of the Schnakenberg model

Similar documents
MS: Nonlinear Wave Propagation in Singular Perturbed Systems

Stationary radial spots in a planar threecomponent reaction-diffusion system

Pattern formation in Nikolaevskiy s equation

The Dynamics of Reaction-Diffusion Patterns

Self-Replication, Self-Destruction, and Spatio-Temporal Chaos in the Gray-Scott Model

1. < 0: the eigenvalues are real and have opposite signs; the fixed point is a saddle point

HOPF DANCES NEAR THE TIPS OF BUSSE BALLOONS. Arjen Doelman. (Communicated by the associate editor name)

system May 19, 2009 MS69: New Developments in Pulse Interactions SIAM Conference on Applications of Dynamical Systems Snowbird, Utah, USA

ME 680- Spring Geometrical Analysis of 1-D Dynamical Systems

system CWI, Amsterdam May 21, 2008 Dynamic Analysis Seminar Vrije Universiteit

Computational Methods in Dynamical Systems and Advanced Examples

Metastability for the Ginzburg Landau equation with space time white noise

LOCALIZED PATTERNS OF THE CUBIC-QUINTIC SWIFT-HOHENBERG EQUATIONS WITH TWO SYMMETRY-BREAKING TERMS. Zhenxue Wei

Solutions for B8b (Nonlinear Systems) Fake Past Exam (TT 10)

Group Method. December 16, Oberwolfach workshop Dynamics of Patterns

One Dimensional Dynamical Systems

Lecture 7: The Swift-Hohenberg equation in one spatial dimension

MAE294B/SIOC203B: Methods in Applied Mechanics Winter Quarter sgls/mae294b Solution IV

B5.6 Nonlinear Systems

Phase Synchronization

An Application of Perturbation Methods in Evolutionary Ecology

Clearly the passage of an eigenvalue through to the positive real half plane leads to a qualitative change in the phase portrait, i.e.

Multiscale Analysis of Many Particle Systems with Dynamical Control

Nonlinear convective stability of travelling fronts near Turing and Hopf instabilities

APPLIED PARTIM DIFFERENTIAL EQUATIONS with Fourier Series and Boundary Value Problems

Numerical techniques: Deterministic Dynamical Systems

Looking Through the Vortex Glass

Lecture 10: Singular Perturbations and Averaging 1

BIFURCATION TO TRAVELING WAVES IN THE CUBIC-QUINTIC COMPLEX GINZBURG LANDAU EQUATION

Projection Methods. Michal Kejak CERGE CERGE-EI ( ) 1 / 29

tutorial ii: One-parameter bifurcation analysis of equilibria with matcont

ENGI 9420 Lecture Notes 4 - Stability Analysis Page Stability Analysis for Non-linear Ordinary Differential Equations

Survey of strong normal-internal k : l resonances in quasi-periodically driven oscillators for l = 1, 2, 3.

Destabilization mechanisms of periodic pulse patterns near a homoclinic limit

The Stability and Dynamics of a Spike in the One-Dimensional Keller-Segel model

Pattern formation in reaction-diffusion systems an explicit approach

Applied Asymptotic Analysis

Quasipatterns in surface wave experiments

LMI Methods in Optimal and Robust Control

Exam in TMA4195 Mathematical Modeling Solutions

Resonant excitation of trapped coastal waves by free inertia-gravity waves

Continuation of cycle-to-cycle connections in 3D ODEs

Dispersion relations, stability and linearization

Solution to Homework #4 Roy Malka

ADIABATIC STABILITY UNDER SEMI-STRONG INTERACTIONS: THE WEAKLY DAMPED REGIME

arxiv: v1 [math.ap] 8 Feb 2018

Chapter 9: Differential Analysis of Fluid Flow

Destabilization mechanisms of periodic pulse patterns near a homoclinic limit

VIII. Phase Transformations. Lecture 38: Nucleation and Spinodal Decomposition

Review for Exam 2 Ben Wang and Mark Styczynski

Synchronization Transitions in Complex Networks

Continuum Modeling of Transportation Networks with Differential Equations

1 Introduction Self-replicating spots and pulses have been observed in excitable reaction-diusion systems [22, 17, 24, 23, 16, 9, 2, 3, 4, 25, 21, 19,

Mathematical Foundations of Neuroscience - Lecture 7. Bifurcations II.

CANARDS AND HORSESHOES IN THE FORCED VAN DER POL EQUATION

NONLINEAR PATTERNS IN URBAN CRIME - HOTSPOTS, BIFURCATIONS, AND SUPPRESSION

Lecture 10: Finite Differences for ODEs & Nonlinear Equations

Exponentially small splitting of separatrices of the pendulum: two different examples. Marcel Guardia, Carme Olivé, Tere M-Seara

Ordinary Differential Equations

Bifurcation Analysis of Non-linear Differential Equations

Nonlinear Control. Nonlinear Control Lecture # 2 Stability of Equilibrium Points

Topic # /31 Feedback Control Systems. Analysis of Nonlinear Systems Lyapunov Stability Analysis

CDS 101/110a: Lecture 2.1 Dynamic Behavior

Half of Final Exam Name: Practice Problems October 28, 2014

Stability in the sense of Lyapunov

Dimensional Analysis - Concepts

Chapter 9: Differential Analysis

B5.6 Nonlinear Systems

4 Insect outbreak model

Lecture 19: Solving linear ODEs + separable techniques for nonlinear ODE s

Lecture 3 : Bifurcation Analysis

Breakdown of Pattern Formation in Activator-Inhibitor Systems and Unfolding of a Singular Equilibrium

Time-dependent variational forms

DIFFERENTIAL EQUATIONS

Turning points and traveling waves in FitzHugh-Nagumo type equations

Introduction to multiscale modeling and simulation. Explicit methods for ODEs : forward Euler. y n+1 = y n + tf(y n ) dy dt = f(y), y(0) = y 0

Solution to Homework #5 Roy Malka 1. Questions 2,3,4 of Homework #5 of M. Cross class. dv (x) dx

MATH 425, FINAL EXAM SOLUTIONS

Discretized Fast-Slow Systems near Pitchfork Singularities

AM 205: lecture 19. Last time: Conditions for optimality, Newton s method for optimization Today: survey of optimization methods

UPPER AND LOWER SOLUTIONS FOR A HOMOGENEOUS DIRICHLET PROBLEM WITH NONLINEAR DIFFUSION AND THE PRINCIPLE OF LINEARIZED STABILITY

Notes for Expansions/Series and Differential Equations

BIFURCATION AND OF THE GENERALIZED COMPLEX GINZBURG LANDAU EQUATION

Topics Covered in Calculus BC

Dispersion relations, linearization and linearized dynamics in PDE models

INTRODUCTION TO PDEs

Homework Solutions:

Numerical Methods for ODEs. Lectures for PSU Summer Programs Xiantao Li

Partial Differential Equations

A Ginzburg-Landau Type Problem for Nematics with Highly Anisotropic Elastic Term

Travelling-wave spatially periodic forcing of asymmetric binary mixtures

The semi-geostrophic equations - a model for large-scale atmospheric flows

Solution of a Fourth Order Singularly Perturbed Boundary Value Problem Using Quintic Spline

13 PDEs on spatially bounded domains: initial boundary value problems (IBVPs)

Finite Difference Methods for Boundary Value Problems

Kramers formula for chemical reactions in the context of Wasserstein gradient flows. Michael Herrmann. Mathematical Institute, University of Oxford

Chapter #4 EEE8086-EEE8115. Robust and Adaptive Control Systems

Dynamics of a mass-spring-pendulum system with vastly different frequencies

Nonlinear Stability, Thermoelastic Contact, and the Barber Condition

Numerical solutions of the small dispersion limit of KdV, Whitham and Painlevé equations

Transcription:

Pulse motion in the semi-strong limit of the Schnakenberg model Newton Institute 2005 Jens Rademacher, Weierstraß Institut Berlin joint work with Michael Ward (UBC) Angelfish 2, 6, 12 months old [Kondo, S. and Asai, R., Nature 1995]

Numerical simulations [Crampin 00] Turing-unstable Schnakenberg model (two species) x t Stable periodic wave train undergoes self-similar sequence of spatial period doublings Logistic growth:

Uniformly exponentially growing line Model for growth rate ρ : additional dilution term [Crampin 00] U t U t + ρ(xu) x Paradigm: growing reaction diffusion system U t = DU xx ρ(xu) x + F (U). Points of view: Weakly nonlinear: Turing instability - not pursued here Strongly nonlinear, here semi-strong limit [Doelman et al 97] Quasi-stationary approximation for slow growth ρ: time as continuation parameter

Slow growth, continuation, onset of splitting Dilution term cancels in rescaling to fixed domain: U t = g(t) 2 DU xx + F (U) ρu, g(t) = e ρt Quasi-stationary approximation: Slow parameter variation of stable solution is continuation, splitting must stem from instability. time independent bifurcation problem in g to find instability. Similarity to previously studied pulse splitting [Reynolds et al 94, Nishiura et al 99] Approximate further to U t = gd xx + F (U)

Interaction and localization u t = u xx + a uv 2 Schnakenberg model v t = δv xx + b + uv 2 v Turing instability: no localization, small sinusoidal pattern, strong interaction but weakly nonlinear. (a = 0.9, b = 0.1, δ 0.11) Semi-strong interaction: v more localized than u. Occurs for δ 0, δ = 0.01, natural limit of Turing case: Weak interaction: same strong localization of u and v. Wave trains near stable pulse - monotone path, no instability no splitting under slow growth.

Splitting in Schnakenberg model For δ = 0.01 numerically: self-similar splitting sequence stable wave trains for large range of periods - not pure modes Not directly related to Turing [Crampin 00] ρ = 0.001, t [1000, 1500], x [0, 1.3]e ρt. instability

Bifurcation and basin of attraction 10 U 9 stable 2 pulse 3.5 3.0 2.5 2.0 1.5 1.0 8 7 0.5 0.0 0.00 0.20 0.40 0.60 0.80 1.00 0.10 0.30 0.50 0.70 0.90 3.5 3.0 2.5 6 5 stable 1 pulse part of basin of attraction 2.0 1.5 1.0 0.5 4 0.0 0.00 0.20 0.40 0.60 0.80 1.00 0.10 0.30 0.50 0.70 0.90 3 0.6 0.8 1 1.2 1.4 1.6 1.8 units of Turing pattern period Quasi-stationary approximation breaks down at the fold point - splitting stems from basin of attraction of 2-pulse... Similar to other semi-strong splitting [Nishura, Doelman...]. But n 2n splitting, n 2: more solutions available

Growth of Turing pattern Near Turing instability solutions stay close to those of real Ginzburg-Landau equation: A t = A xx + µa A A 2 Modulated Turing patterns: A k (x)e ix = r(k, µ)e i(1+k)x, k 2 µ. Domain [0, 2π/g], Neumann b.c. become sin((1 + k)2π/g) = 0 k = gn/2 1, n Z Stable against e ilx for 3k 2 l 2 /2 < µ Interested in solution with k = g 1 as g. Eckhaus instability: g = 1 µ/3 t(k, l). growth µ unstable stable k [Barkley,Tuckerman 90]: instability is a subcritical pitchfork. Quasi-stationary: splitting sequence seems not very robust

Motion of pulses Building block of splitting: motion of pulses when placed away from symmetric, stationary location. Idea: Seek expression for instantaneous speeds Objectives: determine speed depending on location and domain size seek stable equilibria for quasi-stationary approximation relation to splitting onset Ansatz: formal asymptotics in semi-strong limit for slow motion

Formal asymptotics for spike speed Rescale to x ( 1, 1) via ɛ = δ/l: u t = 1/L 2 u xx + a uv 2 v t = ɛ 2 v xx + b + uv 2 v Spike position ansatz: x 0 = x 0 (ɛt), v x (x 0 ) = 0, speed c = ẋ 0. [Doelman et al 00]: Gray-Scott model on R, asymptotic ODE for c Here: obtain asymptotic boundary value problem for c cheap numerics, computations possible beyond single pulse.

Inner problem Rescale: τ = ɛt ɛy = x x 0 (τ) = d/dy = d/dτ v = (Lɛ) 1 V u = Lɛ U ɛ V ẋ 0 V = V V + UV 2 + blɛ ɛ 3 L 2 U ɛl 2 ẋ 0 U = U UV 2 + alɛ For ɛ = 0: asymmetric core problem (similar to Gray-Scott) cv 0 = V 0 V 0 + U 0 V0 2 0 = U 0 U 0 V0 2 Boundary conditions of V 0 : V 0(0) = 0, V 0 0 as y. Boundary conditions of U 0 via outer problem.

Outer problem Distributional approximation of nonlinear term: ( ) x x0 uv 2 = (Lɛ) 1 (U 0 V 0 + O(ɛ)) ɛ 1 f ɛ R f(y)dyδ x0 (x) uv 2 = A/Lδ x0 (x) + O(ɛ), A = R U 0V 2 0 dy. Substitute in u equation: L 2 u xx + a uv 2 = 0 L 2 u xx + a A/Lδ x0 (x) = O(ɛ) ɛ = 0 u xx + al 2 ALδ x0 (x) = 0

Outer problem Approximate u equation: u xx + al 2 ALδ x0 (x) = 0. Integrate: 1 1 u xx dx + 2aL 2 = AL u x 1 1 + 2aL 2 = AL a = A/2L Approximate u equation: u xx + AL/2 ALδ x0 (x) = 0. Solve by Green s function G xx = 1/2 δ x0 (x): u = u c ALG(x; x 0 )

Matching Matching one-sided Taylor series (recall u = ɛlu): u c ALG(x 0 ; x 0 ) (x x 0 )ALG x (x ± 0 ; x 0) ɛlu(y) x x 0 = ɛy u c = ALG(x 0 ; x 0 ) U 0 (y) AG x (x ± 0 ; x 0)y as y ± Here G(x; x 0 ) = x 2 /4 x x 0 /2 + G 0 G x (x ± 0 ; x 0) = (x 0 1)/2 Hence b.c. for U 0 : U 0(± ) = ±A/2(1 x 0 ) = ±al(1 x 0 )

Boundary conditions for U 0 B.c. for U 0 : U 0(± ) = ±al(1 x 0 ) outer region inner region asymptotic b.c. O( ε)

Boundary value problem for speed V + cv + UV 2 = V V U = UV 2 (0) = 0 V (± ) = 0 U (± ) = ±al(1 x 0 ) Spike dynamics: solve BVP and ẋ 0 = c with L = L(ρt). Expect validity for 1 x 0 > O(ɛ), ρ = O(ɛ). Quasi-stationary: spike motion determined by c(x 0, al), in particular the equilibrium curves c(x 0, al) = 0. Symmetry: c( x 0, al) = c(x 0, al) c(0, al) = 0. Self-similarity: same BVP arises for motion of single spike in chain of n spikes, e.g. n = 2 on ( 1, 0): x 0 = (x 0 1)/2, L = 2L.

Computation Continuation with AUTO [Doedel et al]: Alternate between solving BVP and updating x 0, L Map out c(x 0, al) and its domain of definition. Note: V (0) = 0 is interior condition. Resolved by splitting domain into ( 1, 0) for (U, V ) and (0, 1) for (U +, V + ) with b.c.s U + (0) = U (0), V + (0) = V (0) and V ±(0) = 0.

Spike motion before splitting Quasi-stationary: onset of splitting is fold point of spike (x 0 = 0) at al 1.347 [Muratov, Osipov 00,...], stability [Ward et al 03]. c(x 0, al) undefined for al > 1.347. spike should move to x 0 = 0 for small x 0. 0.9 Indeed al = 1: Motion for ρ = 0 and ρ = 0.01: 0.8 0.7 0.6 x 0 0.5 0.4 0.3 0.2 0.1 0 0 1 2 3 4 5 6 7 8 t

Singularities of c(x 0, al) for 1-spikes Onset of splitting: fold point at al 1.347. Continuation of bifurcating branch of fold points in (al, x 0, c): 0 x 0-0.2 fold of 1 spikes Outside of range of validity: -0.4 x = 1 0 U/10-0.6-0.8-1 shifted 1 spikes Invalid in O( ε) strip 1 1.2 1.4 1.6 1.8 2 al motion V

c 0.10 0.00-0.10-0.20 Extending c - other bifurcations x = 0 0 fold of 1 spikes 0.10 x 0 0.00-0.10 2 spikes c=0 c>0 fold of 1 spikes -0.30-0.40 (U/50) -0.20 c=0-0.50 * -0.30-0.60-0.40 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1.90 0.50 2.00 0.75 1.00 1.25 1.50 1.75 2.00 al al Relevance for splitting? If stable: growth is slow motion along curve c = 0... Expect unstable: PDE dynamics after onset is motion to x 0 = ±1.

Todo check against PDE simulations stability of components make (partially) rigorous... (geometric theory [Doelman et al]?, monotonicity) relation to previous work on velocity for Gray-Scott model asymptotics on half-line: onset of splitting?

Conclusion Quasi-stationary approximation of domain length links to previous splitting studies Semi-strong limit yields BVP for spike velocity Motion as expected even from x 0 = ±1 Approach invalid at onset of splitting: speed large and location at boundary Bifurcations at onset of splitting fold curve, velocities towards x 0 = 0 other solutions moving away from x 0 = 0 Bifurcation of c = 0 curve off 2-spike before onset - should be unstable solutions