La lecture à portée de main
Découvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDécouvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDescription
Sujets
Informations
Publié par | profil-zyak-2012 |
Nombre de lectures | 19 |
Langue | English |
Extrait
Stabletransportofinformationnearessentially
unstablelocalizedstructures
T
HIERRY
G
ALLAY
1
,G
UIDO
S
CHNEIDER
2
,H
ANNES
U
ECKER
2
1
InstitutFourier,Universite´deGrenobleI,F-38402Saint-Martind'Heres,France
2
MathematischesInstitutI,Universita¨tKarlsruhe,D-76128Karlsruhe,Germany
June23,2003
Abstract
Whenthesteadystatesatinnitybecomeunstablethroughapatternformingbifurca-
tion,atravellingwavemaybifurcateintoamodulatedfrontwhichistime-periodicina
movingframe.ThisscenariohasbeenstudiedbyB.SandstedeandA.Scheelforaclass
ofreaction-diffusionsystemsontherealline.Undergeneralassumptions,theyshowed
thatthemodulatedfrontsexistandarespectrallystablenearthebifurcationpoint.Here
weconsideramodelproblemforwhichwecanprovethenonlinearstabilityofthese
solutionswithrespecttosmalllocalizedperturbations.Thisresultdoesnotfollowfrom
thespectralstability,becausethelinearizedoperatoraroundthemodulatedfronthas
essentialspectrumuptotheimaginaryaxis.Theanalysisisillustratedbynumerical
simulations.
Contents
1Introduction2
2MainResults4
3ExistenceandstabilityofTuringpatterns11
3.1Existence...................................11
3.2Blochwaves..................................13
3.3Linearstability................................14
3.4Nonlinearstability..............................18
4Constructionofmodulatedfronts18
4.1Theidea....................................19
4.2Transversalityinextendedspace.......................21
1
4.3BifurcationofTuringpatterns........................24
4.4Bifurcationofthemodulatedfront......................24
4.5RemarksoncaseIIandcaseIII........................25
5Thenonlinearstabilityproof25
5.1Theidea....................................26
5.2Theunscaledequations............................28
5.3Therenormalizationprocedure........................30
5.4IterationandConclusion...........................38
6ResultsincaseII
04
7Numericalsimulations42
7.1Modulatedfronts...............................42
7.2Modulatedpulses...............................43
1Introduction
Localizedstructuressuchaspulsesandfrontsplayanimportantroleinthemathematical
theoryofinformationtransport.Typicalsituationswheresuchnonlinearphenomenaarise
arethepropagationofelectromagneticwavesinwiresorbers[AA97,NM92],andthe
motionofelectricpulsesalongnerveaxons[Hux52].Animportantissue,bothfromatheo-
reticalandapracticalpointofview,istherobustnessofthesesolutionswithrespecttosmall
inhomogeneitiesofthepropagationmedium.
Inaremarkablepaper[SS99],B.SandstedeandA.Scheelstudiedanewbifurcationscena-
riofortravelingpulsesinreaction-diffusionsystemsontherealline.Theyinvestigatedthe
situationwherethehomogeneoussteadystateatinnitybecomesunstableandbifurcatesto
aspatiallyperiodicTuringpattern.Theoriginallystablepulsethusundergoesanessential
instability,inthesensethattheessentialspectrumofthelinearizedoperatorcrossesthe
imaginaryaxisatthebifurcationpoint.Undergeneralassumptions,theauthorsshowedthat
theoriginalpulsebifurcatestoamodulatedpulsewhichsitime-periodicinauniformly
translatingframe.Theyalsoprovedthatthisbifurcatingsolutionisspectrallystable[SS00].
However,sincethespectrumofthelinearizationextendsallthewaytotheimaginaryaxis
(withoutgap),thislastresultdoesnotimmediatelyimplythenonlinearstabilityofthemo-
dulatedpulse.Theanalysisof[SS99,SS00]canbegeneralizedtofrontsolutionsconnecting
twodifferentstableequilibria[SS01a,SS01b].Inthiscase,modulatedfrontsmaybifurcate
fromanexistingtravelingwaveifoneorbothofthereststatesatinnitybecomeunstable.
Inthispaper,wegobeyondthelinearstabilityanalysisof[SS00,SS01b]andweshow,
atleastonaspecicexample,thatmodulatedfrontsare
nonlinearly
stablewithrespectto
spatiallylocalizedperturbations.Insimpleterms,thisresultimpliesthatinformationcanbe
transportedinastablemannerevenifthepropagationmediumbecomesunstablethrougha
2
Turingbifurcation.Tokeeptheanalysisassimpleaspossible,wedonotconsiderabstract
reaction-diffusionsystemsasin[SS99],butweprefertoconcentrateonamodelproblem
thatexhibitsallfeaturesofthegeneralcase.Althoughthishasnotbeenprovedsofar,we
certainlyexpectthatallresultsbelowholdtrueforgeneralreaction-diffusionsystemsunder
thesameassumptionsasin[SS00](forpulses)or[SS01b](forfronts).
OurmodelproblemisaChaffee-Infanteequationfortherstvariable
u
coupledtoaSwift-
Hohenbergequationforthesecondvariable
v
,namely:
∂
t
u
=
∂
x
2
u
+
1
(
u
−
c
0
)(1
−
u
2
)+
v,
2)1(∂
t
v
=
−
(1+
∂
x
2
)
2
v
+
αv
−
v
3
−
γvF
(
u
)
,
where
u
(
x,t
)
,v
(
x,t
)
∈
R
,
x
∈
R
,and
t
≥
0
.Thissystemisespeciallyconvenientto
analyze,becauseitcouplestwoscalarequationswhichareratherwellunderstood.Inwhat
follows,thespeedparameter
c
0
andthecouplingparameter
γ
willbexed,with
0
<c
0
<
1
and
γ>
0
nottoobig(seeTheorem2.3below).Ourbifurcationparameter
α
willthenvary
inaneighborhoodofthebifurcationpoint
α
=0
.Tocoverallinterestingcases,weshall
considerthreedifferentfunctions
F
,namely
I)
F
(
u
)=1
−
u
2
,
II)
F
(
u
)=1
−
u,
III)
F
(
u
)=1+
u.
Forallchoicesof
F
,system(1)possessestwospatiallyhomogeneousequilibria
(
u,v
)=
(
±
1
,
0)
andaone-parameterfamilyoffrontsolutions
(
u,v
)=(tanh((
x
−
c
0
t
−
x
0
)
/
2)
,
0)
,x
0
∈
R
,
(2)
connectingtheseequilibria.For
α<
0
,theequilibriaandthefamilyoffrontsolutions
areasymptoticallystablewithsomeexponentialrate.When
α
crossestheoriginfromleft
toright,someoftheequilibriabecomeunstable,dependingontheparticularchoiceof
F
.
IncaseI,thesteadystatesaheadofandbehindthefrontundergoaTuringbifurcationand
spatiallyperiodicequilibriaarecreated.IncaseII,thishappensonlyforthesteadystate
(
u,v
)=(1
,
0)
aheadofthefront,andincaseIIIonlyforthesteadystate
(
u,v
)=(
−
1
,
0)
behindthefront.Inthisrespect,caseIisclosetothecaseofapulse.
1
0
1
0
1-1--50050-50050
Figure1:Modulatedfrontsfor(1)incasesI(left)andII(right).Thesnapshotsshowthe
u
-componentobtainedfromgenericinitialdataatsomelargetime;seealsosection7.
Atthebifurcationpoint
α
=0
,thefrontsolutions(2)becomeessentiallyunstableand,in
casesIandII,afamilyofmodulatedfrontsiscreated.Thesesolutionsaretime-periodicin
3
amovingframewithspeed
c
≈
c
0
,andtheyconnectaspatiallyperiodicTuringpatternat
x
=+
∞
toanotherTuringpattern(caseI)ortotheuniformsteadystate
(
−
1
,
0)
(caseII)at
x
=
−∞
,seeFig.1.WeshallnotconsidercaseIIIanylonger,sincetheanalysisin[SS01a]
showsthatatleastgenericallynomodulatedfrontsexistinthatcase;typicallythepatternis
outrunbythefront,seeg.6onpage44foranillustration.Asforthestability,itturnsout
thatincaseIIthefamilyofmodulatedfrontsisasymptoticallystablewithexponentialrate.
Thiscanbeprovedrathereasilyusingweightedspaces,seesection6.Thusthechallenging
caseinprovingstabilityiscaseI.Inthissituation,thelinearizationaroundthemodulating
frontshascontinuousspectrumuptotheimaginaryaxis.Itisthepurposeofthispaperto
explainhowneverthelessthenonlinearstabilityofthesesolutionscanbeshown.
Remark1.1
IncaseI,themodelproblem(1)seemsnon-generic,sincebothhomogeneous
equilibria
(
±
1
,
0)
undergoaTuringbifurcationatthesamevalueoftheparameter
α
.Infact,
weimplicitlyrestrictouranalysistosystemsforwhichthedestabilizationofbothequilibria
hasthesameorigin
(inourexample,thisisthecouplingofthebistable
u
-equationtoa
single
Swift-Hohenbergequation).Thisalsoexplainswhythewavelengthsofthebifurcating
patternsaheadofandbehindthefrontcoincide.Aswasobservedbyoneofthereferees
ofthispaper,itwouldthenbemorenaturaltoconsiderthecaseofapulseinsteadofa
front.Butthenwewouldhavetoreplacethescalar
u
-equationin(1)bya
2
×
2
system,
whichmakestheanalysisevenmoreintricate.Also,wefounditinterestingtoencompassall
possiblecases(I,II,andIII)inasingle,relativelysimplemodel.
Acknowledgments:
ThisworkwassupportedbytheFrench-Germancooperationproject
PROCOPE00307TKentitledAttractorsforextendedsystems.TheauthorsalsothankB.
SandstedeandA.Scheelforstimulatingdiscussions,andbothrefereesforusefulcomments
andsuggestions.
∂2uu1(uc)(1u2)+v2MainResults
Inthissection,wegiveourresultsinthemostinterestingcase,i.e.when
F
(
u
)=1
−
u
2
in(1).
Tosimplifythenotation,werewrite(1)intheform
∂
t
U
=
L
(
∂
x
)
U
+
N