the riccati equation method with variable expansion coefficients. i. solving the burgers equation

Upload: makosantoniou

Post on 08-Feb-2018

235 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    1/93

    The Riccati equation methodwith variable expansion coefficients.

    I. Solving the Burgers equation

    Solomon M. Antoniou

    SKEMSYSScientific Knowledge Engineering

    and Management Systems

    37 oliatsou Street, Corinthos 20100, [email protected]

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    2/93

    2

    Abstract

    We introduce the Riccati equation method with variable expansion coefficients.

    This method is used to find travelling wave solutions to the Burgers equation

    0uauuu xxxt =+ . The important new feature of the method lies in the fact that

    the ( dependent) coefficients A and B of the Riccati equation 2BYAY +=

    satisfy their own nonlinear ODEs, which can be further solved by one of the

    known methods, like Jacobi's elliptic equation method, the )G/G( expansion

    method, the projective Riccati equation method, and numerous other methods, like

    the reduction to Ermakov's equation. We also introduce the )G/G( expansion

    method with variable expansion coefficients. The second method presupposes

    Riccati's equation method with variable expansion coefficients. The solutions

    obtained by the )G/G( expansion method are expressed in terms of

    hypergeometric function even in the simplest case. More complicated cases might

    require hypercomputing facilities with symbolic capabilities. Both methods are

    new, computationally demanding, the use of which reveals a rich number of

    solutions not known previously.

    Keywords: Riccati method, nonlinear evolution equations, traveling wave

    solutions, Burgers equation, exact solutions.

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    3/93

    3

    1. Introduction.

    Nonlinear partial differential equations arise in a number of areas of Mathematics

    and Physics in an attempt to model physical processes, like Chemical Kinetics

    (Gray and Scott [98]), Fluid Mechanics (Whitham [204]), or biological processes

    like Population Dynamics (Murray [154]). In the recent past, there is a number of

    new methods which have been invented in solving these equations. Among the

    new methods are

    (1)the inverse scattering transform method(Ablowitz, Kaup, Newell andSegur [17], Ablowitz and Clarkson [18], Ablowitz and Segur [19], Drazin

    and Johnson [50], Gardner et. al. [90]-[92], Gelfand and Levitan [93], Kay

    and Moses [121], Lax [132], Marchenko [147], Miura [149]-[152],

    Novikov, Manakov, Pitaevskii and Zakharov [158], Ramm [170], Wadati

    [187], Wadati, Sanuki and Konno [188] )

    (2)Hirotas bilinear method(Hirota [111] and [112], Hientarinen [110],Hereman and Zhuang [109] )

    (3)theAlgebro-Geometricapproach (Belokolos, Bobenko, Enolskii, Its andMatveev [32] and references)

    (4) theBcklund transformation method(Miura [149], Rogers and Shadwick

    [172])

    (5)thetanh-coth method(Malfliet [142], Malfliet and Hereman [144] and[145], Bekir and Cevikel [31], Wazwaz [199], Fan [72], Abdel-All, Razek

    and Seddeek [3])(6)thesn-cn method(Baldwin et al [27])(7)theF-expansion method(Wang and Li [192], Abdou [6] and [7])(8)theJacobi elliptic function method(Abbott, Parkes and Duffy [2], Chen

    and Wang [44], Fan and Zhang [82], Liu and Li [135] and [136], Liu, Fu,

    Liu and Zhao [137] and [138], Lu and Shi [141], Xiang [206], Yan [209]

    and [210],Yang [214])

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    4/93

    4

    (9)theRiccati equation method(Yan and Zhang [213])(10) the Weierstrass elliptic function method(Kudryashov [124])(11) theExp-function method(He and Wu [104], Naher, Abdullah and Akbar

    [155] and [156], Mohyuddin, Noor and Noor [153], Yildirim and Pinar [218],

    Aslan [26], Bekir and Boz [30])

    (11)the G/G expansion method(Abazari and Abazari [1], Borhanibar andMoghanlu [37], Borhanibar and Abazari [38], Elagan, Sayed and Hamed

    [52], Feng, Li and Wan [84], Jabbari, Kheiri and Bekir [118], Naher,

    Abdullah and Akbar [157], Ozis and Aslan [163], Wang, Li and Zhang

    [194], Zayed [220], Zayed and Gepreel [222] )

    (12)thehomogeneous balance method (Fan [71], Wang, Zhou and Li [191],Yan and Zhang [213])

    (13)thedirect algebraic method(Soliman and Abdou [180])(14)thebasic equation method(Kudryashov [125] ) and its variants, like the

    simplest equation method (Vitanov [184]-[186], Yefimova [217])

    (15)the Cole-Hopftransformation method (Cole [48], Hopf [115], Salas andGomez [174])

    (16)theAdomian decomposition method(Adomian [21] and [22], Abdou [4],Cherruault [46], Cherruault and Adomian [47], El-Wakil, Abdou and

    Elhanbaly [58], El-Wakil and Abdou [59], Rach, Baghdasarian and Adomian

    [169], Wazwaz [196] and [197])

    (17)thePainleve truncated method(Weiss, Tabor and Carnevale [202] and[203], Zhang, Wu and Lou [223], Estevez and Gordoa [69] and [70])

    (18)thehomotopy perturbation method(Taghizadeh, Akbari andGhelichzadeh [183], Yahya et al [207], Liao [134])

    (19)thereduced differential transformation method(Keskin and Oturanc[122], Arora, Siddiqui and Singh [25])

    (20)theLie symmetry method(Lie point symmetries, potential symmetries,nonclassical symmetries, the direct method) (Bluman and Kumei [33],

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    5/93

    5

    Bluman and Anco [34], Bluman and Cole [35], Bluman, Cheviakov and

    Anco [36], Cantwell [40], Hydon [116], Olver [160], Ovsiannikov [162],

    Schwarz [175], Steeb [181], Stephani [182])

    (21)the variational iteration method(He [102], Abdou [8], Wazwaz [198])(22)thefirst integral method(Raslan [171], Feng [85], Feng and Wang [86])(23)the integral bifurcation method(Rui, Xie, Long and He [173] and

    references)

    The implementation of most of these methods was made possible only using

    Symbolic Languages (Grabmeier, Kaltofen and Weispfenning [97]) like

    Mathematica (Baldwin et al [28] and [29], Gktaand Hereman [96]), Macsyma

    (Hereman [106], Hereman and Takaoka [108], Hereman and Zhuang [109]),

    Maple, etc. In many cases some special elimination methods (Wang [189]) and

    computational algorithms were also used (Wu [205]).

    In this paper we introduce the Riccati equation method with variable expansion

    coefficients and we find traveling wave solutions of the Burgers equation.The paper is organized as follows: In Section 2 we introduce the basic ingredients

    of the method used. In Section 3 we consider Burgers equation and Riccatis

    equation method of solution where the expansion coefficients are not constants,

    i.e. they depend on the variable . In this section we consider first the Riccati

    expansion method and then the extended Riccati method. In the first case we find

    closed-form solutions for )(u expressed in terms of hyperbolic tangent functions.

    In the second case we establish that A and B are proportional each other. We

    also find a closed form expression for the function )(u expressed in terms of

    )(A . The function )(A is proved to satisfy a second order nonlinear ordinary

    differential equation solved in Appendices A, B and C. In Section 4 we consider

    the various forms the function )(u takes using the various solutions found in the

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    6/93

    6

    Appendices. In Section 5 we consider the )G/G( expansion method with

    variable coefficients as a method of solution of the Burgers equation.

    2. The Method.

    We consider an evolution equation of the general form

    ),u,u,u(Gu xxxt L= or ),u,u,u(Gu xxxtt L= (2.1)

    where u is a sufficiently smooth function. We introduce a new variable given

    by

    )tx(k = (2.2)

    where k and are constants. Changing variables, since

    )(u)k(u t = , )(uku x = , )(uku2

    xx = , (2.3)

    equations (2.1) become ordinary nonlinear differential equations

    =

    L,

    d

    udk,

    d

    duk,uG

    d

    du)k(

    2

    22 (2.4)

    or

    =

    L,

    d

    udk,

    d

    duk,uG

    d

    udk

    2

    22

    2

    22 (2.5)

    Solutions of evolution equations depending on )tx(k are calledtraveling

    wavesolutions.

    Equations (2.4) or (2.5) will be solved considering expansions of the form

    =

    =n

    0k

    kkYa)(u (2.6)

    or

    ==

    +=n

    0kkk

    n

    0k

    kk

    Y

    bYa)(u (2.7)

    where all the expansion coefficients depend on the variable ,

    )(aa kk , )(bb kk for every n,,2,1,0k L=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    7/93

    7

    contrary to the previously considered research where the expansion coefficients

    were considered as constants. The function )(YY satisfy Riccatis

    equations

    2BYA)(Y += (2.8)

    or

    2YRYQP)(Y ++= (2.9)

    where again all coefficients A, B and P, Q, R depend on the variable .

    In solving equations (2.4) or (2.5), we consider the expansions (2.6) or (2.7) and

    then we balance the nonlinear term with the highest derivative term of the function

    )(u , which determines n (the number of the expansion terms). Equating similar

    powers of the function )(Y , we can determine the various coefficients and thus

    find the solution of the equation considered.

    Our motivation in using Riccati's method with variable expansion coefficients,

    was to obtain traveling wave solutions expressed in terms of special functions

    (Hypergeometric, Bessel, etc). Introducing variable expansion coefficients, weobtain variable coefficients A and B of the Riccati equation (2.8). Converting

    Riccati equation into a second order linear ordinary differential equation with

    variable coefficients, its solutions might be expressed in terms of special

    functions.

    3. The Burgers equation and its solutions.

    Burgers equation (Burgers [39]) was introduced in an attempt to model a theory of

    turbulence. There is a number of explicit solutions based on Lie symmetries, either

    classical (Antoniou [24], Liu, Li and Zhang [139], Ouhadan, Mekkaoui and El

    Kinani [161]), or nonclassical (Gandarias [88], Mansfield [146], Qin, Mei and Xu

    [168]). Some solutions have also been found using the tanh-method (Hereman and

    Malfliet [107]) or some of its descendants, like the Exp-Function Method (Ebaid

    [51]) and the Homotopy perturbation method (Taghizadeh, Akbari and

    Ghelichzadeh [183]). On the other hand Burgers equation can be reduced to the

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    8/93

    8

    heat equation through the famous Cole-Hopf transformation (Cole [48], Hopf

    [115], Olver [160] and Mansfield [146]).

    We consider next Burgers equation in the form

    0uauuu xxxt =+ (3.1)

    and try to find traveling wave solutions of this equation. We introduce a new

    variable given by

    )tx(k = (3.2)

    where k and are constants. Changing variables, since

    )(u)k(u t = , )(uku x = and )(uku2

    xx =

    equation (3.1) becomes an ordinary nonlinear differential equation

    0)(uka)(u)(uk)(u)k( 2 =+ (3.3)

    Integrating once the above equation, we obtain

    02 c)(uka)(u

    2

    1)(u)( =+ (3.4)

    where 0c is a constant.

    We consider the following cases in solving equation (3.4).

    3.1. First Case. The Riccati Method.

    We consider the solution of equation (3.4) to be of the form

    =

    =n

    0k

    kkYa)(u (3.5)

    where )(aa kk for every n,,2,1,0k L= .

    3.1.1. Method I. We first consider that )(YY satisfies Riccatis equation

    2BYA)(Y += (3.6)

    where A and B depend on the variable : )(AA and )(BB .

    We substitute (3.5) into (3.4) and take into account Riccatis equation (3.6). We

    then balance the first order derivative term with that of the nonlinear term. The

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    9/93

    9

    order of the nonlinear term )(u2 is n2 and that of the first order derivative term

    is 1n2)1n( +=+ . We thus get the equation 1nn2 += from which we obtain1n= . Therefore

    Yaa)(u 10+= (3.7)

    We first calculate )(u from the above equation, taking into account Riccatis

    equation (3.6) and that the various coefficients 0a , 1a , A and B depend on .

    We find

    2

    1110

    YBaYa)Aaa()(u +++= (3.8)

    Because of (3.7) and (3.8), equation (3.4) becomes, arranged in powers of Y,

    Y}a)a(aka{cakaaa2

    1Aaka 101000

    201 +

    +

    0Y)Bka2a(a2

    1 211 = (3.9)

    We now have to determine the coefficients 0a , 1a , A and B from the above

    equation, equating to zero the coefficients of Y.

    Equating to zero the coefficient of 2Y , we determine the coefficient 1a :

    Bka2a1= (3.10)

    Equate to zero the coefficient of Y,

    0a)a(aka 101 =

    we can determine the coefficient 0a :

    BBka

    aakaa

    1

    10 +=+= (3.11)

    Equate now to zero the constant term:

    0cakaaa2

    1Aaka 000

    201 =+

    (3.12)

    This equation takes on the form

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    10/93

    10

    Dka2

    )2a(aaAa 0001 +

    += (3.13)

    where D is a constant (ka

    cD 0 ).

    Equation (3.13) - because of (3.10) and (3.11) - can be written as

    Dka2

    B

    Bka

    B

    Bka

    B

    BkaA)Bka2( +

    +

    +

    +

    =

    or

    MB

    B

    4

    1

    B

    B

    2

    1AB

    2

    +

    +

    = (3.14)

    where M is another constant (22

    20

    ka4

    c2M

    ).

    What we need, in order to determine )(u , is 0a , 1a and Y (see equation (3.7)).

    On the other hand, Y is determined by Riccati's equation 2BYA)(Y += once

    A and B are known quantities. We thus need four quantities: 0a , 1a , A and B.

    We do have however three equations only, (3.10), (3.11) and (3.14). There is one

    equation missing. However there is no need for an extra equation, as we shall see

    below. First of all, Riccatis equation 2BYA)(Y += under the substitution

    )(w

    )(w

    )(B

    1)(Y

    = (3.15)

    takes on the form of a linear second order ordinary differential equation

    0)(wBA)(wB

    B)(w =+

    (3.16)

    with unknown function )(w .

    We now transform equation (3.16) under the substitution

    )(y)(B)(w = (3.17)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    11/93

    11

    The derivatives of the function )(u transform as

    += yBBy2

    1

    B

    1

    )(w (3.18)

    ++

    = yByBy

    B

    )B(

    4

    1B

    2

    1

    B

    1)(w

    2

    (3.19)

    Equation (3.16), because of (3.14) and (3.17)-(3.19), takes on the form

    +

    +

    ++

    yBBy

    2

    1

    B

    B

    B

    1yByBy

    B

    )B(

    4

    1B

    2

    1

    B

    12

    0yBMB

    B

    4

    1

    B

    B

    2

    12

    =

    +

    +

    +

    which gives us upon multiplying by B , the simple equation (some miraculous

    cancellations take place)

    0)(yM)(y =+ (3.20)

    Considering various forms of the constant M, we can determine )(y and then

    )(w from (3.16). For 2nM= , equation (3.20) admits the general solution

    )nsin(C)ncos(C)(y 21 += , whereas for2nM = admits the general solution

    )nsinh(C)ncosh(C)(y 21 += . If 0M= then 21 CC)(y += .

    The function )(Y can be determined from (3.15) and then )(u from (3.7),

    calculating first 1a and 0a from (3.10) and (3.11) respectively.

    We thus find that

    +

    +=+=

    )(w

    )(w

    )(B

    1Bka2

    B

    BkaYaa)(u 10

    or

    )(w

    )(wka2

    B

    Bka)(u

    += (3.21)

    Since )(y)(B)(w = , we obtain

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    12/93

    12

    )(y

    )(y

    )(B

    )(B

    2

    1

    )(w

    )(w

    +

    =

    Using the above expression, we obtain from (3.21) that

    )(y

    )(yka2)(u

    = (3.22)

    This is quite aremarkable result :no matter what the coefficients of Riccatis

    equation are, we arrive at the equation (3.22) where )(y satisfies equation (3.20).

    We thus obtain the following three solutions, depending on the values of the

    constant M (which appears in equation (3.20))

    )ntan(C

    )ntan(C1nka2)(u

    +

    = for 2nM= (3.23)

    )ntanh(C

    )ntanh(C1nka2)(u

    +

    += for 2nM = (3.24)

    +=

    C1

    Cka2)(u for 0M= (3.25)

    where 21 C/CC = , with 1C and 2C are the two arbitrary constants of the general

    solution of the differential equation (3.20).

    3.1.2. Method II. We consider instead of (3.6), the equation

    2YRYQPY ++= (3.26)

    where P, Q and R are functions depending on .

    In this case considering again ( )(aa 00 = , )(aa 11 = )

    Yaa)(u 10+= (3.27)

    we find

    211110 RYaY)Qaa()Paa()(u ++++= (3.28)

    taking into account equation (3.26). Equation (3.4) then becomes

    +++ Y)}Qaa(kaa)a{(Y)Rka2a(a2

    11110

    211

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    13/93

    13

    01000 cPakaaka)2a(a2

    1=+ (3.29)

    Equating to zero the coefficients of the powers of Y, we obtain the following

    system of equations

    Rka2a1= (3.30)

    0)Qaa(kaa)a( 1110 =+ (3.31)

    01000 cPakaaka)2a(a2

    1= (3.32)

    Equation (3.31), because of (3.30), takes on the form

    ++=

    R

    RQkaa0 (3.33)

    From (3.32) we obtain

    Dka2

    )2a(aaPa 0001 +

    += (D is a constant)

    Using (3.30) and (3.33), we obtain from the above equation

    MRRQ

    41

    RRQ

    21PR

    2

    +

    ++

    += (3.34)

    Using the usual substitution

    w

    w

    R

    1Y

    = (3.35)

    equation (3.26) transforms into

    0wPRw

    R

    RQw =+

    + (3.36)

    Substituting the product PR in the above equation given by (3.34), we arrive at

    0wMR

    RQ

    4

    1

    R

    RQ

    2

    1w

    R

    RQw

    2

    =

    +

    ++

    ++

    + (3.37)

    We introduce another function f by

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    14/93

    14

    +=

    R

    RQ

    f

    f (3.38)

    Equation (3.37) then becomes

    0wMf

    f

    4

    1

    f

    f

    2

    1w

    f

    fw

    2

    =

    +

    +

    +

    (3.39)

    Under the substitution

    yfw= (3.40)

    equation (3.39) takes on the form

    0yMy =+ (3.41)

    We thus have obtained

    ++=

    R

    RQkaa0 , Rka2a1= and

    w

    w

    R

    1Y

    =

    where

    yfw= with

    +=

    R

    RQ

    f

    f and 0yMy =+ .

    Therefore

    +

    +=

    w

    w

    R

    1Rka2

    f

    fka)(u

    or

    w

    wka2

    f

    fka)(u

    += (3.42)

    Since, using (3.40),

    y

    y

    f

    f

    2

    1

    w

    w +

    =

    (3.43)

    we obtain from (3.39) that

    y

    yka2)(u

    = (3.44)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    15/93

    15

    We thus arrive again at the same result: no matter what the coefficients of

    Riccatis equation (3.26) are, we arrive at the equation (3.41), where )(y satisfies

    equation (3.41), which are equations (3.22) and (3.20) respectively of Method I.

    Conclusion. Burgers equation 0uauuu xxxt =+ under the substitution

    )tx(k = transforms into 02 c)(uka)(u

    2

    1)(u)( =+ . Considering

    the expansion Yaa)(u 10+= where Y satisfies either one of the Riccati

    equations 2BYA)(Y += or 2YRYQPY ++= , where all the coefficients

    depend on , we obtain the three solutions (3.23)-(3.25), depending on the values

    of the constant M.

    3.1.3. Method III. Equation (3.4) is a Riccati equation by itself and under the

    substitutionw

    w)ka2(u

    = transforms into the linear second order equation

    0wka2

    cw

    kaw 0 =

    + (3.45)

    with constant coefficients. Let22

    02

    ka

    cka2+ be the discriminant of the

    auxiliary to (3.45) equation. (I) If 202 ncka2 =+ , n real, then

    +

    =

    ka2exp

    ka2

    nsinC

    ka2

    ncosCw 21 and thus

    +

    =

    ka2

    ntanC

    ka2

    ntanC1

    nu , 21 C/CC = . (II) If 202 ncka2 =+ , n real,

    then

    +

    =

    ka2exp

    ka2

    nsinhC

    ka2

    ncoshCw 21

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    16/93

    16

    and thus

    +

    +

    =

    ka2

    ntanhC

    ka2

    ntanhC1

    nu , 21 C/CC = . (III) If 0cka2 02

    =+ ,

    then

    +=

    ka2exp)CC(w 12 and thus

    +=

    C1

    Cka2u ,

    21 C/CC = . We thus see that we derive the same set of solutions as in the

    previous two sections: it suffices to perform the rescaling n)ka2(n .

    The reader might wonder why we have considered all these three cases instead toconsider only the last one, since we get identical solutions. The reason is that the

    first two methods will be used as a prelude to the extended Riccati equation

    method, considered next, where we obtain quite different solutions.

    3.2. Second Case. The Extended Riccati Method.

    In this case we consider the expansion

    ==

    +=n

    1k k

    kn

    0k

    kk

    Y

    bYa)(u

    and balance the first order derivative term with the second order nonlinear term of

    (3.4). We then find 1n= and thus

    Y

    bYaa)(u 110 ++= (3.46)

    where again require all the coefficients 0a , 1a and 1b depend on , and Y

    satisfies Riccatis equation2

    BYAY += . From equation (3.42) we obtain (taking

    into account 2BYAY += )

    2112

    11110Y

    bA

    Y

    bBYaYa)BbAaa()(u

    ++++= (3.47)

    Therefore equation (3.4), under the substitution (3.46) and (3.47), becomes

    ++++ )BbAaa(ka)ba2a(2

    1a)( 11011

    200

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    17/93

    17

    +++ 211110 Y)Bka2a(a2

    1Y}akaa)a{(

    0211110 c

    Y

    )Aka2a(b

    2

    1

    Y

    bkab)a(=

    ++

    + (3.48)

    Equating the coefficients of Y to zero, obtain a system of differential equations

    from which we can determine the various coefficients. Equating to zero the

    coefficients of 2Y and 2Y/1 , we find that

    Bka2a1= and Aka2b1 = (3.49)

    respectively. Equating to zero the coefficients of Y and Y/1 and taking into

    account the values of 1a and 1b , we find

    B

    Bkaa0

    += and

    A

    Akaa0

    += (3.50)

    respectively. The above two equations imply that A and B are proportional each

    other:

    AsB 2 = (3.51)

    with 2s being the proportionality factor with s real. We do not consider the

    case AsB 2 = , because this choice leads to imaginary type of solutions (see

    equation (3.58) below).

    The zeroth order term appearing in (3.48) takes on the form, taking into account

    the values of 1a and 1b :

    2

    )2a(aakaAB)ka8( 000

    22 +=

    The above equation, after introducing the value of 0a expressed in terms of A,

    takes on the form

    22

    m2A

    A

    2

    1

    A

    ABA8

    +

    = (3.52)

    where m is defined by

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    18/93

    18

    ka2

    1m

    = , 0m> (3.53)

    From equation (3.52), since AsB 2 = , we obtain the following second order

    nonlinear ordinary differential equation

    0)(As8)(Am2))(A(2

    3)(A)(A 42222 =+ (3.54)

    Equation (3.54) can be solved using a number of methods (the auxiliary equation

    method, the G/G expansion method, the reduction to Ermakovs equation, the

    projective Riccati equation expansion method, etc.). This has been done in

    Appendix A, where we have found a considerable number of different solutions.

    We now have to determine the function )(w from the equation (3.16)

    0)(wBA)(wB

    B)(w =+

    (3.55)

    This equation written in terms of A, using (3.47), takes on the form

    0)(wAs)(w

    A

    A)(w 22 =

    (3.56)

    The previous equation can be written as (see also the Notebelow, for another

    method of solution) 0wAsA

    Aw

    A

    w 22

    =

    which is equivalent to

    0wAsA

    w 2 =

    (3.57)

    Multiplying byA

    w, we obtain 0wws

    A

    w

    A

    w 2 =

    which is equivalent to

    0)w(sA

    w 222

    =

    and from this by integration

    0wsA

    w 222

    =

    (3.58)

    where we have put the constant of integration equal to zero. We thus get

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    19/93

    19

    )(As)(w

    )(w=

    (3.59)

    (Note. Another method of solution of equation (3.57) would be the following:

    Under the substitutionA

    w= , since

    2A

    wA

    A

    w

    = , equation (3.57) takes on the

    form 0ww

    s2 =

    , which is equivalent to 0)w(s)( 222 = and by

    integration (we have put the integration constant equals to zero) 0ws 222 =

    and then )(ws)( = , i.e. )(wsA

    w

    =

    , which is (3.59)).

    Therefore

    s

    1

    )(w

    )(w

    )(As

    1

    )(w

    )(w

    )(B

    1)(Y

    2 =

    =

    = (3.60)

    We then obtain the following expression for the function )(u using (3.46), (3.50),

    (3.51) and (3.60):

    )(Aska4)(A)(Aka)(u +=

    (3.61)

    We thus arrive at the following Theorem:

    Theorem. Burgers equation 0uauuu xxxt =+ under the substitution

    )tx(k = transforms into 02 c)(uka)(u

    2

    1)(u)( =+ . Considering

    the expansionY

    bYaa)(u 110 ++= , where Y satisfies the Riccati equation

    2YBA)(Y += , with all the coefficients 0a , 1a , 1b and A, B depending on ,

    we obtain that AsB 2 = and )(u given by the formula

    )(Aska4)(A

    )(Aka)(u

    += , where )(A satisfies the differential equation

    0)(As8)(Am2))(A(2

    3)(A)(A

    42222 =+

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    20/93

    20

    4. The final solutions.

    Using the various solutions of the function )(A calculated in Appendices A, B

    and C, we are able to find the different forms of the function )(u from equation

    (3.61).

    4.I. First Family of Solutions.

    Using (A.10) for )(A , we obtain from (3.61)

    )]mtanh()Cmaa()maCa[(])mtanh(C[

    )]m(tanh1[maka)C1()(u

    1010

    221

    2

    ++

    +=

    +

    +

    +

    )mtanh(C

    1)mtanh(Cmaaska4 10

    where the coefficients 0a and 1a satisfy the relations (four different

    combinations)2

    221

    s16a

    = ,

    2

    220

    s16

    ma = .

    (4.Ia)For

    s4

    a1

    = and

    s4

    ma0 = , we obtain

    )]mtanh(1[])mtanh(C[

    )]m(tanh1[mka)C1()(u

    2

    ++

    ++=

    +

    +

    )mtanh(C

    1)mtanh(C1mka (4.1)

    (4.Ib)Fors4

    a1

    = ands4

    ma0 = , we obtain

    )]mtanh(1[])mtanh(C[

    )]m(tanh1[mka)C1()(u

    2

    ++

    =

    +

    ++

    )mtanh(C

    1)mtanh(C1mka (4.2)

    (4.Ic)Fors4

    a1

    = ands4

    ma0 = , we obtain

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    21/93

    21

    )]mtanh(1[])mtanh(C[

    )]m(tanh1[mka)C1()(u

    2

    ++

    =

    +

    ++

    )mtanh(C

    1)mtanh(C1mkam (4.3)

    (4.Id)Fors4

    a1

    = ands4

    ma0 = , we obtain

    )]mtanh(1[])mtanh(C[

    )]m(tanh1[mka)C1()(u

    2

    +

    +=

    +++)mtanh(C1)mtanh(C1mka (4.4)

    4.II. Second Family of Solutions.

    Using (A.24) for )(A , we obtain from (3.61)

    +

    +

    +=

    )(H

    )(Faa)(Ha

    )(FDaa

    am4exp)}(Fa)(Hma4{m4

    ka)(u

    102

    0

    00

    120

    231

    2

    +

    )(H

    )(Faaska4 10

    where2

    21

    s16

    1a = and

    2

    220

    s4

    ma = (four different combinations).

    (4.IIa) For s4

    1

    a1= and s2

    m

    a0= , we obtain

    +

    +

    +=m

    m2m22

    e)(H)(H

    )(Fm2

    e)(H

    )(FDs2e

    )(H

    )(Fs8mm

    s

    ka2)(u

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    22/93

    22

    +

    )(H

    )(Fm2ka (4.5)

    where = m22

    es

    mD)(F and ++= m2e

    s2

    mDE)(H (E, D constants).

    Since

    )m(tanhDs2

    mED

    s2

    mE

    )m(tanhs

    mD

    s

    mD

    )(H

    )(F

    22

    ++

    ++

    ++

    =

    equation (4.5) can be written as

    +=s

    ka2)(u

    )(Z)(Z

    )(Xm2

    )mtanh()(Z

    )(X)Dm4(s2m

    )(Z

    )(X)Dm4(s2m 223223

    +

    ++

    +

    +

    )(Z

    )(Xm2ka (4.6)

    where

    )m(tanhs

    mD

    s

    mD)(X

    22

    ++

    = (4.7)

    )m(tanhD

    s2

    mED

    s2

    mE)(Z

    ++

    ++= (4.8)

    (4.IIb) Fors4

    1a1= and

    s2

    ma0 = , we obtain

    +

    +=m

    m2m22

    e)(H)(H

    )(Fm2

    e)(H

    )(FDs2e

    )(H

    )(Fs8mm

    s

    ka2)(u

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    23/93

    23

    +

    )(H

    )(Fm2ka (4.9)

    where = m22

    es

    mD)(F and += m2e

    s2

    mDE)(H (E, D constants).

    Since

    )m(tanhDs2

    mED

    s2

    mE

    )m(tanhs

    mD

    s

    mD

    )(H

    )(F

    22

    ++

    +

    +

    =

    equation (4.9) can be written as

    +=s

    ka2)(u

    )(Z)(Z

    )(Xm2

    )mtanh()(Z

    )(X)Dm4(s2m

    )(Z

    )(X)Dm4(s2m 223223

    ++

    +

    )(Z

    )(Xm2ka (4.10)

    where

    )m(tanhs

    mD

    s

    mD)(X

    22

    +

    = (4.11)

    )m(tanhD

    s2

    mED

    s2

    mE)(Z

    ++

    += (4.12)

    (4.IIc) Fors4

    1a1 = and

    s2

    ma0 = , we obtain

    +

    +=m

    m2m22

    e)(H)(H

    )(Fm2

    e)(H

    )(FDs2e

    )(H

    )(Fs8mm

    s

    ka2)(u

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    24/93

    24

    )(H

    )(Fm2ka (4.13)

    where += m22

    es

    mD)(F and ++= m2e

    s2

    mDE)(H (E, D constants).

    Since

    )m(tanhDs2

    mED

    s2

    mE

    )m(tanhs

    mD

    s

    mD

    )(H

    )(F

    22

    +

    ++

    +

    =

    equation (4.13) can be written as

    =s

    ka2)(u

    )(Z)(Z

    )(Xm2

    )mtanh()(Z

    )(X)Dm4(s2m

    )(Z

    )(X)Dm4(s2m 223223

    ++

    )(Z

    )(Xm2ka (4.14)

    where

    )m(tanhs

    mD

    s

    mD)(X

    22

    += (4.15)

    )m(tanhD

    s2

    mED

    s2

    mE)(Z

    +

    ++= (4.16)

    (4.IId) Fors4

    1a1 = and

    s2

    ma0 = , we obtain

    +

    +

    =m

    m2m22

    e)(H)(H

    )(Fm2

    e)(H

    )(FDs2e

    )(H

    )(Fs8mm

    s

    ka2)(u

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    25/93

    25

    +

    )(H

    )(Fm2ka (4.17)

    where += m22

    es

    mD)(F and += m2e

    s2

    mDE)(H (E, D constants).

    Since

    )m(tanhDs2

    mED

    s2

    mE

    )m(tanhs

    mD

    s

    mD

    )(H

    )(F

    22

    +++

    +

    +

    +

    =

    equation (4.17) can be written as

    =s

    ka2)(u

    )(Z)(Z

    )(Xm2

    )mtanh()(Z

    )(X)Dm4(s2m

    )(Z

    )(X)Dm4(s2m 223223

    +

    ++

    +

    +

    )(Z

    )(Xm2ka (4.18)

    where

    )m(tanhs

    mD

    s

    mD)(X

    22

    +

    += (4.19)

    )m(tanhD

    s2

    mED

    s2

    mE)(Z

    +++

    += (4.20)

    4.III. Third Family of Solutions.

    Using (A.31) for )(A , we obtain from (3.61)

    +=

    222m212

    m421

    222

    2

    sm16)eCCm2(

    ]eC)s4C(m4[m2ka)(u

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    26/93

    26

    222m212

    m21

    2

    sm16)eCCm2(

    eCm4ska4

    (4.21)

    The above expression can also be written as

    }Cmska16);s,m(U{);s,m(V

    )]m(tanh1[mka2)(u 1

    22

    ++= (4.22)

    where

    )m2tanh(]sm16C)1m4[(sm16C)1m4();s,m(U 22212222

    12 += (4.23)

    = 21212 )}mtanh()CCm2()CCm2({);s,m(V m

    )]m(tanh1[)]m2(tanh1[sm16 222 + (4.24)

    4.IV. Fourth Family of Solutions.

    The projective Riccati equation method provides us with eight families of

    solutions, each family containing some subfamilies of solutions. In Appendix A,

    (Section A.IV) we have found twenty one solutions in total. To save space, the

    reader can easily write down the various expressions for )(u , using the different

    solutions of )(A found in that Section and (3.61). The complete set is to be

    reported in electronic form somewhere else.

    4.V. Fifth Family of Solutions.

    We now use (A.71)-(A.74) for )(A and the general expression (3.61) and we

    find four additional families of solutions.

    (4.Va)Using (A.71) and (3.61), we have

    +

    ++=

    m2

    m1

    m

    m2

    m1

    m2

    eCeC

    eska4

    eCeC

    eCm2ka)(u

    or

    )mtanh()CC()CC(

    )]mtanh(1[Cmka2)(u

    2121

    2

    ++

    +=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    27/93

    27

    )mtanh()CC()CC(

    )mtanh(1ska4

    2121 ++

    + (4.25)

    (4.Vb)Using (A.72) and (3.61), we have

    =

    mm1

    m

    mm1

    m1

    2

    es2emC

    emska4

    es2emC

    eCm2ka)(u

    or

    )mtanh()s2mC()s2mC(

    )]mtanh(1[Ckam2)(u

    11

    12

    ++

    +=

    )mtanh()s2mC()s2mC()mtanh(1skam4

    11 ++ (4.26)

    (4.Vc)Using (A.73) and (3.61), we have

    +

    +=

    mm1

    m

    mm1

    m1

    2

    es2emC

    emska4

    es2emC

    eCm2ka)(u

    or

    )mtanh()s2mC()s2mC(

    )]mtanh(1[Ckam2)(u

    11

    12

    ++

    +=

    )mtanh()s2mC()s2mC(

    )mtanh(1skam4

    11 ++

    (4.27)

    (4.Vd)Using (A.74) and (3.61), we have (we find 0)(A = in this case)

    ++

    ++=

    m2

    22m1

    m2

    m1

    2

    eCs4s4eCsm2

    eCsm2sm2eCmska4)(u

    or

    kam2)(u = (4.28)

    4.VI. Sixth Family of Solutions.

    Using the results of Appendix B and (3.61), we have

    ])(aa[ska4)(aa

    )(aka)(u 10

    10

    1 ++

    += (4.29)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    28/93

    28

    where )( is a solution of the differential equation (B.2), the various coefficients

    corresponding to the Solutions 1-11 of Appendix B. The final formulas are too

    lengthy to write them down here.

    4.VII. Seventh Family of Solutions.

    Using the results of Appendix C, we obtain

    ++

    +=

    )s,m(F

    )s,m(F

    eDE

    e)s,m(FD

    G

    G and

    ++=

    )s,m(F

    )s,m(F2

    eDE

    e)s,m(F

    G

    G

    We thus have

    ++

    +=

    +=

    )s,m(F

    )s,m(F010

    eDE

    e)s,m(FD

    s4

    1aG

    Gaa)(A

    and

    =

    +

    +

    =

    G

    Gaa

    G

    G

    G

    Gaa

    )(A

    )(A

    10

    2

    10

    ++

    +

    ++

    +

    ++

    =

    )s,m(F

    )s,m(F

    0

    2

    )s,m(F

    )s,m(F

    )s,m(F

    )s,m(F2

    0

    eDE

    e)s,m(FD

    s4

    1a

    eDE

    e)s,m(FD

    eDE

    e)s,m(F

    s4

    1a

    The final expression for )(u , follows from (3.61) and the above formulas.

    The coefficient 0a satisfies equation (C.7), which is actually equation (A.1) and

    therefore admits allthe solutions reported in Appendices A, B and C.

    5. Solution of the Burgers equation using the G'/G-expansion.

    In this section we solve Burgers equation using the

    G

    Gexpansion. It will be

    clear by the end of this section that we need the solutions found previously using

    the Riccati equation methods. Using the expansion

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    29/93

    29

    +=

    G

    Gaa)(u 10 (5.1)

    assuming that the coefficients 0a and 1a are dependent, Burgers equation

    (3.4) becomes

    ++

    +

    2

    1010G

    Gaa

    2

    1

    G

    Gaa)(

    0

    2

    110 cG

    G

    G

    Ga

    G

    Gaaka =

    +

    + (5.2)

    Upon expanding the above equation and equating the coefficients of G to zero,

    we obtain

    Coefficient of 0G :

    00200 cakaa

    2

    1a)( =+ (5.3)

    Coefficient of 1G :

    0GakaGakaGaaa)(11101

    =+ (5.4)

    Coefficient of 2G :

    0)G(aka)G(a2

    1 21

    221 =+ (5.5)

    From equation (5.3) we conclude that )(aa 00 is a solution of the equation

    (3.4). Assuming that 0a1 we obtain from (5.5) that

    ka2a1 = (5.6)

    showing that 1a is a constant.

    Finally from equation (5.4) we obtain

    0G)aka(Gka 0 =++ (5.7)

    The above equation can be reduced to a linear first order differential equation and

    can be solved in principle provided that 0a is a known quantity. However the

    coefficient 0a satisfies equation (5.3), which has been solved, using the Riccati

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    30/93

    30

    equation method with variable expansion coefficients. The solutions of equation

    (5.3) are actually the various expressions of the function )(u of Section 4. The

    solutions of (5.3) substituted into (5.7) lead to equations with rather complicated

    solutions, despite the fact that equation (5.7) is a linear differential equation.

    It is obvious that we could use some other methods in solving equation (5.3),

    using, say, the Riccati equation method with constant coefficients. Using the

    expansion Ybba 100 += with2

    YMLY += ( 0b , 1b , L, and M constants) and

    substituting into (5.3), after equating to zero the coefficients of Y, we obtain the

    following set of equations

    0bMkab2

    111 =

    , 0)b(b 01 = , 0

    2010 cb

    2

    1bLkab =+

    From the previous equations we obtain the following two sets of solutions

    Solution 1.

    =0b , Mka2b1= ,M

    1

    ka4

    c2L

    220

    2

    +

    =

    Solution 2.

    =0b ,L

    1

    ka2

    c2b 0

    2

    1 +

    = ,L

    1

    ka4

    c2M

    220

    2

    +

    =

    For Solution 1, Riccati's equation 2YMLY += becomes

    2

    220

    2

    YMM

    1

    ka4

    c2Y +

    +=

    which admits the solution

    +

    ++= )C(

    ka2

    c2tanh

    Mka2

    c2Y 1

    02

    02

    Therefore

    +

    ++= )C(

    ka2

    c2tanhc2a 1

    02

    02

    0

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    31/93

    31

    For Solution 2, Riccati's equation 2YMLY += becomes

    222

    0

    2

    YL1

    ka4c2LY +=

    which admits the solution

    +

    +

    += )C(

    ka2

    c2tanh

    c2

    Lka2Y 1

    02

    02

    Therefore

    +++= )C(

    ka2c2tanhc2a 1

    0

    2

    02

    0 (5.8)

    We thus see that 0a is given by the same expression in both solutions.

    Because of the expression (5.8) for 0a , equation (5.7) becomes

    KG2

    KtanhKK1G =

    ++ (5.9)

    where we have put 0c0 = , 0C1= and kaK = for simplicity.

    Equation (5.9) is a rather complicated equation, which admits a closed-form

    solution expressed in terms of the hypergeometric function. Details of the solution

    are given in Appendix D. Using the results of that Appendix, we obtain the

    following expression for the function )(u :

    = ka2

    2

    Ktanh)(u

    22)K1(

    12

    2)K1(1

    22

    K)1K3K2(K2);K(ZeK)1K(4);K(VK

    2

    Khsec]eK)1K3K2(K2);K(UK[

    +++

    ++

    (5.10)

    The functions );K(U , );K(V and );K(Z are defined in Appendix D. The

    function );K(Z is expressed in terms of the hypergeometric function.

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    32/93

    32

    We thus see that even in the simplest case, we obtain very complicated equations.

    In more complex situations, we might need supercomputing facilities with

    symbolic capabilities. We also conclude that in order to find solutions using the

    )G/G( expansion method with variable expansion coefficients, we have to

    consider the Riccati expansion method with variable coefficients first.

    Appendix A.

    In this Appendix we shall solve equation (3.54):

    0)(As8)(Am2))(A(

    2

    3)(A)(A 42222 =+ (A.1)

    A.I) First Method. We consider an expansion of the form

    =

    =n

    0k

    kk )(a)(A (A.2)

    where )( satisfies Jacobi's differential equation

    42)(

    d

    d++=

    (A.3)

    (In Appendix B we consider a full fourth order polynomial).

    Upon substitution of (A.2) into (A.1) and balancing 4A with either AA or

    2)A( , and taking into account (A.3), we obtain 1n= . We thus substitute

    )(aa)(A 10 += (A.4)

    Since

    4211 a)(a)(A ++== and )2(a)(A

    31 +=

    equation (A.1) becomes

    ++++ 3212

    1042

    122

    1 )as16(aa2)2

    1as8(a

    ++++++ )4

    1mas8(aa4)

    12

    1m

    3

    1as8(a6

    220

    210

    2220

    221

    0a2

    3as8am2

    21

    40

    220

    2 =+ (A.5)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    33/93

    33

    Equating to zero the coefficients of the different powers of , we find

    = 22

    1 s16

    1

    a ,2

    2

    2

    0 ms16

    1

    a = ,2

    m2= ,4

    m= (A.6)

    We thus obtain that

    4222

    2

    4

    m2m

    )( +

    = (A.7)

    where we have redefined 2 .

    The last equation can be written as

    =

    22 m)( (A.8)

    This is a Riccati equation which can be transformed into a second order ODE,

    admitting solution

    )mtanh(C

    1)mtanh(Cm)(

    +

    +

    = (A.9)

    Therefore )(A is given by (A.4) with )( given by (A.9):

    +

    +

    +=

    )mtanh(C

    1)mtanh(Cmaa)(A 10 (A.10)

    where the coefficients 0a and 1a satisfy the relations (four different

    combinations)

    2

    221

    s16a

    = ,

    2

    220

    s16

    ma = (A.11)

    A.II) Second Method. We consider the )G/G( expansion of the form

    kn

    0kk

    G

    Ga)(A

    =

    = (A.12)

    with ka ( n,,1,0k L= ) constants and )(GG = .

    (In Appendix C we consider variable expansion coefficients: )(aa kk ).

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    34/93

    34

    Upon substitution of (A.12) into (A.1) and balancing 4A with either AA or

    2

    )A( , we obtain 1n= . We thus substitute

    +=

    G

    Gaa)(A 10 (A.13)

    Notice that 0a and 1a have nothing to do with the similar coefficients appearing

    in (3.7), (3.24) or (3.42).

    Since

    =

    2

    1 G

    G

    G

    G

    a)(A and

    +

    =

    3

    21 G

    G

    2G

    GG

    3G

    G

    a)(A

    equation (A.1) becomes

    +

    +

    3

    2101 G

    G2

    G

    GG3

    G

    G

    G

    Gaaa

    +

    ++

    +

    22110

    20

    2422

    21

    G

    Ga

    G

    Gaa2am2

    G

    G

    G

    G

    G

    G2

    G

    Ga

    2

    3

    0G

    Ga

    G

    Gaa4

    G

    Gaa6

    G

    Gaa4as8

    441

    3310

    221

    201

    30

    40

    2 =

    +

    +

    +

    +

    From the above equation we obtain the following set of coefficients, equating each

    one of them to zero:

    Coefficient of 4G : 0)G(as8)G(a2

    3)G(a2 441

    2421

    421 = (A.14)

    Coefficient of 3G : 0)G(aas32)G(aa2 33102310 = (A.15)

    Coefficient of 2G :

    221

    2221

    2110 )G(am2)G(a

    2

    3)G()G(a)G()G(aa)3( ++

    0)G(aas48 22120

    2 = (A.16)

    Coefficient of 1G : 0Gaas32Gaam4Gaa 130

    210

    210 =+ (A.17)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    35/93

    35

    Coefficient of 0G : 0as8am2 4022

    02 = (A.18)

    The above system (A.14)-(A.18) is equivalent to

    2

    21

    s16

    1a = ,

    2

    220

    s4

    ma = , Gm4G 2 = ,

    0am4G

    Ga2

    G

    Ga 1

    20

    2

    1 =+

    +

    (A.19)

    The last equation is a quadratic equation with respect to the ratio G/G and

    because its discriminant is zero, we obtain

    1

    0

    a

    a

    G

    G=

    (A.20)

    The above equation combined with the equation Gm4G 2 = , gives us

    0

    12

    a

    am4

    G

    G=

    (A.21)

    and by integration

    ++=

    0

    12321aam4expCCCG (A.22)

    Therefore

    ++

    +

    =

    0

    12321

    0

    12

    0

    1232

    a

    am4expCCC

    a

    am4exp

    a

    am4CC

    G

    G

    or, adjusting the above expression and redefining the constants (i.e. 320 C/CaD =

    and 310 C/CaE= ), we obtain

    ++

    =

    0

    120

    0

    121

    2

    a

    am4expaDE

    a

    am4expam4D

    G

    G

    We thus have

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    36/93

    36

    ++

    +=

    +=

    0

    120

    0

    121

    2

    1010

    a

    am4expaDE

    a

    am4expam4D

    aaG

    G

    aaA (A.23)

    where2

    21

    s16

    1a = and

    2

    220

    s4

    ma = (four different combinations).

    Equation (A.23) can also be written as

    )(H

    )(FaaA 10

    += (A.24)

    where

    =

    0

    121

    2

    a

    am4expam4D)(F (A.25)

    ++=

    0

    120

    a

    am4expaDE)(H (A.26)

    A.III) Third Method. Using the transformation

    )(w

    1)(A2

    = (A.27)

    equation (A.1) transforms into the Ermakov equation (Ermakov [68])

    322 ws4)(wm)(w = (A.28)

    Equation 0)(wm)(w 2 = admits two independent solutions = m1 ew and

    = m2 ew . We consider one of them,= m2 ew and introduce the

    transformation =

    2w

    d, i.e.

    = m2em2

    1, )m2(ln

    m2

    1= and )(wez m = , )(zz = .

    Since

    })(w)(wm{ed

    dz m +=

    and })(w)(wm{ed

    zd 2m32

    2

    +=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    37/93

    37

    Ermakovs equation (A.28) transforms into the equation

    32

    2

    2

    zs4d

    zd = (A.29)

    Using one of the standard methods, i.e. multiplying both members byd

    dz2 or

    introducing the variable p by

    =d

    dzp , we arrive at the equation

    12

    22

    C

    z

    s4

    d

    dz+=

    (A.30)

    From the above equation we obtain the two equations

    12

    2

    Cz

    s4

    d

    dz+=

    and 12

    2

    Cz

    s4

    d

    dz+=

    The first of the above equations admits the solution 2122

    1 CCs4zC +=+

    while the second 2122

    1 CCs4zC +=+ .

    We thus have, squaring both the previous two equations and solving with respect

    to 2z

    12

    222m2122

    Cm4

    sm16)eCCm2(z

    =

    and then, since )(wez m = ,

    = m2

    12

    222m2122 e

    Cm4

    sm16)eCCm2()(w

    We thus obtain, using the above expression and (A.27), that )(A is given by

    222m212

    m21

    2

    sm16)eCCm2(

    eCm4)(A

    =

    (A.31)

    The above expression can be written in terms of )mtanh( . In fact since

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    38/93

    38

    2m2m1

    m2

    12

    )esm4()eCeCm2(

    Cm4)(A

    =

    we can easily transform the above expression into

    22221212

    21

    2

    )]m(tanh1[sm16)]m(tanh)CCm2()CCm2([

    )]m(tanh1[Cm4)(A

    =

    m

    A.IV. Fourth Method. We shall now use the projective Riccati equation

    method (see for example Abdou [5]) in solving equation (A.1). We consider the

    expansion

    ])(gb)(fa[)(fa)(AN

    1iii

    1i0

    =

    ++= (A.32)

    where the functions )(f and )(g satisfy the system

    )(g)(fp)(f = (A.33)

    )(fr)(gpq)(g2 += (A.34)

    ++= )(f

    q

    r)(fr2q

    p

    1)(g 2

    22 (A.35)

    The system of equations (A.33) and (A.34) admitsfive familiesof solutions given

    by

    (I) If 22 = and 0qp < then

    ++=

    qpcoshqpsinhr

    q)(f (A.36)

    +++=qpcoshqpsinhr

    qpsinhqpcoshp

    qp)(g (A.37)

    with

    ++= )(f

    q

    r)(fr2q

    p

    1)(g 2

    2222 (A.38)

    (II) If 22 = and 0qp > then

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    39/93

    39

    ++=

    qpcosqpsinr

    q)(f (A.39)

    ++

    +=

    qpcosqpsinr

    qpsinqpcos

    p

    qp)(g (A.40)

    with

    += )(f

    q

    r)(fr2q

    p

    1)(g 2

    2222 (A.41)

    (III) If 0q= , then

    ++

    =2

    2

    rp1)(f (A.42)

    ++

    +=

    2

    2

    rp

    rp

    p

    1)(g (A.43)

    with

    )(fp

    r2

    p)(fp

    r2

    )(g2

    2

    22

    += (A.44)

    where and are free parameters.

    (IV) If 1p = and 2r= then

    )(rp

    2

    r6

    q)(f += (A.45)

    )(12q

    )(12)(g

    +

    = (A.46)

    where )( satisfies Weirstrassequation

    216

    qp)(

    12

    q)(4))((

    3232 =

    with solution )()( = .

    The relation between f and g is given by

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    40/93

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    41/93

    41

    Upon expanding and rearranging, we obtain an equation which contains a constant

    term and powers of the functions )(f and )(g . We further substitute )(g2 by

    (A.35) wherever powers of the function )(g enter. We thus obtain finally an

    equation which will contain powers of g no greater than one. Equating the

    coefficients of this equation to zero, we obtain the following.

    Constant coefficient:

    + 2122

    12

    042

    02 )b(m

    p

    q2)b()a(s

    p

    q48)a(m2

    0)a(s8)b(sp

    q8 40

    441

    4

    2

    2 = (A.53)

    Coefficient of f:

    +++ 212

    042

    11044

    14

    2102 )b()a(s

    p

    r96)b(aas

    p

    q96)b(s

    p

    rq32aam4

    0aaqp)b(rqa)a(s32)b(mp

    r410

    211

    30

    421

    2 =+ (A.54)

    Coefficient of g:

    0bam4b)a(s32)b(asp

    q3210

    21

    30

    4310

    4 =+ (A.55)

    Coefficient of 2f :

    ++

    ++ 2

    12

    22

    12

    142

    12

    04

    2

    )b(mqp

    )r(2)b()a(s

    p

    q48)b()a(s

    qp

    )r(48

    +

    + 414

    2

    41

    4

    2

    22

    11042

    1

    2

    )b(s

    p

    16)b(s

    p

    r48)b(aas

    p

    r192)b(

    2

    r5

    0aarp3)a(2

    qp)a()a(s48)b(2)a(m2 10

    21

    21

    20

    421

    21

    2 =++++ (A.56)

    Coefficient of 3f

    ++

    +

    10

    22

    12

    144

    14

    2

    2

    aaq

    )r(p2)b()a(s

    p

    r96)b(s

    qp

    )r(r32

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    42/93

    42

    ++

    + 31042

    1

    22

    1104 )a(as32)b(

    q

    )r(r2)b(aas

    qp

    d96

    0)b(aasqp

    r96 2110

    42

    =+ (A.57)

    Coefficient of 4f

    +

    +

    + 2

    12

    22

    1

    22

    12

    12

    14

    2

    )b(q2

    )a(q2

    rp)a(

    q2

    p)b()a(s

    qp

    )r(48

    +

    + 414

    22

    24

    14

    22

    44

    142

    12

    42

    12

    2

    )b(sqp

    8)b(s

    qp

    r8)a(s8)b(

    q2

    r)b(

    q

    r

    0)b(sqp

    r16 41

    4

    22

    2

    =

    (A.58)

    Coefficient of gf

    ++ 3104

    11112

    04

    10112 )b(as

    p

    r64baqpba)a(s96baprbam4

    0)b(as

    p

    q32 311

    4 =+ (A.59)

    Coefficient of gf2

    ++

    +

    10

    23

    104

    2

    baq

    )r(p2)b(as

    qp

    )r(32

    0b)a(as96)b(asp

    r64bapr 1

    210

    4311

    411 =+ (A.60)

    Coefficient of gf3

    0b)a(s32baq

    )r(p)b(as

    qp

    )r(321

    31

    411

    23

    114

    2

    =+

    +

    (A.61)

    Solving the system of the nine simultaneous equations (A.53)-(A.61), using any of

    the known Computer Algebra Systems (Axiom, Macsyma, Maple, Mathematica),

    we obtain a considerable number of solutions, from which we select only the

    nontrivial ones:

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    43/93

    43

    Solution I.20 s4

    ma = ,

    m2

    ba 11

    = , 11 bb = , 1

    2bs8p= ,

    12

    2

    bs2

    mq =

    where is any root of the equation 22 r+= .

    Solution II.20

    s4

    ma = ,

    m2

    ba 11

    = , 11 bb = , 1

    2bs8p = ,1

    2

    2

    bs2

    mq =

    where is any root of the equation 22 r+= .

    Solution III.20 s4

    ma = , 11 aa = , 11 bb = , 2

    1

    21

    221

    2

    )b(

    )b(r)a(m4 = ,

    12bs8p= ,

    12

    2

    bs2

    mq =

    Solution IV.20

    s4

    ma = , 11 aa = , 11 bb = , 2

    1

    21

    221

    2

    )b(

    )b(r)a(m4 = ,

    12bs8p = ,

    12

    2

    bs2

    mq=

    Solution V.20 s4

    ma = , 0a1= , 11 bb = , 1

    2bs4p= ,

    12

    2

    bs4

    mq =

    Solution VI.20

    s4

    ma = , 0a1= , 11 bb = , 1

    2bs8p= ,

    12

    2

    bs2

    mq =

    and r is any root of 0r2 =+ .

    Solution VII. 20 s4

    m

    a = , 0a1= , 11 bb = , 12

    bs8p = ,1

    2

    2

    bs2

    m

    q=

    and r is any root of 0r2 =+ .

    Solution VIII. 0a0= ,ms8

    pa

    21

    = , 0b1= , pp = ,

    p

    m4q

    2

    =

    and is any root of 22 r+= .

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    44/93

    44

    For each set of coefficients and parameters (Solution I Solution VIII) there

    correspond in principle five families of solutions. However not all of them appear

    in the final solutions, because there is a conflict between the values of the

    coefficients and the values of the parameters of the various solutions.

    Family I. This family corresponds to the Solution I

    20 s4

    ma = ,

    m2

    ba 11

    = , 11 bb = , 1

    2bs8p= ,

    12

    2

    bs2

    mq =

    where is any root of the equation 22 r+= .

    From (A.51) and Solution I, we obtain

    )(gb)(fm2

    b

    s4

    m)(A 1

    12

    +

    += (A.62)

    with 12bs8p= ,

    12

    2

    bs2

    mq = and satisfies the equation 22 r+= .

    Sub-family Ia. Since 0m4qp 2 >= , we consider first the choice 22 = .

    The functions )(f and )(g in (A.62) are given by (A.36) and (A.37)respectively:

    )(U

    1

    bs2

    m)(f

    12

    2

    = and

    )(U

    )(U2

    bs2

    m)(g

    12

    2

    =

    where

    )|m|2(cosh)|m|2(sinhr)(U ++=

    We thus obtain

    )(U

    )(Um

    s4

    1

    s4

    m)(A

    22

    =

    where is any root of the equation 2222 r+= .

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    45/93

    45

    Sub-family Ib. If 1p = and 2r= , we have 0= and then 0a1= . In this

    case we have )(gbs4m)(A 12 += where )(g is given by (A.46).

    (Ib.a) For 1p= , we have21

    s8

    1b = and 2m4q = . Therefore

    2m4)(12

    )(12)(g

    = and thus

    222 m4)(12

    )(

    s2

    3

    s4

    m)(A

    +=

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(3

    m4)(4))((6432 +=

    (Ib.b) For 1p = , we have21 s8

    1b = and 2m4q= . Therefore

    2m4)(12

    )(12)(g

    +

    = and thus

    222m4)(12

    )(

    s2

    3

    s4

    m)(A

    +

    =

    Sub-family Ic. If 1p = and25

    r2= , is satisfies the equation

    25

    r24 22 =

    while )(f and )(g are given by (A.48) and (A.49) respectively.

    (Ic.a) For 1p= , we have21

    s8

    1

    m2a

    = ,

    21s8

    1b = and 2m4q = . Therefore

    )(9

    m10

    r3

    m10)(f

    42

    += and

    )(]m4)(12[

    )(m4)(g

    2

    2

    =

    We thus have the following expression for )(A :

    ]m4)(12[)(

    )(

    s2

    m

    r

    1

    )(3

    m

    s24

    m5

    s4

    m)(A

    22

    22

    22

    +

    +=

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 +=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    46/93

    46

    (Ic.b) For 1p = , we have21 s8

    1

    m2a

    = ,

    21 s8

    1b = and 2m4q= . Therefore

    )(9

    m10

    r3

    m10)(f

    42

    = and

    )(]m4)(12[

    )(m4)(g

    2

    2

    =

    We thus have the following expression for )(A :

    ]m4)(12[)(

    )(

    s2

    m

    r

    1

    )(3

    m

    s24

    m5

    s4

    m)(A

    22

    22

    22

    +

    +=

    Note.If 22 = and 0qp > leads to 0m2 < . The case 0q= cannot be

    considered either, since it leads to 0m = and then the coefficient of )(f in (A.62)

    becomes infinite.

    Family II. This family corresponds to the Solution II

    20 s4

    ma = ,

    m2

    ba 11

    = , 11 bb = , 1

    2bs8p = ,1

    2

    2

    bs2

    mq =

    where is any root of the equation 22 r+= .

    From (A.51) and Solution II, we obtain

    )(gb)(fm2

    b

    s4

    m)(A 1

    12

    +

    += (A.63)

    with 12bs8p = ,

    12

    2

    bs2

    mq = and satisfies the equation 22 r+= .

    Sub-family IIa. Since 0m4qp 2 >= , we first consider the case 22 = .

    The functions )(f and )(g are given by (A.39) and (A.40) respectively:

    )(V

    1

    bs2

    m)(f

    12

    2

    = and

    )(V

    )(Z

    bs4

    |m|2)(g

    12

    =

    where

    )|m|2(sin)|m|2(cos)(Z +=

    )|m|2(cos)|m|2(sinr)(V ++=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    47/93

    47

    We thus obtain

    )(V

    )(Z|m|m

    s4

    1

    s4

    m

    )(A 22

    +

    =

    where is any root of the equation 2222 r = .

    Sub-family IIb. If 1p = and 2r= then 0= and thus 0a1= .

    In this case we have )(gbs4

    m)(A 12

    += where )(g is given by (A.46).

    (IIb.a) For 1p= , we have2

    1

    s8

    1b = and 2m4q= . Therefore

    2m4)(12

    )(12)(g

    +

    = and thus

    222 m4)(12

    )(

    s2

    3

    s4

    m)(A

    +

    =

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 =

    (IIb.b) For 1p = , we have21 s8

    1b =

    and 2m4q = . Therefore

    2m4)(12

    )(12)(g

    = and thus

    222m4)(12

    )(

    s2

    3

    s4

    m)(A

    +=

    Sub-family IIc. If 1p = and25

    r2= then satisfies the equation

    25

    r24 22 = .

    The functions )(f and )(g are given by (A.48) and (A.49) respectively.

    (IIc.a) For 1p= , we have 21 s8

    1

    b = , 21 s8

    1

    m2a

    = and2

    m4q= . We then

    have the following expressions for )(f and )(g

    )(9

    m10

    r3

    m10)(f

    42

    += and

    )()](12m4[

    )(m4)(g

    2

    2

    +

    =

    Therefore

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    48/93

    48

    )](12m4[)(

    )(

    s2

    m

    )(3

    m

    r

    1

    s24

    m5

    s4

    m)(A

    22

    22

    22 +

    +

    +

    =

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 =

    (IIc.b) For 1p = , we have21 s8

    1b = ,

    21 s8

    1

    m2a

    = and 2m4q = . We then

    have the following expressions for )(f and )(g

    )(9

    m10

    r3

    m10)(f

    42

    = and

    )()](12m4[

    )(m4)(g

    2

    2

    +

    =

    Therefore

    )](12m4[)(

    )(

    s2

    m

    )(3

    m

    r

    1

    s24

    m5

    s4

    m)(A

    22

    22

    22 +

    +

    +

    =

    Note.If 22 = and 0qp < leads to 0m2 < . The case 0q= cannot be

    considered either, since it leads to 0m = and then the coefficient of )(f in (A.63)

    becomes infinite.

    Family III. This family corresponds to the Solution III

    20s4

    ma = , 11 aa = , 11 bb = , 2

    1

    21

    221

    2

    )b(

    )b(r)a(m4 = ,

    12bs8p= ,

    12

    2

    bs2

    mq =

    From (A.51) and Solution III, we obtain

    )(gb)(fas4

    m)(A 112

    ++= (A.64)

    where2

    1

    21

    221

    2

    )b(

    )b(r)a(m4 = , 1

    2bs8p= ,1

    2

    2

    bs2

    mq =

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    49/93

    49

    Sub-family IIIa. Since 0m4qp 2

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    50/93

    50

    222 m)(3

    )(

    s8

    3

    s4

    m)(A

    =

    Sub-family IIIc. If 1p = and25

    r2= , i.e.

    21

    21

    221

    22

    )b(

    )b(r)a(m4

    25

    r =

    then )(f and )(g are given by (A.48) and (A.49) respectively.

    (IIIc.a) For 1p= , we have21

    s8

    1b = and 2m4q = and thus

    )(9

    m10

    r3

    m10)(f

    42

    += and

    )(]m4)(12[

    )(m4)(g

    2

    2

    =

    Therefore

    ]m)(3[)(

    )(

    s8

    m

    )(3

    m

    r

    1

    3

    ma10

    s4

    m)(A

    22

    2221

    2

    +

    ++=

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 +=

    with21

    221

    2br24am100 = .

    (IIIc.b) For 1p = , we have21 s8

    1b = and 2m4q= and thus

    )(9

    m10

    r3

    m10)(f

    42

    = and

    )(]m4)(12[

    )(m4)(g

    2

    2

    =

    Therefore

    ]m)(3[)(

    )(

    s8

    m

    )(3

    m

    r

    1

    3

    ma10

    s4

    m)(A

    22

    2221

    2

    +

    +=

    Note. If 22 = and 0qp > leads to 0m2 < . The choice 0q= leads to

    0m = .

    Family IV. This family corresponds to the Solution IV

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    51/93

    51

    20 s4

    ma = , 11 aa = , 11 bb = , 2

    1

    21

    221

    2

    )b(

    )b(r)a(m4 = , 1

    2bs8p = ,

    12

    2

    bs2

    mq=

    From (A.51) and Solution IV, we obtain

    )(gb)(fas4

    m)(A 112

    ++= (A.65)

    where21

    21

    221

    2

    b

    bram4 = , 1

    2bs8p = ,

    12

    2

    bs2

    mq=

    Sub-family IVa. Since 0m4qp 2

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    52/93

    52

    where )( is the Weirstrassfunction, satisfying the equation

    27

    m8)(3

    m4)(4))((

    64

    32 +=

    (IVb.b) For 1p = , we have21 s8

    1b = , 2m4q= and thus from (A.46) we get

    2m4)(12

    )(12)(g

    +

    = and then

    222m)(3

    )(

    s8

    3

    s4

    m)(A

    +

    +=

    Sub-family IVc. If 1p = and25r

    2

    = , i.e. 21

    2

    1

    22

    1

    22

    bbram4

    25r = then )(f and

    )(g are given by (A.48) and (A.49) respectively.

    (IVc.a) For 1p= , we have21

    s8

    1b = , 2m4q = and then

    )(9

    m10

    r3

    m10)(f

    42

    += and

    )(]m4)(12[

    m4)(g

    2

    2

    =

    Therefore

    ]m)(3[)(

    )(

    s8

    m

    )(3

    m

    r

    1

    3

    ma10

    s4

    m)(A

    22

    2221

    2

    ++=

    where )( is the Weirstrassfunction, satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 +=

    with42

    221

    sm800r3a = . This last equation comes from equating the two different

    expressions of and using the value of 1b .

    (IVc.b) For 1p = , we have21 s8

    1b = , 2m4q= and then

    )(9

    m10

    r3

    m10)(f

    42

    = and

    )(]m4)(12[

    m4)(g

    2

    2

    =

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    53/93

    53

    Therefore

    ]m)(3[)(

    )(

    s8

    m

    )(3

    m

    r

    1

    3

    ma10

    s4

    m)(A 22

    222

    12

    +=

    Note. The case 22 = and 0qp > leads to 0m2 < . The choice 0q=

    leads to 0m = .

    Family V. This family corresponds to the Solution V

    20s4

    ma = , 0a1= , 11 bb = , 1

    2bs4p= ,

    12

    2

    bs4

    mq =

    From (A.51) and Solution V, we obtain

    )(gbs4

    m)(A 12

    += (A.66)

    where 12bs4p= ,

    12

    2

    bs4

    mq =

    Sub-family Va. Since 0m4qp 2

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    54/93

    54

    where )( is Weirstrassfunction satisfying the equation

    216

    m)(12

    m)(4))((

    64

    32 +=

    (Vb.a) For 1p = , we have21 s4

    1b = , 2mq= and then

    2m)(12

    )(12)(g

    +

    = .

    Therefore222

    m)(12

    )(

    s

    3

    s4

    m)(A

    +

    =

    Sub-family Vc. If 1p = and25

    r2= then )(g is given by (A.49).

    (Vc.a) For 1p= , we get21

    s4

    1b = , 2mq = and then

    ]m)(12[)(

    )(m)(g

    2

    2

    = .

    Therefore

    ]m)(12[)(

    )(

    s4

    m

    s4

    m)(A

    22

    2

    2

    +=

    (Vc.b) For 1p = , we get2

    1

    s4

    1b = , 2mq= and then

    ]m)(12[)(

    )(m)(g

    2

    2

    = . Therefore

    ]m)(12[)(

    )(

    s4

    m

    s4

    m)(A

    22

    2

    2

    +=

    Note. The case 22 = and 0qp > leads to 0m2 < . The case 0q= leads

    to 0m = .

    Family VI. This family corresponds to the Solution VI

    20s4

    ma = , 0a1= , 11 bb = , 12bs8p= ,

    12

    2

    bs2mq =

    and r is any root of 0r2 =+ .

    From (A.51) and Solution VI, we obtain

    )(gbs4

    m)(A 12

    += (A.67)

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    55/93

    55

    where 12bs8p= ,

    12

    2

    bs4

    mq = and r is any root of 0r2 =+ .

    Sub-family VIa. Since 0m4qp 2

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    56/93

    56

    (VIc.a) For 1p= , we get21 s8

    1b = , 2m4q = and then

    ]m4)(12[)(

    )(m4)(g

    2

    2

    = .Therefore

    ]m)(3[)(

    )(

    s8

    m

    s4

    m)(A

    22

    2

    2

    +=

    (VIc.b) For 1p = , we get21

    s8

    1b = , 2m4q= and then

    ]m4)(12[)(

    )(m4)(g

    2

    2

    = .

    Therefore]m)(3[)(

    )(

    s8

    m

    s4

    m)(A22

    2

    2 += .

    Note. The case 22 = and 0qp > leads to 0m2 < . The case 0q= leads

    to 0m = .

    Family VII. This family corresponds to the Solution VII

    20s4

    ma = , 0a1= , 11 bb = , 1

    2bs8p = ,

    1

    2

    2

    bs2

    mq=

    and r is any root of 0r2 =+ .

    From (A.51) and Solution VII, we obtain

    )(gbs4

    m)(A 12

    += (A.68)

    where 12bs8p = ,

    12

    2

    bs4

    mq= and r is any root of 0r2 =+ .

    Sub-family VIIa. Since 0m4qp 2

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    57/93

    57

    Therefore)(U

    )(U

    s4

    1

    s4

    m)(A

    22

    = where r satisfies the equation

    0r 222 =+ .

    Sub-family VIIb. If 1p = and 2r= (i.e. r any real) then )(g is given by

    (A.46).

    (VIIb.a) For 1p= , we have21

    s8

    1b = , 2m4q = and thus

    2m4)(12

    )(12)(g

    = . Therefore

    222 m)(3

    )(

    s8

    3

    s4

    m)(A

    =

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 +=

    (VIIb.b) For 1p = , we have21 s8

    1b = , 2m4q= and thus

    2m4)(12

    )(12)(g

    +

    = .

    Therefore 222 m)(3

    )(

    s8

    3

    s4

    m

    )(A +

    +=

    Sub-family VIIc. If 1p = and25

    r2= , i.e. 0

    25

    rr

    22 = then )(g is given by

    (A.49).

    (VIIc.a) For 1p= , we get21 s8

    1b = , 2m4q = and then

    ]m4)(12[)()(m4)(g

    2

    2

    = .Therefore

    ]m)(3[)()(

    s8m

    s4m)(A

    22

    2

    2 =

    (VIIc.b) For 1p = , we get21 s8

    1b = , 2m4q= and then

    ]m4)(12[)(

    )(m4)(g

    2

    2

    = . Therefore

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    58/93

    58

    ]m)(3[)(

    )(

    s8

    m

    s4

    m)(A

    22

    2

    2

    = .

    Note. The case 22 = and 0qp > leads to 0m2 < . The case 0q= leads

    to 0m = .

    Family VIII. This family corresponds to the Solution VIII

    0a0 = ,ms8

    pa

    21

    = , 0b1= , pp = ,p

    m4q

    2

    =

    and is any root of 22 r+= .

    From (A.51) and Solution VIII, we obtain

    )(fms8

    p)(A

    2

    = (A.69)

    where pp = ,p

    m4q

    2

    = and is any root of the equation 22 r+= .

    Sub-family VIIIa. Since 0m0qp 2 >< , the choice 22 = , i.e.

    2222 r+= , then )(f is given by (A.36):)(U

    1

    p

    m4)(f

    2

    = where

    )|m|2(cosh)|m|2(sinhr)(U ++= .Therefore)(U

    1

    s2

    m)(A

    2

    =

    where satisfies the equation 2222 r+= .

    Sub-family VIIIb. If 1p = and25

    r2= , leads to 22 r

    25

    24= and then )(f is

    given by (A.48).

    (VIIIb.a) For 1p= , we have 2m4q = and thus)(9

    m10

    r3

    m10)(f

    42

    += .

    where )( is Weirstrassfunction satisfying the equation

    27

    m8)(

    3

    m4)(4))((

    6432 +=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    59/93

    59

    Therefore

    +

    =

    )(3

    m

    r

    1

    s12

    m5)(A

    2

    2, and is any root of the equation

    25

    r24 22 = .

    (VIIIb.b) For 1p = , we have 2m4q= and thus)(9

    m10

    r3

    m10)(f

    42

    = .

    Therefore

    =

    )(3

    m

    r

    1

    s12

    m5)(A

    2

    2.

    Note. The case 22 = and 0qp > leads to 0m2 < . The case 0q= leads

    to 0m = . The case 2r= cannot be considered since in this case 0= and then

    0a1= (i.e. 0)(A = ).

    A.V) Fifth Method. We consider now an expression of the form

    ++

    ++=

    m

    10

    m

    1

    m10

    m1

    edcec

    ebaea)(A (A.70)

    and substitute back into (A.1). We then derive an equation which contains powers

    of the exponentials me . Equating all the coefficients of these exponentials to

    zero, we obtain the following set of solutions

    11 aa = , 0a0 = , 0b1= , 11 cc = , 0c0 = , 11 dd =

    11 aa = , 00 aa = , 11 bb = ,m

    sa2c 11 = ,

    m

    sa2c 00 = ,

    m

    sb2d 11 =

    0a1= , 0a0 = , 11 bb = , 11 cc = , 0c0 = ,m

    sb2d 11 =

    0a1= , 0a0 = , 11 bb = , 11 cc = , 0c0 = ,m

    sb2d 11=

    s2

    cma 11= , 00 aa = , 11 bb = , 11 cc = ,

    m

    sa2c 00 = ,

    m

    sb2d 11=

    We thus have

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    60/93

    60

    +=

    m1

    m1

    m1

    edec

    ea)(A ,

    =m1m

    1

    m1

    em

    bs2ec

    eb)(A

    +

    =m1m

    1

    m1

    em

    bs2ec

    eb)(A ,

    ++

    ++

    =m10m

    1

    m10

    m1

    em

    bs2

    m

    as2ec

    ebaes2

    cm

    )(A

    or

    +

    =m

    2

    m

    1

    m

    eCeC

    e)(A ,

    1

    11

    a

    cC = ,

    1

    12

    a

    dC = (A.71)

    =

    mm1

    m

    es2eCm

    em)(A ,

    1

    11

    b

    cC = (A.72)

    +=

    mm1

    m

    es2eCm

    em)(A ,

    1

    11

    b

    cC = (A.73)

    2m2

    2m1

    m2

    m1

    2

    s4eCs4eCsm2

    sm2eCsm2eCm)(A

    ++

    ++=

    ,0

    11

    a

    cC = ,

    0

    12

    a

    bC = (A.74)

    Appendix B. We consider an expansion of the form

    )(aa)(A 10 += (B.1)

    where )( satisfies Jacobi's differential equation

    44

    33

    2210 nnnnn)(

    d

    d++++=

    (B.2)

    Contrary to the previously considered case (Appendix A, equation (A.3)) we

    consider the right hand side of (B.2) to be a complete fourth degree polynomial.

    Since

    44

    33

    221011 nnnnna)(a)(A ++++==

    and

    )n4n3n2n(a2

    1)(A 34

    23211 +++=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    61/93

    61

    equation (A.1) becomes

    +++

    3

    4

    2

    1

    2

    10

    42

    1

    2

    4

    2

    1 )nas16(aa2as8n2

    1

    a

    +

    ++ 2310

    21

    20

    22

    21

    21

    2 naa2

    3aas48na

    2

    1am2

    +++ )aas32naanaaam4( 130

    22101

    2110

    2

    0naa2

    1na

    2

    3as8am2 1100

    21

    40

    220

    2 =++ (B.3)

    Equating to zero all the coefficients, we find

    0as8n2

    1a 21

    24

    21 =

    0)nas16(aa2 421

    210 =+

    0naa2

    3aas48na

    2

    1am2 310

    21

    20

    22

    21

    21

    2 =+

    0aas32naanaaam4 130

    22101

    2110

    2 =+

    0naa2

    1na

    2

    3as8am2 1100

    21

    40

    220

    2 =+

    From the above system we consider only the following set of nontrivial solutions,

    assuming that 0a1 :

    Solution 1.

    00 aa = , 11 aa = , 31

    03120

    21

    220

    0a

    )anaas48am4(an

    += ,

    21

    03120

    21

    20

    1a

    )an3aas128am8(an

    += ,

    1

    03120

    21

    2

    2a

    an3aas96am4n

    +=

    33 nn = ,21

    24 as16n = .

    Solution 2.

    00 aa = ,2

    220

    2

    031

    nm4as96

    an3a

    += ,

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    62/93

    62

    23

    22

    220

    220

    22

    2

    0n27

    )nm4as96()as48nm8(n

    ++= ,

    3

    222

    022

    02

    22

    1n3

    )nm4as96()as32nm4(n

    ++= ,

    22 nn = , 33 nn = , 22

    220

    2

    23

    20

    2

    4)nm4as96(

    nas144n

    +=

    Solution 3.

    00aa = ,

    1

    20

    22

    20

    1 n

    )as32nm4(aa

    += ,

    22

    220

    2

    21

    20

    22

    2

    0 )nm4as32(3

    n)as48nm8(n

    +=

    11 nn = , 22 nn = ,1

    20

    22

    220

    22

    2

    3n3

    )as96nm4()as32nm4(n

    +++= ,

    21

    22

    220

    220

    2

    4n

    )nm4as32(as16n

    =

    Solution 4.

    00 aa = ,s4

    a1 = , 24

    420242302020

    n

    )nas12nmnas(as64n +=

    4

    20

    22300

    1n

    )as32m2nas3(as16n

    +=

    4

    420

    24

    230

    2n

    )nas24nmnas3(4n

    +=

    where is any root of the equation 0n4

    2 = .

    Solution 5.

    00 aa = ,s4

    a1

    = ,4

    20

    22

    220

    2

    0n3

    )as48nm8(as16n

    +=

    +=

    )as32nm4(as4n

    20

    22

    20

    1 ,0

    20

    22

    2

    3as12

    )as96nm4(n

    ++=

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    63/93

    63

    where is any root of the equation 0n42 = .

    Solution 6.

    00 aa = ,s4

    a1

    = , )as64mas16n(n

    as

    3

    4n 30

    3201

    4

    00 +=

    0

    201

    30

    3

    2as4

    mas16nas128n

    += ,

    20

    2

    412

    030

    3

    3as48

    nnmas32as512n

    +=

    where is any root of the equation 0n42 = .

    Solution 7.

    00 aa = ,s4

    a1

    = ,0

    4020

    2240

    4

    1as4

    nn3ams64as256n

    +=

    20

    2

    4020

    2240

    4

    2as16

    nn3ams128as768n

    += ,

    30

    3

    4020

    2240

    4

    3as64

    )nnams64as768(n

    +=

    where is any root of the equation 0n42 = .

    Solution 8.

    00 aa = , 01 aa = ,2

    120

    22 m4nas32n += ,

    )ans16nm4nm24nans384(n9

    1n 201

    21

    20

    221

    200

    2

    03 ++=

    )as16m4n(n

    as

    3

    16n 20

    221

    0

    20

    2

    4 +=

    where is any root of the equation 0m4as16nn3 2202

    12

    0 =+

    Solution 9.

    00 aa = , 01 aa = ,3

    220

    2

    20

    2211

    0n3m8as128

    )as16m4n(n

    3

    1n

    +=

    21

    20

    22 m4nas32n += , )n3m8as128(

    n

    as16n 3

    220

    2

    1

    20

    2

    4 =

    where is any root of the equation 0n3)m8as128(n 322

    022

    1 =+

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    64/93

    64

    Solution 10.

    00 aa = , 01 aa = ,2

    120

    22 m4nas32n +=

    )ans16nm4nm24nans384(n9

    1n 201

    21

    20

    221

    200

    2

    03 ++=

    )as16m4n(n

    as

    3

    16n 20

    221

    0

    20

    2

    4 +=

    where is any root of the equation 0m4as16nn3 2202

    12

    0 =+ .

    Solution 11.

    00 aa = , 01 aa = ,

    +=

    220

    20

    2

    1

    m4as16n3n

    220

    20

    22 m8as48n3n += ,

    220

    24 as16n =

    where is any root of the equation 0n)m4as48(n 322

    023

    0 =+ .

    B.1. Solutions of the differential equation (B.2).

    For the first three Solutions, we use the following Lemma:

    Lemma: If (B.2) is expressed as

    )Dqp()Dqp()n(r)(d

    d++=

    (B.4)

    then its solution is given by

    Dq4e)pn(4e

    Dqn4e)pnpDq(4en2KrKr2

    2Kr22Kr2

    +++

    ++=

    (B.5)

    where

    Dq)pn(K22 += (B.6)

    The proof of the Lemma is based on the following formula

    +

    =++

    Dq)pn(

    1

    )Dqp()Dqp()n(

    d

    22

    +++++

    + }Dq)p(Dq)pn()pn()n(Dq)pn({

    n

    2ln

    222222

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    65/93

    65

    Integrating equation (B.4) using the above integral, we find (we set the integration

    constant equals to zero)

    =++++++

    }Dq)p(Dq)pn()pn()n(Dq)pn({n

    2 222222

    )Dq)pn(r(exp 22 +=

    Solving the above equation with respect to , we obtain (B.5).

    The solution (B.5) can also be expressed in terms of the hyperbolic tangent

    function as

    )Kr(tanh1)pn(4)Krtanh()Dq41()Dq41(

    )Kr(tanh1)pnpDq(4)Krtanh()Dq41(n)Dq41(n

    222

    22222

    +++

    ++=

    (B.7)

    For the Solution 1, we have

    +=++++

    2

    1

    021

    244

    33

    2210

    a

    aas16nnnnn

    +21

    221

    2

    3210

    21

    221

    2

    3210

    as32

    D

    as32

    nsaa32

    as32

    D

    as32

    nsaa32

    where

    21

    23

    210 )asm16()nsaa64(D

    Equation (B.2) then gives by integration either (B.5) or (B.7) where

    2

    1

    2

    32

    10

    as32

    nsaa32p

    = and

    2

    1

    2

    as32

    1q=

    For the Solution 2, we have

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    66/93

    66

    +

    =++++2

    222

    02

    23

    20

    24

    43

    32

    210)nm4as96(

    nas144nnnnn

    ++

    2

    3

    222

    02

    n3

    nm4as96

    ++

    +

    320

    2

    222

    02

    320

    2

    22

    222

    02

    nas288

    D)nm4as96(

    nas288

    )nm4()nm4as96(

    +

    +

    3

    2

    0

    22

    220

    2

    3

    2

    0

    22

    22

    220

    2

    nas288

    D)nm4as96(

    nas288

    )nm4()nm4as96(

    where

    20

    22

    220

    2 )asm48()nm4as96(D +

    Equation (B.2) then gives by integration either (B.5) or (B.7) where

    320

    2

    22

    222

    02

    nas288

    )nm4()nm4as96(p

    += ,

    320

    2

    222

    02

    nas288

    nm4as96q

    +=

    For the Solution 3, we have

    =++++21

    22

    220

    220

    24

    43

    32

    210n

    )nm4as32(as16nnnnn

    ++

    2

    20

    22

    21

    as32nm4

    n

    +

    )nm4as32(as96

    Dn

    )nm4as32(as96

    )m4n(n

    222

    022

    02

    1

    222

    022

    02

    221

    )nm4as32(as96

    Dn

    )nm4as32(as96

    )m4n(n

    222

    022

    02

    1

    222

    022

    02

    221

    where

    20

    22

    220

    2 )asm48()nm4as96(D +

    Equation (B.2) then gives by integration either (B.5) or (B.7) where

  • 7/22/2019 The Riccati equation method with variable expansion coefficients. I. Solving the Burgers equation

    67/93

    67

    )nm4as32(as96

    )m4n(np

    222

    022

    02

    221

    = ,

    )nm4as32(as96

    nq

    222

    022

    02

    1

    =

    For the rest of the Solutions (Solution 4-Solution 11), we may use Section 11 (pp.

    209-211), of Chapter 7 of H. D. Davis: "Introduction to Nonlinear Differential

    and Integral Equations", Dover 1962. In all these cases we shall also need the

    algorithm of determining the roots of a fourth degree polynomial. The reader may

    consult my paper "Solving Third, Fourth and Fifth Degree Polynomial Equations"

    available from www.docstoc.com. The solutions are then expressed in terms ofJacobi's or Weirstrass functions or even in terms of Elliptic Integrals of various

    kinds.

    Appendix C. In this Appendix we shall solve equation (A.1) using the G/G

    expansion method with variable coefficients. We substitute

    +=

    G

    Gaa)(A 10 (C.2)

    where 0a and 1a are dependent quantities, )(aa 00 = and )(aa 11 = , while

    )(GG = . Since

    +