Nonlinear dynamics and pattern formation w ith applications to ecology M eiji U niversity, N ov. 25-28, 2008
Ehud M eron Blaustein Institutes for D esert R esearch & Physics D epartm ent Ben G urion U niversity, Israel
T he w orld w e perceive w ith our senses is highly nonlinear: Clim ate phenom ena (tornados, hurricans), patterns in nature (clouds, vegetation), oscillations and synchronization (heart beats, hands clapping, stock-m arket dynam ics), are all nonlinear phenom ena. T o study phenom ena of this kind w e need the tools of dynam ical system theory
Focus on pattern formation phenomena, their nonlinear aspects, and their relevance to the ecology of w ater-limited systems G eneral outine B
Part I: Basic m echanism s of pattern form ation e n
G
Basic concepts: dynam ical system s, instabilities, etc. u r i o
Pattern form ing system s n U
1. Class I – instabilities of uniform states n i v 2. Class II – m ultistability of states, front dynam ics e r s i t y ,
Part II: Controling patterns by tem poral and spatial periodic forcing E h
Frequency locking, w avenum ber locking in spatially extended system s u d
Class I and class II pattern form ation m echanism s induced by the M e r
Forcing o n - w
Part III: M ultim ode localized structures w w . Localized structures, m ultiple instabilities, instabilities of single- b g u
m ode to dual-m ode localized structures . a c . i l /
Part IV : A pplications to ecology ~ e h
V egetation pattern form ation in w ater-lim ited system s, plant u d interactions and biodiversity. O utine of Part I: Basic mechanisms of pattern formation B e
Basic concepts: O pen non-equilibrium system s, dynam ical system s, n
G
nonlinear system s, m ultiplicity of states, instabilities, hysteresis u r i o n
Pattern form ation in nature: S ym m etry breaking, universality U n i v e r s
Canonical m odels of pattern form ing system s: S w ift-H ohenberg, i t y ,
FitzH ugh-N agum o
E h u d
Pattern form ing system s class I: Instabilities of uniform states M e
Exam ples, linear stability analysis, nonlinear analysis – am plitude r o equations, secondary instabilities n - w w w
Pattern form ing system s class II: M ultiplicity of stable states . b g u
Exam ples, fronts, front instabilities, labyrinthine patterns, spiral . a c .
w aves, spiral turbulence. i l / ~ e h u d Basic concepts: open non-equilibrium systems B e
T he natural system s w e m entioned represent open non-equilibrium n
G
system s, that continuously exchange m aterials and energy w ith u r i their environm ents. o n U n i O ne of the first tasks in exploring such system s is defining w hat v e r constitutes the “system ” and w hat constitutes the system ’s s i t y
“environm ent”. ,
E h u T he distinction betw een the system and its environm ent depends d
M
on the processes they involve, w hich can be divided into tw o e r o general groups: n - w
1. G roup I: Processes that m utually affect one another through w w . various feedbacks. b g u .
2. G roup II: Processes that affect G roup-I processes but are a c . i
hardly affected by them . l / ~ e h
U sing this process classification w e can introduce the concept of a u dynam ical system . d Basic concepts: defining a dynamical system
D ynamical system - a set of group-I processes, quantified by B e n dynam ical variables, U (t), that evolve in tim e according to som e law s i G u of m otion. T hese law s describe the m utual relationships am ong the r i o dynam ical variables and the m anners by w hich they are affected by n U param eters, P (t), representing group-II processes, w hich define the n
i i v e
system ’s environm ent. r s i t y ,
E
P1(t) h u d
M U1(t) e r o n - w U (t) P2(t) 3 w w .
D ynamical U2(t) b g
S ystem u . a
G roup-I processes c . i l (dynam ical variables) Environment / ~ e h
G roup-II processes (param eters) u d Ben G urion U niversity, Ehud M eron - w w w .bgu.ac.il/~ehud x t
: n v ) s o n i t o i t t n a u l l m l i e o c L s
s o
m v
l
d b a n
e m c
p - o q i e m 0 p
r a n = t y i D
o T v s
. i F
n y g
t
e s c ( m i e
l r , c s . f r a e b e
o
c c x s a , i f r
s r
o i h m e n
f a t c
a
o
i n l i o y d n h w x o w t . a r t i
a t a y i s c n n i l t p i e i c
m e d o x p s r w i b d o
I u
d
l n f l t g L
l m r I n u - e a u 2 a e - u o t - r a
v t i i r d i s h p i l s v e t u ’ n a s t u v
p n c t . n
e a b a o m i i d l e ( a l o m f p r r r a a
- h n t a o n r
e F G G
t
m A = o w
g e n b
. s e : o e n t n n v ) o t i i
t
d m d s e m
N s
t t e r d : n
g e y d n a
) h e o l n y t s e e l i n t t
s i m i b a t s t l
c y
e
c a e a n r l . s s d t r i e r d c l
a x i s o i e j l a p n
e y c , s a l b b a e e m l n ) i s s s c i u r a n i a ( r l e
o e i s
l , n
v c l
n s r ) m m s g y e t s
s o p s = a e i ( n m : d
e . d n
n x s t a
q e x v s u d d d ( s y
e r d L t d e t t
, e a
e i a d s r s = n y l p
e i n c ) y e m a t o p a t e
i a b o i s n u
(
i a c r c c t m s x m l f l i e n
o p n s a a e o : s l
h
e a e t a o t c l , o e I
t i l e d s c
n w m l - / p v s e a y
i
m o e - - x o p p
i w s c a m r d h n d d u
t s i
m s n i e i c a a e o = d s i e a y s ) h r h e e h f n t h h a x o ( D p B o G w a T B p w E T v B Basic concepts: instabilities
N onlinearity can induce instabilities to new states B W > WC e W < W n C T he buckling of a beam : G u r i o
T here is a critical w eight, WC, n U
beyond w hich the beam buckles. n i v e r
Let, s i t y ,
u - m easure the degree of buckling
E h l = (W - W ) W - the deviation from the critical point u
C C d
M e
Close to the critical point the system can be described by the r o n
nonlinear dynam ical system -
3 w w
u# = lu - u = f (u) (u# ô du dt ) u w + . u b g u
(u = 0): .
S teady state solutions a # u0 c . i l / = - the unbuckled state l ~ u0 0 e h u
u d u ê = ê l - tw o sym m etric buckled states - Basic concepts: linear stability
T he steady state solutions tell us that beyond the critical w eight, B e n
or w hen l>0 , the buckled states appear, but they do not tell us G u r
that the system w ill necessarily evolve tow ards these states. T his i o type of inform ation is contained in the stability properties of the n U n
various states. i v e r s
Linear stability: A solution (state) u is stable if any sm all i
s t y , perturbation of decays in the course of tim e. us E h u
A graphical view : A solution us is stable (unstable) if the slope d
3 M f’(us) of u = f (u) = lu - u at us is negative (positive) e # r o n -
l < 0 l > 0 w u# u# w f(u) w . b g u
f(u) . a c . i l / u u ~ e h u d S table U nstable S table Basic concepts: bifurcation diagrams
W e describe such an instability by a l < 0 l > 0 B e n
diagram that show s the various G u steady-state solutions as functions r i o of the control param eter l, and n U n
their stability properties: i v
~ l e r
+ l s i
u t y ,
E h u d
D iagram s of this kind are M
l e called bifurcation diagram s r o n -
T his particular instability w
- l w
is called the pitchfork w . b
Stable Unstable bifurcation g u . a c .
In practice the tw o buckled states need not be perfectly sym m etric, i l / and the instability m ay favor one state over the other. ~ e h
M athem atically, such im perfections are described by additional u d term s that break the sym m etry u ç -u Basic concepts: the cusp singularity
3 2 B
u = lu - u + au + b = 0 e
# n
G
u Cusp singularity u r i o n
b U n i v e r s i t y ,
E h u d
M
l e r o n - w
Cuts at constant b values: w w . b g u .
u u u a c . i l / ~ e h u b > 0 b = 0 b < 0 d l l l Basic concepts: the cusp singularity
3 2 B
u = lu - u + au + b = 0 e
# n
G
u Cusp singularity u r i o n
b U n i v e r s i t y ,
E h u d
M
l e r o n - w
Cuts at constant l values: w w . b g u .
u u u a c . i l / ~ e h u l < 0 l = 0 l > 0 d b b b Basic concepts: hysteresis B e
u n
H ysteresis phenom ena: u+(b) G u r i
D ifferent transition points upon o n
increasing and decreasing b. U n i v
Introduced in the context of e r
u-(b) s m agnetism , exists in num erous i t y ,
other contexts
E h
b b u bmin max d
M e r
Coexistence of o n
S table states - w w
In ecology - catastrophic shifts (S cheffer et al. Nature 2001): w . b g
D esertification – sudden decrease in biological productivity as a u . a c
result of clim ate fluctuations (droughts) and hum an- related . i l / activities (over-grazing). ~ e h
S udden loss of transparency and vegetation in shallow lakes u d subjected to hum an-induced eutrophication. Basic concepts: hysteresis B
H ysteresis in Escher’s art e n
Birds G u r i o n U n i v e r s i t y ,
E
Fish to birds h u d
M
Coexistence e r o of birds & fish n - w
Birds to fish w w . b g u . a c . i l / ~ e h u Fish d Pattern formation in nature B e n
G u r i o n U n i v e r s i t y ,
E h u d
M e r o n - w w w . b g u . a c . i l / ~
o Symmetry breaking e h u o Universality d
o Emergence Salt formation in the Atacama desert (Marcus Hauser) Canonical models of pattern forming systems
N um erous experim ental and theoretical studies support the B e n interpretation of natural patterns as sym m etry breaking phenom ena. G u r i
O f these, studies based on m athem atical m odels have been very o n
instrum ental in understanding the m echanism s underlying these U n i phenom ena. v e r s i
In som e contexts the m odel equations are based on firm theories and t y ,
reproduce observed behaviors w ith quantitative accuracy. Exam ples E h
are the N avier-S tokes equations of fluid dynam ics, and the M axw ell u d
equations in nonlinear optics. M e r o
In m ost other contexts, especially w hen the system ’s com plexity n increases, m athem atical m odels are less accurate. H ow ever, they can - w w
still be very helpful in capturing the qualitative behavior of a system , w . b
and in understanding the specific m echanism s responsible for it. g u . a c .
A long w ith m odels that are geared to reproduce and predict i l / behaviors of specific system s, sim plified m odels, focusing on general ~ e h aspects of pattern form ation rather than on the details of a specific u d system , have appeared. A few of them have becom e canonical m odels. Canonical models of pattern forming systems
T he S w ift-H ohenberg m odel B e n
2 G 3 2 2 u r
u = l u - u - (∞ + k ) u i t c o n U n i v T he FitzH ugh-N agum o m odel e r s i t y
3 2 , 3
u t = l u - u - v + ∞ u E
t h u
2 d
v t = e ( u - a 1 v - a 0 ) + d ∞ v M e r o n -
2 2 w ïu 2 ï ï - Ñ < x < Ñ w w
u = ∞ = + . t , 2 2 b -Ñ < y < Ñ g
ït ïx ïy u . a c . i l / ~ e
u(x,y,0) u(x,y,t) h u v(x,y,0) v(x,y,t) d Classes of pattern forming systems
M odels of this kind help us understand basic m echanism s of pattern B e n
form ation and classify pattern form ing system s accordingly. G u r i o W e w ill distinguish here am ong tw o classes w hich w e illustrate w ith n U
the follow ing m odel sim ulations: n i v e r
S andstede s i t y ,
E h u d
M e r o n - w w w . b g u . Class I: Instabilities Class II: M ultistable Class III: Excitable a c . i l of uniform states system s system s / ~ e h u d Class I: Instabilities of uniform states
T herm al convection (R ayleigh-Bénard): heating a fluid layer from B e n
below G u r i Below a critical tem perature o n
Ttop difference, U n i v
Tbot e d DT = T - T < DT r Flui bottom top C s i t y ,
the fluid is at rest – heat is E h
transferred by m olecular conduction u d
M e
A bove DTC convection sets in the form of ordered parallel roles r o n - w w w . b g u . a c . i l /
Ttop ~ e
Cold h bot > top V isualization by illum inating from Tbot T T u
H ot d fluid below (index of refraction changes fluid w ith tem perature periodically). Class I: Instabilities of uniform states B
W hy do w e have an instability w hen the tem perature difference is e n large enough? G u r
top plate i o n U n i T v e e m r D s p
d e T1+ T i ç e Colder fluid layer heavier t T2 t t y u E r h e T : A fluid layer w ith a hotter fluid particle u 1 T1+dT d M e r o n Bottom plate - w w Im agine a fluctuation in w hich a fluid particle at som e height has a w . b tem perature w hich is slightly higher than the surrounding fluid at g u . that height. Because of therm al expansion the fluid particle w ill a c . i have a density low er than that of the surrounding fluid (i.e. w ill be l / ~ lighter) and w ill tend to m ove upw ard. A s it m oves upw ard the e h u surrounding fluid becom es yet colder and the buoyancy force d upw ards increases. Class I: Instabilities of uniform states W hy does not the instability occur w hen the tem perature B e n difference is not large enough? G u r i T here are processes that stabilize the rest state: o n U Fluid viscosity: induces transfer of linear m om entum from the up- n i v m oving fluid particle to its neighborhood, thus reducing its e r s m om entum and speed. i t y , T herm al conduction: induces diffusion of heat from the fluid E h particle to its colder neighborhood, thus reducing the buoyancy u d force that drives the fluid particle upw ard. M e r o n Ω T here should be a critical - w DT w tem perature difference, c, at w . w hich the stabilizing factors just b g u . balance the destabilizing buoyancy a c . i force. l / ~ e h Bodenschatz lab. 1997: sm all Prandtl u num ber fluid experim ents d Class I: Instabilities of uniform states Farady surface w aves B e n (stripes and squares) G u r i o n U n i v e r s i t y , S tephen W . M orris E h u d Chem ical reactions M e r o n - w w CIM A reaction: w . b T uring patterns g u V igil, O uyang and S w inney . a c . i l / ~ e h u d Class I: Linear stability analysis of the uniform state A pow erful m ethod that provides inform ation about the instability B e n threshold of a uniform state and the nature of the m ode that grow s G u at the instability point is linear stability analysis. Let’s illustrate this r i o m ethod using the S w ift-H ohenberg (S H ) m odel: n U n i 2 v 2 e ï ≈ ï ’ r u 3 2 s i = l u - u - ∆ + k ÷ u - Ñ < x < Ñ t ∆ 2 0 ÷ y , ï t ï x « ◊ E h u d Consider the stationary uniform solution us=0. T his solution is M e linearly stable if any infinitesim ally sm all perturbation decays in the r o n course of tim e. It is linearly unstable if there exists a perturbation - that grow s in tim e. w w w . W riting u(x, t )= u s + du(x, t ), w here du(x, t ) is an infinitesim ally b g u . sm all perturbation, w e obtain after linearizing in du(x, t ): a c . i l 2 / 2 ~ ï d u ≈ ï ’ e 2 h = ld u - ∆ + k ÷ d u u ∆ 2 0 ÷ d ï t « ï x ◊ Class I: Linear stability analysis of the uniform state S ince any perturbation can be represented as a Fourier integral and B e n since the equation for du is linear, it is sufficient to study the G u stability of u =0 to the grow th of any Fourier m ode, r s i o n ikx ikx U ≈ = ’ n du(x, t) = A(t)e + c.c. | A |ç 0 e i v ∆ ÷ e ∆ ÷ r cos kx + i sin kx s « ◊ i t y In other w ords w e study the , 2p E grow th or the decay of sinusoidal k = h u l d perturbations w ith w avenum ber k M or w avelength l=2p/ k: e l r o n T he am plitude of the perturbation satisfies the equation - w w 2 w 2 2 . b A# = s (k ) A, s = l - (k 0 - k ) g u . a c . w here s is the grow th rate of the perturbation. T he solution to this i l / equation is ~ e s (k )t h = u A(t) A(0)e d Class I: Linear stability analysis of the uniform state B T he solution us=0 is stable if s(k)<0 for all w avenum bers k. Plotting e n the grow th rate s =s(k) for various l values w e find: G u r i o n s U l < 0 l = 0 l > 0 n i v e r s i t 0 0 0 y , E h u d M e r k0 k0 k o n s<0 for all k s(k0)=0 , us=0 is s(k)>0 for a band of - w us=0 is stable m arginally stable m odes,us=0 is unstable w w . b Ω T he solution u = loses stability at l = g s 0 0 u . a to the grow th of a m ode w ith w avenum ber k w hich c 0 . i l / breaks the translational sym m etry of the system . ~ e h u T his is a m ajor m echanism of sym m etry breaking in d nonequilibrium system s Class I: Linear stability analysis of the uniform state A nother w ay to describe the instability is by plotting a neutral B e 2 n 2 2 stability curve, by setting s = 0 l = (k - k ) G Ω 0 u r i o l n U A ll m odes w ithin the n i v parabola grow e r s i t us=0 is unstable, a y , band of m odes E Dk ~ l h u around k0 grow d M e 0 r us=0 is stable, all o k0 k n m odes decay - w w w Linear stability analysis provides inform ation about instability . b g thresholds, l = 0 in the case, and about the m odes that grow , a u . a c finite-w avenum ber m ode k in this case. . 0 i l / ~ It does not provide inform ation about the new state that the e h system evolves to because such inform ation can only be derived u d by considering large deviations from the original state us=0. Class I: Nonlinear analysis, amplitude equations G oing back to the S H m odel (now in tw o space dim ensions) B e n G u 3 2 2 2 r i = l - - ∞ + o u t u u ( k 0 ) u -Ñ w e consider approxim ate solutions of the form r s i t y , u(x,t) @ A(x, y,t) exp(ik x) + c.c. 0 < A << 1 0 E h u d w here the am plitude A is assum ed to be sm all but finite. W e M e further assum e that A varies w eakly in space and in tim e. S uch a r o n solution describes w eak m odulations of a stripe pattern, either - w w w . longitudinal or transverse or both. b g u . a c . i l / U sing various m athem atical m ethods, e.g. m ultiple tim e scales, a ~ e h nonlinear equation for the am plitude A can be derived from the u original S H equation: d Class I: Nonlinear analysis, amplitude equations T he equation, know n as the N ew ell-W hitehead-S egel (N W S ) B e n equation, reads: G u r i 2 o ≈ 2 ’ n ïA ï ï 2 U = lA + ∆ 2k0 - i ÷ A - 3 | A | A n ∆ 2 ÷ i v ït « ïx ïy ◊ e r s i t y , Equations of this kind are called “am plitude s E l = 0 h equations”. T hey have universal form s that are u d 0 determ ined by the types of instabilities the M e system s go through. A ny system that goes r o n through a finite-k instability of the form - k0 k w w w ill be described close to the instability point by a N W S equation. w . b g M ore generally, different system s that experience the sam e type u . a c of instability are described by the sam e am plitude equation and w ill . i l / behave sim ilarly near the instability point. ~ e h u T his explains the universal nature of m any of the patterns observed d in nature. Class I: Nonlinear analysis, amplitude equations T he am plitude equation has periodic solutions of the form : B e n 2 2 G u A = A exp(iqx), A = (l - 4k q )/ 3 r 0 0 0 i o n U n w hich exist w ithin the parabola i k = k ê q v 0 e 2 2 r l = s 4k 0 q i t y , l l E h u d M or e r o n - 0 q k k w 0 w w . b g Close to the instability point (0 2 2 c for the grow ing m odes obtained from the linear . l = (k - k ) i 0 l / stability analysis. T his does not im ply that all m odes that grow from ~ e h u the zero state evolve to periodic solutions – som e of them m ay be d unstable. Class I: Nonlinear analysis, amplitude equations T he am plitude equation can be used to study the linear stability of B e n periodic solutions w hich exist inside the blue parabola. S uch an G u analysis gives the follow ing boundaries of longitudinal and transverse r i o instabilities: n U n Periodic solutions are stable: i v e r s i t y , l Eckhaus instability: Periodic E h solutions lose stability to u d longitudinal m odulations M e r o n - w w w . 0 b g k k u 0 . a c . i l / Z igzag instability: Periodic solutions lose ~ e stability to transverse m odulations h u d Read m ore in Cross & H ohenberg, Rev. M od. Phys. 1993. Class I: S ummary B e n G Instabilities m ay break translational sym m etry and give rise to u r i spatial patterns even w hen the system and the forces it is subjected o n to are uniform . U n i v e T he spatial patterns induced by instabilities are universal; the sam e r s i t patterns appear in different system s that go through the sam e y , instability. E h u d T he dynam ics close to instability points are described by universal M e equations – the so called am plitude equations , e.g. N W S equation for r o n a system close to a stationary finite- w avenum ber instability. - w w A m plitude equations are useful for studying secondary instabilities, w . b such as Eckhaus and zigzag. g u . a c . i l / ~ e h u d Class II: M ulti-stable systems M ultistability of uniform states can result in asym ptotic patterns B e n even if all the uniform states are linearly stable. G u r i T he FIS (ferrocyanide-iodate-sulfite) reaction: A bistable system of o n high (w hite) and low (black) pH . D evelops patterns by a transverse U n i front instability (class II). v e r s i t y , E h u d M e r o n - w w w . Bistability and b g hysteresis in the u . a FIS reaction c . i l / ~ e h u d Lee & S w inney, PRE (1995) Class II: Bistable systems W e w ill study pattern form ation phenom ena in bistable system s using B e n the FitzH ugh-N agum o (FH N ) m odel: G u r 3 2 i o u t = u - u - v + ∞ u n U 2 n i v = e (u - a v - a ) + d ∞ v v t 1 0 e r s i T his is an exam ple of an A ctivator-Inhibitor system : u activates the t y , grow th of itself and v, w hile v inhibits the grow th of itself and u. E h u d T he FH N m odel captures all 3 classes of pattern form ing system s: M e r v v v o e << 1 n - (u+ ,v+ ) w w w . b ( ) g u0 ,v0 u (u ,v ) u u 0 0 u (u- ,v- ) . (u- ,v- ) a c . i l / ~ e Class I Class II Class III h u d 3 u - a1v - a0 = 0 u - u - v = 0 (nullclines) Class II: Bistable systems - fronts W e focus here on class II, corresponding to bistable system s that B e n consist of tw o stable uniform stationary states: u = u - , v = v- and G u u = u+ , v = v+ r i o v n S uch system s support in general U n i v front solutions u(x,t) that are bi- e (u+ ,v+ ) r s asym ptotic to the tw o stable states: i t y , u(x,t)çu+ v(x,t)çv+ as x ç-Ñ E ( ) h u0 ,v0 u u d u(x,t)çu- v(x,t)çv- as x ç+Ñ M (u- ,v- ) e r o n - y (u+ ,v+ ) + - (u- ,v- ) w w U p state D ow n state w . b g u x W e w ill see in the follow ing that . a c . pattern form ation phenom ena in i l u+ / bistable system s strongly ~ v+ e h depend on the properties of u v x d - these fronts solutions. u- Class II: Bistable systems - fronts Let’s begin w ith a sim pler system , assum ing v is constant, B e n 3 G u ut = u - u - v + u xx [Ω a gradient system ] r i o n 3 U u -u - v = 0 n i v v e For a certain range of v there are tw o stable r s i t stationary uniform solutions, u+(v) and u-(v), y , and an unstable solution u0(v) in betw een: E h u u d Front solutions that are biasym ptotic to the u- u0 u+ M tw o stable states propagate at constant e r o speeds, w hich depend on the value of v: n - w v<0 v=0 v>0 w w u u u+ . + + [ Fronts propagate so b g u as to m inim ize a . a u c u- u- - . Lyapunov functional ] i l / (+) invades (–) (–) invades (+) ~ e h u N ote: front speed is uniquely determ ined by the value of v. d Class II: Bistable systems – front instabilities T he situation becom es com pletely different w hen the equation for B e n the inhibitor v is added: G u r i u o + n 3 v u - a1v - a0 = 0 u -u - v = 0 v+ U n i v v e - r (u+,v+) s u i - t y x , (u-,v-) u E h u d M e Front solutions connecting the up-state (+) (at xç -Ñ) to the dow n- r o xç -Ñ n state (-) (at ) can them selves undergo instabilities. - w w T w o front instabilities: w . b g A transverse instability A longitudinal instability u . a c . u+ i l / v+ ~ + - e h u v- d u- Class II: Bistable systems – front instabilities B T o understand these instabilities and their effects on pattern e n form ation w e first need to understand how the curvature of a front G u r k , affects its norm al velocity, C (velocity in a direction norm al to the i o n front line). U n i v T he curvature affects the diffusion of the e r - s i activator and the inhibitor and thereby the t y , front velocity C: E + h R u d C = C(k ) k = 1 R M e Radius of curvature r o n - w w T he actual form of the velocity-curvature relation w ill be affected w . b by the param eters that control these diffusion processes: d and e. g u . a c . i U sing a singular perturbation analysis in the range e/d << 1 , w hich l / ~ is often m et in actual system s, velocity-curvature relationships can e h u be derived. d Class II: Bistable systems – front instabilities B e n G u r i o n U n i C v e r s i t y , E h k u d M e r o n - 0 w w w 0 . b g u . a c . i l / ~ e h u d H agberg & M eron, Chaos (1994), N onlinearity (1994) Class II: Bistable systems – stationary patterns B Front stability to transverse perturbations e n G u C C = C0 - Dk r i + - + - o Front is stable n U (d small - slow v n i Ω v 0 k diffusion) e r s i t y , 3 d large C C = C0 + D1k - D3k E + - + - h u Front is unstable d (d large - fast v Ω M e k diffusion) r 0 o n - w w U nder these conditions stationary labyrinthine patterns develop: w . b g u . N ote: dom ains do not m erge a c . i due to fast v diffusion l / ~ e h u d vf<0 vf=0 M im ura 2000 Class II: Bistable systems – traveling w aves Front stability to longitudinal perturbations B e n G u r i o S tationary labyrinthine n U patterns are found here n i d v C w here is large (fast v e r diffusion) and e not too s i t sm all. y , k E h Rem iniscent of the behavior u d near a cusp singularity or M pitchfork bifurcation e r o n - w w w . b g u . a c . i l / T he solid line is the threshold of a longitudinal front instability of ~ e h the pitchfork type com m only referred to as the “N on-equilibrium u d Ising-Bloch” (N IB) bifurcation. Class II: Bistable systems – traveling w aves T he N on-equilibrium Ising-Bloch (N IB) bifurcation: B e n u+ Ising front (Coullet) G u v r + i o v v n - U u- n i v e r s i t y h = ed , u E h u Bloch fronts d M e Ikeda, M im ura, N ishiura, 1989 r o H agberg and M eron 1994 n - In Ising regim e (h>hc) fronts have In Bloch regim e (h unique direction of propagation propagate in opposite directions w . b g C C C u . a c . i l / ~ hc h hc h hc h e h u d (+) invades (–) or (–) invades (+) (+) invades (–) and (–) invades (+) Class II: Bistable systems – traveling w aves B W hat are the im plications for pattern form ation? e n G 1. O nset of traveling w aves u r i o n T he N IB designates a transition from stationary or uniform U n i patterns to traveling w aves v e r s i Ising regim e: dom ains can t y , expand, shrink or becom e E h stationary u (+) invades (–) (–) invades (+) d M e r o Bloch regim e: dom ains can n - travel w (–) invades (+) (+) invades (–) w w . b g u . a 2. S piral w aves: c . i l / ~ e h u Point e d Gilli and Frisch (1995) Class II: Bistable systems – traveling w aves B Close to the N IB bifurcation – m ore com plex dynam ics e n H agberg & M eron, Chaos (1994), PRL (1994); Elphick, H agberg & M eron PRE (1995). G u r k=0 i 3. Bloch-front turbulence: Positive slope at o n im plies transverse U n instability i C v e r Point s d i t y , E h V ortex u d pair k M e r o n - w w w . G row ing transverse A fter spiral vortex b g perturbations nucleation u . a c . i l / ~ S im ulations of the FH N e h m odel close to the front u d bifurcation Class II: Bistable systems – traveling w aves B 4. S pot splitting: e n G u r i C Initial spot o n U n i v e r s i t y , 0 k E h u d M e Local front reversals induce r o Point b n nucleation of spiral-vortex pairs - w w w . S im ulations of the FH N m odel in the Ising b g regim e close to the front bifurcation u . a c . i U p state (+) D ow n state (-) l / ~ e h u H agberg and M eron, PRL (1994), N onlinearity (1994), d PRL (1997), Physica D (1998) Class II: S ummary B e Bistability of states allow s for dom ain patterns, stationary or n G traveling patterns, consisting of alternate dom ains of the different u r i stable states. o n U n i D om ain patterns are strongly affected by front instabilities: v e r s i t A transverse front instability that leads to stationary y , labyrinthine patterns. E h u d A longitudinal instability – the N IB bifurcation – that designates M e the onset of traveling w aves. r o n - Coupling of the tw o instabilities leads to spiral turbulence, spot w w splitting, etc. w . b g u . a T he m ultistable states need not be stationary and uniform - c . i l periodic states in space or tim e, can be reduced to stationary / ~ e uniform states using am plitude equations. h u A d A cknow ledgement A ric H agberg, Christian Elphick (theory) B e n H arry S w inney, Q i O uyang, Bill M cCorm ick (experim ents) G u r i o n Funds U n i v e BS F – U S -Israel Binational S cience Foundation r s i t y , E R elevant references h u d 1. Meron E., “Pattern Formation in Excitable Media”, Physics Reports 218, 1-66 M e r (1992). o n 2. M.C. Cross and P.C. Hohenberg, “Pattern Formation Outside Equilibrium”, Rev. - Mod. Phys. 65, 851 (1993). w w 3. Hagberg A. and Meron E., “Domain Walls in Nonequilibrium Systems and the w . b Emergence of Persistent Patterns”, Phys. Rev. E 48, 705-708 (1993). g u . 4. Hagberg A. and Meron E., “Pattern Formation in Non-Gradient Reaction Diffusion a c . i Systems: The Effects of Front Bifurcations”, Nonlinearity 7, 805-835 (1994). l / ~ 5. Hagberg A. and Meron E., “From Labyrinthine Patterns to Spiral Turbulence”, Phys. e h Rev. Lett. 72, 2494-2497 (1994). u d R elevant references B e 6. Hagberg A. and Meron E., “Complex Patterns in Reaction Diffusion Systems: n A Tale of Two Front Instabilities”, Chaos 4, 477-484 (1994). G u r 7. Elphick C., Hagberg A. and Meron E., “Dynamic Front Transitions and i o Vortex Nucleation”, Phys. Rev. E 51, 1899-1915 (1995). n U n 8. Lee K.J. and Swinney H.L., “Lamellar structures and self replicating spots in a i v e reaction-diffusion systems”, Phys. Rev. E 51, 3052-3058 (1995). r s i 9. Hagberg A., Meron E., Rubinstein I, and Zaltzman B., “Controlling Domain t y , Patterns Far From Equilibrium”, Phys. Rev. Lett. 76, 427-430 (1996). E 10. Haim D., Li G., Ouyang Q., McCormick W.D., Swinney H.L., Hagberg A., h u d and Meron E., “Breathing Spots in A Reaction Diffusion System”, Phys. Rev. M Lett. 77, 190-193 (1996). e r o 11. Hagberg A. and Meron E., “The Dynamics of Curved Fronts: Beyond n Geometry”, Phys. Rev. Lett. 78, 1166-1169 (1997). - w 12. Hagberg A., Meron E., Rubinstein I, and Zaltzman B., “Order Parameter w w Equations for Front Transitions: Planar and Circular Fronts”, Phys. Rev. E, 55, . b g 4450-4457 (1997). u . a 13. Hagberg A. and Meron E., “Order Parameter Equations for Front Transitions: c . i l Nonuniformly Curved Fronts”, Physica D 123, 460-473 (1998). / ~ 14. Hagberg A. and Meron E., “Kinematic Equations for Front Motion and Spiral- e h u Wave Nucleation”, Physica A 249, 118-124 (1998). d