ContentslistsavailableatScienceDirect
Applied Mathematics and Computation
journalhomepage:www.elsevier.com/locate/amc
A modification to the first integral method and its applications
Hong-Zhun Liu
Zhijiang College, Zhejiang University of Technology, Shaoxing 312030, PR China
a rt i c l e i nf o
Article history:
Received 30 January 2019 Revised 3 December 2020 Accepted 3 December 2021 Available online 18 December 2021 Keywords:
First integral method
Conformable fractional diffusion-reaction equation
Duffing-van der Pol oscillator Complex cubic-quintic Ginzburg–Landau equation
Equation for surface waves KdV–Burgers–Fisher equation
a b s t ra c t
In thisarticle,we modifyFeng’s firstintegral method(FIM)for the purposeof enlarg- ingitsapplications.ComparedwithoriginalFIM,ourmodifiedFIMismorestraightand canalsobeemployedtofindfirstintegralofhigher-orderordinarydifferentialequations (ODEs).WeemployourmodifiedFIMintofivedifferentialequations,namely,thedensity- dependentconformablefractionaldiffusion-reactionequation,theDuffing-vanderPolos- cillator,the complexcubic-quintic Ginzburg–Landauequation,thewell-knownnonlinear evolutionequationfordescriptionofsurfacewavesinaconvectingliquid,andtheKdV–
Burgers–Fisherequation.Consequently, wegetthesamefirstintegralobtainedbyFeng’s FIM forthe firstequation; for the secondequation,we reobtaincertainimportantfirst integral reportedpreviously;forthe thirdequation, weconstructanewfirstintegralof complexcubic-quinticGinzburg–Landauequation;andforthefourthandfifthequations, weshowtheeffectivenessofourapproachtothird-orderODEsandreobtainthesamefirst integralrecentlypresentedbyKudryashovforthefourthequation,andforthefifthequa- tion, twonewfirstintegralsarepresented.Alltheabovefullyrevealtheeffectivenessof ourmodification.
© 2021ElsevierInc.Allrightsreserved.
1. Introduction
In2002,forconstructingtravelingwavesolutionsofBurgers–KdVequation,Feng[1,2]proposedapopularmethodcalled firstintegralmethod(FIM)whichmakesuseofthealgebraictheoryofcommutativeringandhasattractedmuchattention fromresearcherstoinvestigatefirstintegralandthereaftertravelingwavesolutionsofpartialdifferentialequations(PDE)in nonlinearscience,seereferences[3–31]andthereferencestherein.Forconvenience,we firstintroduceFeng’sfirstintegral methodinbrief.
1.1. OutlineofFIM
SupposethatweinvestigateanonlinearPDEwhichreads
E
(
u,ux,ut,uxx,...)
=0, (1)whereu=u(x,t)isadependentvariabletobedetermined,E isapolynomial ofuandits partialderivativesinwhichthe highestorderderivativesandnonlineartermsareinvolved.
E-mail addresses: [email protected] , [email protected] https://doi.org/10.1016/j.amc.2021.126855
0 096-30 03/© 2021 Elsevier Inc. All rights reserved.
• Step1
Carryingoutthefollowingtravelingwavetransformation
u
(
x,t)
=u( ξ )
,ξ
=kx+ct, (2)wegetanonlinearordinarydifferentialequation(ODE)
Q
(
u,u,u,· · ·)
=0, (3)wheretheprimemeansthederivativewithrespecttoindependentvariable
ξ
.• Step2
Wesupposethat(3)admitsthefollowingparticularsolution
u
( ξ )
=X( ξ )
, (4)then(3)canbechangedtoanonlinearplaneautonomousequationsas
X
( ξ )
=Y( ξ )
,Y
( ξ )
=H(
X( ξ )
,Y( ξ ))
, (5)• Step3
AccordingtothequalitativetheoryofODEs[32],wecandirectlyconstructthegeneralsolutionsto(3)bymakinguseof its twoindependentfirstintegrals.Nevertheless,thereisneitheralogicalwaynorsystematictheoriestoconstructitsfirst integralsforagivenplaneautonomoussystem,soitisnoteasyforustogetevenonefirstintegral.Inreference[1],Feng employedtheDivisionTheoremtosearchforonefirstintegralfor(3)whichreduces(3)toafirst-orderODE.Subsequently, exacttravelingwavesolutionsto(1)wereconstructedbysolvingthefirst-orderODEdirectly.Forthesakeofcompleteness, wepresentthefamousDivisionTheoreminwhatfollows.
DivisionTheorem. SupposethatM(
ω
,z)isanirreduciblepolynomialinC[ω
,z],andN(ω
,z)isalsoapolynomialinC[ω
,z]; IfthepolynomialN(ω
,z)vanishesatallzeropointofM(ω
,z),thenthereexistsapolynomialG(ω
,z)inC[ω
,z]suchthatN
( ω
,z)
=G( ω
,z)
·M( ω
,z)
.Indetail,wesupposethat p(X(
ξ
),Y(ξ
))=0isthefirstintegralfor(5)andp(X(ξ
),Y(ξ
))isalsoanirreduciblepolyno- mialinC[X,Y]suchthatp
(
X,Y)
= mi=0
ai
(
X)
Yi. (6)Accordingtothedivisiontheorem,theremustexistW(X,Y)=h(X)Y+g(X)inC[X,Y],suchthat
dp
(
X,Y)
d
ξ
=(
h(
X)
Y+g(
X))
·m
i=0
ai
(
X)
Yi, (7)from which we determine each ai(X),i=1...m. Subsequently, we obtain the desired first integral p(X(
ξ
),Y(ξ
))=0 for plane autonomous system (5). As a result, exact traveling wave solutions to (1) can be constructed by solving p(X(ξ
),Y(ξ
))=0directly.1.2. AnobservationofFIM
Wenoticethefollowingfact,namely
Observation1. Becauseofthedivisiontheorem,thesupposedirreduciblefunction p(X,Y)isabinarypolynomial.Thus,the finalemployedODE(3)isexactlyasecond-orderODE,namelyQ(u,u,u)=0.Asaresult,wecannotapplyFIMtoamore higher-orderODE,sayathird-orderODE.
So, howtomodify FIMto letit beapplied tothe higher-orderODEisstill an open problem.In thisarticle,wetry to answerthisquestion.
Rest ofthe articleisarranged asfollow:inSection2,we givethedescriptionofourmodificationto FIM.In Section3, the firstintegrals ofdensity-dependentconformable fractional diffusion-reaction(cfDR) equation, the Duffing-vander Pol (DVDP) oscillator,the(1+1)dimensionalcomplexcubic-quinticGinzburg–Landau(cqGL)equation,thenonlinearequation forsurfacewavesarisingina convectingfluid,andtheKdV–Burgers–Fisherequation areillustrated asapplications ofour modifiedFIM,respectively.Section4concludesourarticle.
2. AmodificationtoFIM
ThefirststepisthesameasoriginalFIM.Werewrite(3)intheform
Q
(
u,u,u,...,u(n))
=0, (8)wherethehighestorderofderivativesncanbegreaterthan2.
Withoutlossofgenerality,inwhatfollows,weonlyconsiderthecaseofn=3,namely
Q
(
u,u,u,u)
=0, (9)• Step2
Wesupposethat(9)admitsthefollowingparticularsolution
u
( ξ )
=X( ξ )
, (10)then(9)canbechangedtoasystemofnonlinearODEsas X
( ξ )
=Y( ξ )
,Y
( ξ )
=Z( ξ )
,Z
( ξ )
=H(
X( ξ )
,Y( ξ )
,Z( ξ ))
, (11)whichisnotaplaneautonomousinthiscase.
• Step3
Wesupposethat(9)admitsafirstintegralintheform p=0, where p
( ξ
,X,Y,Z)
de f=m
i=0
ai
( ξ
,X,Y)
Zi, (12)whereai(
ξ
,X,Y)(i=1,...,m−1)arearbitraryfunctionstobedetermined.Thenbymakinguseof(11),wehave dp
( ξ
,X,Y,Z)
d
ξ
=R( ξ
,X,Y,Z)
, (13)whichmustbeequalto0sincep(
ξ
,X,Y,Z)=0isthefirstintegralfor(11). Inordertospecifyai(ξ
,X,Y)(i=1,...,m),wesupposeR
( ξ
,X,Y,Z)
=f(
p( ξ
,X,Y,Z))
, (14)where f(·)isafunctionadmittingthefollowingconditionf(0)=0,andhereby f(p(
ξ
,X,Y,Z))isexactlyequalto0.In order to eliminate termsof Z,we can take f(p(
ξ
,X,Y,Z))such that both ofR(ξ
,X,Y,Z) and f(p(ξ
,X,Y,Z))have termswhichinvolveZincommon.Thusfrom(14),wecangetanODE
G
( ξ
,X,Y)
=0. (15)From(15),wedetermineai(
ξ
,X,Y)(i=1,...,m).Finally,wegettherequiredfirstintegral p(
ξ
,X,Y,Z)=0withdeterminedai(ξ
,X,Y)(i=1,...,m). 3. Applications3.1. Thedensity-dependentcfDRequation
Thedensity-dependentcfDRequationreads
∂
αu(
t,x)
∂
tα +ku(
t,x) ∂
u(
t,x)
∂
x =D∂
2u(
t,x)
∂
x2 +au(
t,x)
−bu2(
t,x)
, (16)where0<
α
≤1,t>0anda,b,kandDarepositiveparameters[24].Weshouldmentionthat(16)canbeservedasamodel arisinginthenonlinearscienceandhasattractedmuchgreatefforts[24,33,34].Assumingthat fisacontinuousfunction f:(0,∞)−>,thentheconformablefractionalderivativeoforder
α
isintro-ducedas[35]
Dαf
(
t)
=lim →0f
(
t+t1−α
)
−f(
t)
, (17)
whichbearsfollowingproperties[24,35]
(1) Dα(a f+bg)=aDα(f)+bDα(g),
∀
a,b∈. (2) Dα(tμ)=μ
tμ−α,∀ μ
∈.(3) Dα(f g)= f Dα(g)+gDα(f). (4) Dα(gf)= gDαfg−2fDαg.
Furthermore,ifgisaarbitrarydifferentialfunctiondefinedintherangeof f,and fisdifferentiableand
α
-differentiable, thenwehaveDαf
(
t)
=t1−αdfdt, (18)
Dα
(
f◦g)(
t)
=t1−αg(
t)
f(
g(
t))
. (19)Nowlet
u
(
x,t)
=U( ξ )
,ξ
=mx− cα
tα, (20)wecanchange(16)tothefollowingODE[24]
Dm2U+cU−kmUU+aU−bU2=0. (21)
Wesupposethat(21)admitsaparticularsolutionintheform
U
( ξ )
=X( ξ )
, (22)then(21)canbechangedtoaplaneautonomoussystemofnonlinearODEsas X
( ξ )
=Y( ξ )
,Y
( ξ )
= 1 m2D −cY+kmXY−aX+bX2, (23)
In2018,Rezazadeetal.[24]investigated(23)byusingFIMandobtainedthefollowingfirstintegral Y=− ak
2mbDX+ k
2mDX2, (24)
alongwiththefollowingcondition c=
(
4b2D+ak2)
m2kb . (25)
Inwhatfollows,weemployourmodificationmethodto(21).Forthesake ofsimplicity,wesupposethat(23)admitsa firstintegralintheform
p=0, where pde f=A
(
X)
+Y, (26)whereA(X)isafunctiontobedetermined.
Thenwehave dp
d
ξ
=AY+Y=AY+1 m2D
−cY+kmXY−aX+bX2=
A− c m2D+ k
mDX
Y− a
m2DX+ b
m2DX2. (27) Accordingtoourmodification,wetake
A− c m2D+ k
mDX
Y− a
m2DX+ b
m2DX2= f
(
p)
=A− c m2D+ k
mDX
p=
A− c m2D+ k
mDX
(
A(
X)
+Y)
, (28) whichcanbereducedtoAA+ k
mDXA− c
m2DA+ a
m2DX− b
m2DX2=0. (29)
Byobservation,wetake
A=k0+k1X+k2X2, (30)
wherekisarearbitraryconstantstobespecified.
Substituting theansatz (30)into (29)andequating thecoefficientsofpowers X withzero,we canobtain anonlinear over-determined systemof algebraicequations. Then by solving the over-determinedsystem, we get the followingthree cases
(a) k0=−mka ,k1= mkb ,k2=0,underconditionofc=mbDk .
(b) k0=k2=0,k1=mkb ,underconditionofc= (b2D+bkak2)m.
(c)k0=0,k1=2mbDak ,k2=−2mDk ,underconditionofc= (4b2D2+bkak2)m. Inaword,wehaveconstructedthreefirstintegralsto(21),namely
U+ b mkU− a
km=0, (31)
U+ b
mkU=0, (32)
U+ ak
2mbDU− k
2mdU2=0. (33)
Remark 1. The third first integral (33)is the same as (24)obtained by Rezazade etal, butthe formertwo along with correspondingconstraintconditionshavenotbeenpresentedbyresearcherstothebestofourknowledge.
3.2. TheDVDPoscillator
TheDVDPoscillatorbeingconsideredreads
x+
( α
+β
x2)
x−γ
x+x3=0, (34)wherethe primedenotesderivative withrespecttotimet and
α
,β
andγ
arerealparameters.In2010, bymeans ofthe Prelle–Singermethod[36],Fengetal.[37]obtainedthefollowingfirstintegralx+
α
−β
3x+
β
3x3=Iexp
−3t
β
, (35)whereIisaconstant,alongwiththefollowingcondition
α
= 3β
−βγ
3 . (36)
Inwhatfollows,weemployourmodificationmethodto(34). Atthebeginning,wesupposethat(34)hasasolutionintheform
x
(
t)
=X(
t)
. (37)Then(34)ischangedtoanonlinearplaneautonomoussystemas X=Y,
Y=−
( α
+β
X2)
Y+γ
X−X3, (38)Forthesakeofsimplicity,wesupposethat(38)admitsafirstintegralintheform
p=0, where pde f=G
(
t)
+A(
X)
+Y. (39)Sowehave dp
dt =G+AY+Y=G+AY−
( α
+β
X2)
Y+γ
X−X3=G+γ
X−X3+(
A−α
−β
X2)
Y. (40) Accordingtoourmodification,wetakeG+
γ
X−X3+(
A−α
−β
X2)
Y=(
A−α
−β
X2)
p=(
A−α
−β
X2)(
G(
t)
+A(
X)
+Y)
, (41) whichcanbereducedtoG−
(
A−α
−β
X2)
G+γ
X−X3+( α
+β
X2)
A−AA=0. (42) Tosolve(42),wetakeγ
X−X3+( α
+β
X2)
A−AA=0, (43)G−
(
A−α
−β
X2)
G=0. (44)Tosolve(43),forthesakeofsimplicity,wetake
A=aX+bX3. (45)
Substituting(45)into(43),wehave
γ
X−X3+( α
+β
X2)
A−AA=(
bβ
−3b2)
X5+(
aβ
+bα
−1−4ab)
X3+( γ
+aα
−a2)
X=0. (46) Equating the coefficientsof powers X with zero,we obtain thefollowing nonlinear over-determined systemof algebraic equationsb
β
−3b2=0,a
β
+bα
−1−4ab=0,γ
+aα
−a2=0. (47)Bysolving(47),weobtainthefollowingnontrivialcase a=−
γ β
3 ,b=
β
3, (48)
alongwiththefollowingcondition
α
=9−γ β
23
β
, (49)whichisthesameas(36).
Inthiscase,(44)issimplifiedas G+
α
+βγ
3
G=0. (50)
Solving(50),weobtain G=Cexp
−
α
+βγ
3
t
, (51)
whereCisanintegrationconstant.
Inaword,wegetafirstintegralto(34)intheform x−
γ β
3 x+
β
3x3+Cexp
−
α
+βγ
3
t
=0. (52)
Remark2. Thefirstintegral(52)isexactlythesameas(35).ComparedwiththePrelle–Singermethod,ourmethodismore straightforwardandeffective.
3.3. The(1+1)dimensionalcomplexcqGLequation The(1+1)dimensionalcomplexcqGLequationreads
iux+D
2utt+
|
u|
2u=iδ
u+i|
u|
2u+iβ
utt+iα|
u|
4u−γ |
u|
4u, (53) where,inoptics,u(x,t)iscomplexamplitude,tistheretardedtime,xisthepropagationdistance,theparameterδ
islinearlossorgain,
β
describesdiffusioncoefficient,D=±1isthedispersioncoefficientandparametersα
,γ
anddetermineits
shapeofnonlinear[23].
Introducingthefollowingtravelingwavetransformation
u=
v ( ψ )
exp(
iη )
,η
=px+st,ψ
=kx+ct, (54)wherek,c,pandsarearbitraryconstantstobespecified,alongwithconstraintconditions[23]
r=−4
β
sDc =k+Dsc
β
c2 ,l=2p+Ds2
Dc2 =−
β
s2−δ
β
c2 , (55)m=− 2
Dc2 =−
β
c2,n=−2
γ
Dc2 =−
α
β
c2, (56)wecanreduce(53)intothefollowingODE
v
−rv
−lv
−mv
3−nv
5=0. (57)Inreference[23],G.Akrametal.constructed(57)’sexactsolutions byFeng’sfirstintegralmethod[1].Inwhatfollows,we employourmethodtoobtainitsfirstintegral.
Atthebeginning,wesupposethat(57)hasasolutionintheform
v ( ψ )
=X( ψ )
. (58)Then(57)canbechangedtoaplaneautonomoussystemofnonlinearODEsintheformof X=Y,
Y=
γ
Y+lX+mX3+nX5, (59)Wesupposethat(59)admitsafirstintegralintheform
p=0, where pde f=G
( ψ )
+A(
X)
+Y. (60)Sowehave dp
d
ψ
=G+AY+Y=G+AY+γ
Y+lX+mX3+nX5=G+lX+mX3+nX5+(
A+γ )
Y. (61) Accordingtoourmodification,wetakeG+lX+mX3+nX5+
(
A+γ )
Y=(
A+γ )
p=(
A+γ )(
G(
t)
+A(
X)
+Y)
, (62) whichcanbesimplifiedasG−
(
A+γ )
G+lX+mX3+nX5−γ
A−AA=0. (63)Tosolve(63),wetake
lX+mX3+nX5−
γ
A−AA=0, (64)G−
(
A+γ )
G=0. (65)From(64),AisacubicpolynomialofX,soweassume
A=aX+bX3, (66)
whereaandbarearbitraryconstantstobespecified.
Nowsubstituting(66)into(64),weobtain
(
n−3b2)
X5+(
m−rb−4ab)
X3+(
l−a2−ra)
X=0. (67)Thus,equatingthecoefficientsofpowersXwithzero,weobtain n−3b2=0,
m−rb−4ab=0,
l−a2−ra=0. (68)
Solving(68),wehave a=−r
4±m 4
3
n,b=± n
3, (69)
alongwiththefollowingconstraintcondition l=3m2
16n ±mr 8
3 n−3r2
16. (70)
ThusweobtainthevalueofAas A=
−r 4±m
4 3 n
X± n
3X3. (71)
Substituting(71)into(65),andsolvingG(
ψ
),weget G( ψ )
=Cexp(
34rψ )
exp±√ 3n
(
X2+ m 4n)
dψ
. (72)
Inaword,undertheconstraintcondition(70),wefindthefirstintegralof(57)intheform
v
+−r 4±m
4 3 n
v
± n3
v
3+Cexp3r4
ψ
±√3n( v
2+ m 4n)
dψ
=0. (73)
IfwetakeC=0,(73)isreducedtothefollowingBernoulliequation
v
−r 4∓m
4 3 n
v
=∓ n3
v
3, (74)whosegeneralsolutioncanbeobtainedintheform
v
2= 1C1exp
(
−12(
r∓m3n
) ψ )
±r∓4m√√n33 n, (75)
whereC1isanarbitraryintegrationconstant.
Remark3. Tothebestofourknowledge,firstintegral(73)isnew.
Remark4. Weshouldpointoutthat(57)admitsgeneralsolution(75),whichisalsonovel.
3.4. Theequationforsurfacewavesinaconvectingfluid
In this subsection,we will considera higher-orderdifferential equation, namely,the equation forsurface wavesin a convectingfluidwhichreads[38]
ut+
μ
uxx+σ
uxxx+δ
uxxxx+α (
u2)
x+χ (
u2)
xx=0. (76) Inaconvectingfluid,theaboveequationdepictsalongshallowwaveevolutionasandwhentheRayleighnumberfraction- allysurpassitscriticalvalueandhasbeeninvestigatedmanytimes,seereference[38]andthereferencestherein.Atχ
=0, (76)isreducedtotheknownKuramoto–Sivashinskyequation.Takingintoaccounttravelingwave solutionu=μχy(z),z=
μδx−C0δμ2t alongwithtworedefinedparameters
σ
andα
,wehavethenonlinearODEintheform[38]
yzzz+9
σ
yzz+yz+2yyz+α
y2−C0y+C1=0. (77) In orderto constructthe auxiliaryequation, Kudryashov [38] usedthe Fuchsindex which exists inthe Painlevé test and foundthat(77)admitsafirstintegralintheformyzz+
(
9σ
−α )
yz+y2+(
1+α
2−9ασ )
y+Cα
1 +4C2exp(
−α
z)
=0. (78)Inwhatfollows,weuseourapproachtoreobtainthisfirstintegral.
Wesupposethat(77)hasasolutionintheform
y
(
z)
=X(
z)
. (79)Then(77)canbechangedtoasystemofnonlinearODEsas X=Y,
Y=Z,
Z=−9
σ
Z−Y+2XY−α
X2+C0X−C1, (80)Forthesakeofsimplicity,wesupposethat(80)admitsafirstintegralintheform
p=0, where pde f=G
(
z)
+A(
X,Y)
+Z. (81)Sowehave dp
dz =G+A1Y+A2Z+Z=G+A1Y+A2Z−9
σ
Z−Y+2XY−α
X2+C0X−C1=G+A1Y−Y+2XY−
α
X2+C0X−C1+(
A2−9σ )
Z. (82) Accordingtoourmodification,wetakeG+A1Y−Y+2XY−
α
X2+C0X−C1+(
A2−9σ )
Z=(
A2−9σ )
p=(
A2−9σ )(
G(
z)
+A(
X,Y)
+Z)
, (83) whichcanbesimplifiedasG−
(
A2−9σ )
G+A1Y−Y−2XY−α
X2+C0X−C1−AA2+9σ
A=0. (84)Tosolve(84),wetake
A1Y−Y−2XY−
α
X2+C0X−C1−AA2+9σ
A=0, (85)G−
(
A2−9σ )
G=0. (86)ByobservingthedegreeofY inthelefthandsideof(85),weassume
A
(
X,Y)
=M(
X)
+kY, (87)whereM(X)isafunctionofX tobedetermined,andkisaconstant.
Substituting(87)into(85),wehave
[M−1−2X+9
σ
k−k2]Y−α
X2+C0X−C1−kM+9σ
M=0 (88)Therefore,wetake
M−1−2X+9
σ
k−k2=0 (89)−
α
X2+C0X−C1−kM+9σ
M=0. (90)Solving(90),wehave M=
α
9
σ
−kX2−C0
9
σ
−kX+C1
9
σ
−k, (91)and
M= 2
α
9
σ
−kX−C0
9
σ
−k. (92)Substituting(92)into(89),wehave
2α
9
σ
−k−2X+9
σ
k−k2−9σ
C0−k−1=0. (93)Sowehave 2
α
9
σ
−k−2=0, (94)σ
k−k2−9σ
C0−k−1=0. (95)Solving(94)and(95)comesto
k=9
σ
−α
, C0=α (
9σα
−α
2−1)
. (96)SubstitutingvaluesofkandC0 into(91),wehave M=X2+
(
1+α
2−9σα )
X+C1α
, (97)andthusweobtain
A=X2+
(
1+α
2−9σα )
X+C1α
+(
9σ
−α )
Y. (98)Andinthiscase,(86)turnstobethefollowingODE
G+
α
G=0. (99)Solving(99)gives
G
(
z)
=C2exp(
−α
z)
. (100)Asaresult,weobtainthefollowingfirstintegral Z+
(
9σ
−α )
Y+X2+(
1+α
2−9ασ )
X+C1α
+C2exp(
−α
z)
=0, (101)i.e.,
yzz+
(
9σ
−α )
yz+y2+(
1+α
2−9ασ )
y+C1α
+C2exp(
−α
z)
=0. (102)Remark5. Thefirstintegral(102)isinaccordwith(78)whichhasbeennewlypresentedbyKudryashov[38]. 3.5. TheKdV–Burgers–Fisherequation
Recently, Kocak proposed a third-order dispersion-dissipation-reaction model called the KdV–Burgers–Fisher equa- tion[39],namely
ut+
ε
uux−ν
uxx+μ
uxxx=ru(
1−u)
, (103)where
ε
,ν
,μ
,andrarerealparametersforconvection,diffusion,dispersion,andreactionterms,respectively.TheKdV–Burgers–FisherequationcanbereducedtotheKdVequationwhen
ν
=r=0,Burgersequationwhenμ
=r=0, Fisherequationwhenε
=μ
=0,Burgers–Fisherequationwhenμ
=0,anddispersive-Fisherequationwhenε
=ν
=0.Applyingthefollowingtravelingwavetransformation
u=u
( ξ )
,ξ
=kx+ω
t+ξ
0 (104)to(103),wehave
ω
u+ε
kuu−ν
k2u+μ
k3u=ru(
1−u)
. (105)Kocakusedthe tanhmethodtoinvestigateexactsolutionsof(105).Inwhatfollows,we employourmethodtoobtainits firstintegral.
Wesupposethat(105)hasasolutionintheform
u
( ξ )
=X( ξ )
. (106)Then(105)canbechangedtoasystemofnonlinearODEsas X=Y,
Y=Z,
Z=rX
(
1−X)
−ω
Y−ε
kXY+ν
k2Zμ
k3 , (107)Wesupposethat(107)admitsafirstintegralintheform
p=0, where pde f=Z+f
(
X,Y)
. (108)Sowehave dp
d
ξ
=f1Y+f2Z+Z=f1Y+f2Z+rX
(
1−X)
−ω
Y−ε
kXY+ν
k2Zμ
k3 =f1Y+rX
(
1−X)
−ω
Y−ε
kXYμ
k3 +f2+
ν μ
k Z.(109) Accordingtoourmodification,wetake
f1Y+rX
(
1−X)
−ω
Y−ε
kXYμ
k3 +(
f2+ν
μ
k)
Z=(
f2+ν
μ
k)
·p=(
f2+ν
μ
k)
·(
Z+f(
X,Y))
, (110)whichcanbesimplifiedas
f1Y+rX
(
1−X)
−ω
Y−ε
kXYμ
k3 =(
f2+ν
μ
k)
·f. (111)Byobservation,weconsiderthefollowingtwocases:
• CaseI: f= f(X)
Inthiscase,(111)becomes rX
(
1−X)
μ
k3 +(
f−ω
+ε
kXμ
k3)
Y=ν
μ
k·f. (112)Therefore,wetake f−
ω
+ε
kXμ
k3 =0. (113)Solving(113),weobtain f=2
ω
X+ε
kX22
μ
k3 (114)Substituting(114)to(112),weobtain rX
(
1−X)
=ν
μ ( ω
kX+
ε
2X2
)
. (115)Thus,weobtainthefollowingconstraintconditions
ω
k =−
ε
2 =r
μ
ν
. (116)Inoneword,underconstraintconditions(116),wehaveobtainedthefollowingfirstintegralof(105). u+
ω
μ
k3(
u−u2)
=0. (117)Multiplying2uandintegrating(117)once,wehave
(
u)
2= 2ω
3
μ
k3u3−ω
μ
k3u2+C, (118)whereCisintegrationconstant.Thenwehave u=6k3
μ
ω
℘+1, (119)where℘=℘(
ξ
;g2,g3)isWeierstrassellipticfunction,whichadmitsthefollowingequation:℘˙2=4℘3−g2℘−g3. (120)
Theinvariantsg2 andg3aredeterminedby g2=
ω
212
μ
2k6,g3=ω
3−6Cμω
2k3216
μ
3k9 . (121)• CaseII: f=a(X)+b(X)Y Inthiscase,(111)becomes
bY2+
a−
b2+
ν
bμ
k+ω
+ε
kXμ
k3Y+rX
(
1−X)
μ
k3 −(
b+ν
μ
k)
a=0. (122)Sowehave
b=0 (123)
a=b2+
ν
bμ
k+ω
+ε
kXμ
k3 (124)rX
(
1−X)
μ
k3 =(
b+ν
μ
k)
a. (125)From(123),weobtainthatbisaconstanttobedetermined.From(124),wetake a
(
X)
=(
b2+ν
bμ
k+ω
μ
k3)
X+ε
2
μ
k2X2. (126)Substituting(126)into(125),weobtain rX
(
1−X)
μ
k3 =(
b+ν μ
k)
·b2+
ν
bμ
k+ω
μ
k3X+
ε
2
μ
k2X2(127)
From(127),weobtain r
μ
k3 =−(
b+ν μ
k)
·ε
2
μ
k2 (128)b2+
ν
bμ
k+ω
μ
k3 =−ε
2
μ
k2. (129)From(128),weobtain b=−2r
ε
k−ν
μ
k. (130)Substituting(130)into(129),weobtainthefollowingconstraintcondition
8k
μ
r2+4krεν
+2ωε
2+kε
3=0. (131)Inoneword,undertheconstraintcondition(131),wehaveobtainedthefollowingfirstintegralof(105). u−
2rε
k+ν μ
ku−
ε
2
μ
k2(
u−u2)
=0. (132)Remark 6. Weshouldpointout thatexcepttheveryfirst,therelationsofparameters inthe restsixsolutionsof(105)in Kocak[39]alsomeetourconstraintcondition(131).
Remark7. Tothebestofourknowledge,bothoffirstintegrals(117)and(132)arenew.