φ be state-continuous, uniformly in finite time: Φ t (x) is continuous in x uniformly for t in bounded subsets of J. Theorem Let A be the compact attractor of compact sets in X. Then A is the compact attractor of neighborhoods of compact sets in X. Further A is stable. Open Problem: Is asymptotic smoothness necessary for existence of a compact attractor of compact sets? H. R. Thieme (ASU) GCOE 21 / 37

Global attractors are connected Theorem Let Φ be state-continuous. (a) If K is the compact attractor of a connected set B, K B X, then K is connected. (b) Let X be the closed convex subset of a Banach space. If K X is the compact attractor of compact sets in X, then K is connected. Proof. (b) The closed convex hull of K is compact by a theorem by Mazur and thus attracted by K. H. R. Thieme (ASU) GCOE 22 / 37

Total trajectories Recall Ĵ = J ( J). Let Φ be a semiflow on X. Definition φ : Ĵ X is a total Φ-trajectory if Φ(t,φ(s)) = φ(t + s), t J,s Ĵ. In ODEs, a total trajectory is a solution that exists for all times. Theorem A X is invariant if and only if for every x A there is a total Φ-trajectory φ : Ĵ A with φ(0) = x. H. R. Thieme (ASU) GCOE 23 / 37

Attractors and persistence Let J = R + or J = Z + = {0} N. Assume that the semiflow Φ : J X X is state-continuous, ρ : X R + is continuous and ρ Φ is continuous. (H0) Φ has a compact attractor A which attracts all points in X. (H1) There exists no total trajectory φ with range in A such that ρ(φ(0)) = 0 and ρ(φ( r)) > 0 and ρ(φ(t)) > 0 for some r,t J. The following set is closed and forward invariant (though possibly empty), X 0 = {x X; t J : ρ(φ(t,x)) = 0}. Recall that Φ is uniformly weakly ρ-persistent, if there exists η > 0 s.t. lim supρ(φ t (x)) > η, whenever ρ(x) > 0, t and is uniformly ρ-persistent if we can replace lim sup by lim inf above. H. R. Thieme (ASU) GCOE 24 / 37

An attractor of points facilitates persistence The following result says roughly that uniform weak persistence plus a compact attractor of points implies uniform strong ρ-persistence. Theorem If X 0 =, then ρ(x) > 0 for all x A and there exists some η > 0 such that lim inf t ρ(φ(t,x)) η for all x X. Theorem If X 0 and Φ is uniformly weakly ρ-persistent, then Φ is uniformly ρ-persistent. H. R. Thieme (ASU) GCOE 25 / 37

Attractors and persistence J = Z + or J = R +, the semiflow Φ : J X X is state-continuous, uniformly in finite time. Assume that Φ has a compact attractor, A, of compact sets in X. A is the compact attractor of neighborhoods of compact sets in X. Definition Y X is called uniformly ρ-positive if inf ρ(y ) > 0. The semiflow Φ is eventually uniformly ρ-positive on D X if Φ(J r D) is uniformly ρ-positive for some r J where J r = {t J : t r}. H. R. Thieme (ASU) GCOE 26 / 37

The partition theorem Assume that X 0, Φ is uniformly weakly ρ-persistent, and (H1) there exists no total Φ-trajectory φ with range in A such that ρ(φ( r)) > 0, ρ(φ(0)) = 0, and ρ(φ(s)) > 0 with r,s J, Then the attractor A is the disjoint union A = A 0 C A 1 of three invariant sets A 0, C, and A 1. A 0 and A 1 are compact and (a) A 0 = A X 0 is the compact attractor of neighborhoods of compact subsets in X 0. A 0 attracts every subset of X 0 that is attracted by A. H. R. Thieme (ASU) GCOE 27 / 37

Partition theorem continued (b) A 1 is uniformly ρ-positive and is the compact attractor of neighborhoods of compact sets in X \ X 0. A 1 is stable. A 1 also attracts all sets that are attracted by A and on which Φ is eventually uniformly ρ-positive. (c) If x X \ A 1 and φ is a total Φ-trajectory through x with pre-compact range, then φ(t) A 0 as t. If x X \ A 0 and φ is total Φ-trajectory through x with pre-compact range, then φ(t) A 1 as t. C consists of those points x A through which there exists a total trajectory φ with φ( t) A 0 and φ(t) A 1 as t. H. R. Thieme (ASU) GCOE 28 / 37

Connectedness of the persistence attractor A 1 persistence attractor of Φ, A 0 the extinction attractor of Φ. Existence of persistence attractor: Hale&Waltman 1990, Zhao 2003, Magal&Zhao 2005. Theorem Let X be the closed convex subset of a Banach space. Let the assumptions of the Partition Theorem be satisfied and the persistence function ρ be concave. Then the persistence attractor, A 1, is connected. Proof. If ρ is concave, the closed convex hull of A 1 is uniformly ρ-positive (and compact) and thus attracted by A 1. H. R. Thieme (ASU) GCOE 29 / 37

Existence of non-trivial fixed points via persistence Assumptions: J = Z + or J = R + and Φ is continuous. X is the closed convex subset of a Banach space, ρ : X R + is continuous and concave. Φ is uniformly weakly ρ-persistent, has a compact attractor of compact sets in X. { t = 1 if J = Z+ Φ t is compact (or condensing) for. t (0,1] if J = R + (H1) there exists no total Φ-trajectory with pre-compact range such that ρ(φ( r)) > 0, ρ(φ(0)) = 0, and ρ(φ(s)) > 0 with r,s J. Conclusion Then there exists x X with ρ(x ) > 0 and Φ(t,x ) = x for all j J. [Similar results by Zhao (2003) and Magal and Zhao (2005).] H. R. Thieme (ASU) GCOE 30 / 37

Back to the endemic model with vaccination I =σ(1 I R V )I (γ + µ)i, R =γi (θ + µ)r, V =κ(1 I R) (κ + η + µ)v. state space X = {(I,R,V ) R 3 + ; I + R + V 1}. X is compact, so solution semiflow Φ has a compact attractor of bounded sets in X (including X itself). disease persistence ρ(i, R, V ) = I. X 0 = {(0,R,V ); R 0, V 0, V + R 1}. X 0 has a compact attractor of bounded sets, A 0. Since X 0 invariant (not only forward invariant), A 0 = A X 0. H. R. Thieme (ASU) GCOE 31 / 37

Disease-free dynamics Consider bounded everywhere defined solutions of I 0, R = (θ + µ)r, V =κ(1 R) (κ + η + µ)v. Then R 0 and V κ κ + η + µ =: V. Theorem The disease free equilibrium (0,0,V ) attracts X 0. H. R. Thieme (ASU) GCOE 32 / 37

Disease persistence Theorem Let lim inf t V κ (κ + η + µ)v. V V, R I θ + µ. I (t) I(t) σ(1 I R V ) (γ + µ). 1 < σ(1 V ) γ+µ =: R. (net replacement ratio) Then the disease is uniformly persistent: there exists some ǫ > 0 such that for all solutions with I(0) > 0. lim inf t I(t) ǫ H. R. Thieme (ASU) GCOE 33 / 37

If R > 1, there is a disease persistence attractor. If R 1, there is still the overall attractor. To describe them, consider everywhere defined bounded solutions. Transform to an integral equation, with ( I (t) = I(t) σ 0 ) I(t s)m(ds), t R, σ = σ(1 V ) (γ + µ) = (R 1)(γ + µ). Let m be a signed Borel measure on R +, m({0}) 0, whose variation m satisfies R + (1 + s) m (ds) < and whose Fourier transform satisfies inf R e ist m(dt) > 0. s 0 R + H. R. Thieme (ASU) GCOE 34 / 37

A result for integro-differential equations Theorem If σ 0, every bounded non-negative solution is identically 0. If σ > 0, every positive solution that is bounded and bounded away from 0 is constant. Proof: Adapt Fourier transform techniques by London (1983) to everywhere defined solutions. For σ > 0 another transformation is required. Set I σ = m(r + ), Set u(t) = ln I(t) I, u (t) = I (t) I(t) = 0 0 (I I(t s))m(ds). g(u(t s))m(ds), g(u) = I (e u 1). H. R. Thieme (ASU) GCOE 35 / 37

Global stability of equilibria in the endemic model Theorem If R 1, then the disease-free equilibrium attracts the whole state space X and is stable. If R > 1, there is a unique endemic equilibrium. It attracts every subset X {I ǫ} and is stable. H. R. Thieme (ASU) GCOE 36 / 37

Conclusion A global compact attractor facilitates persistence. Persistence creates a tripartition of the global compact attractor. This is a stepping-stone for proving global asymptotic stability (in a uniform sense) of equilibria. H. R. Thieme (ASU) GCOE 37 / 37