Symmetry Analysis, Explicit Power Series Solutions and Conservation Laws of The Space-Time Fractional Variant Boussinesq System
Symmetry Analysis, Explicit Power Series Solutions and Conservation Laws of The Space-Time Fractional Variant Boussinesq System
Symmetry Analysis, Explicit Power Series Solutions and Conservation Laws of The Space-Time Fractional Variant Boussinesq System
Abstract. In this study, the classical Lie symmetry method is successfully applied to investigate the sym-
metries of the space-time fractional variant Boussinesq system which was introduced as a model of water
waves. With the help of the obtained symmetries, the governing system is reduced into the system of
nonlinear fractional ordinary differential equations (NLFODEs) which contains Erdèlyi-Kober fractional
differential operators via Riemann-Liouville fractional derivative. The system is also studied for the ex-
plicit power series solution. The obtained power series solution is further examined for the convergence.
The conservation laws of the governing system are constructed by using the new conservation theorem
and generalization of the Noether operators. The numerical approximation for the fractional system is also
found by using the residual power series method (RSPM). Some figures are also presented to explain the
physical understanding for both explicit and approximate solutions.
1 Introduction
The branch of mathematics that deals with derivatives and integrals of arbitrary order real or complex numbers is
called “fractional calculus”. The research area of nonlinear fractional order partial differential equations (NLFPDEs)
has been remarkably active for the past few decades. The physical phenomena in various engineering and scientific
fields, specifically statistical mechanics, fluid flow, viscoelasticity, material science, biosciences, medicine, geology,
electrochemistry, electromagnetics described by differential models is examined for the fractional order. There are
many methods to find solutions of NLFPDEs such as the Homotopy method [1], G /G the expansion method [2],
the sub-equation method [3], the exp function method [4] and the Lie symmetry method [5, 6]. Exact solutions of
fractional differential equations (FDEs) have attracted substantial interest in the literature. A lot of information of
physical phenomena modeled by differential equations has been supplied by Lie symmetry analysis and conservation
laws [7]. The Lie symmetry method is systematically used in different nonlinear scientific fields and plays an important
role for finding exact solutions of nonlinear partial differential equations (NLPDEs). The theory of the Lie group for
obtaining the group invariant solutions of NLPDEs is one of the most powerful methods. We found lot of books and
survey articles about applications of the Lie group [5, 8–12]. The investigation process of conservation laws has been
given in some articles [13–15]. In order to find conservation laws the fractional order Noether operators have been
generalized by using a new conservation theorem [16].
Nowadays researchers have shown much interest in space-time NLFPDEs in place of NLPDEs of integer order
for a better understanding of nonlinear phenomena. There have been many approaches for the solution of fractional
partial differential equations (FPDEs) [14,17–19]. In this paper, we study the system of a space-time fractional variant
Boussinesq system for Lie symmetry reduction, explicit power series solution and conservation laws.
The Boussinesq equation was originally introduced to describe the continuation of long waves in the shallow
water [20–22]. In this work, we consider the space-time fractional variant Boussinesq system,
∂γ H ∂β H ∂β U
γ
+H β
+ = 0,
∂t ∂x ∂xβ
∂γ U ∂β U ∂β H ∂3H
+ H + U + = 0, (1)
∂tγ ∂xβ ∂xβ ∂x3
a
e-mail: sachin1jan@yahoo.com (corresponding author)
Page 2 of 19 Eur. Phys. J. Plus (2018) 133: 520
is a mathematical model which describes the propagation of long surface waves, where H is the velocity and U is the
height of the free wave surface for fluid in the trough. The Boussinesq equations have been studied by the various
authors by many different approaches. In [22], the authors studied the Boussinesq equation for soliton solutions.
In [23], the traveling wave solution of the variant Boussinesq is found. In [20, 24] the variant Boussinesq system has
been studied for Lie symmetry reductions, exact solutions and conservation laws.
The outline of the paper is as follows: In sect. 2, the Lie symmetry method is introduced to deal with a system of
space-time FPDEs. The Lie symmetry method is employed to obtain symmetry reduction of the considered system of
space-time fractional equations in sect. 3. Section 4 contains explicit power series solutions of governing system (1). In
sect. 5, the explicit power series is analysed for convergence. Conservation laws are found for the governing system in
sect. 6. The residual power series method (RPSM) is applied to find a residual approximate solution in sect. 7. Section 8
contains the numerical result and discussion of the approximate solution by interesting graphs of the approximate
solution. Finally, in sect. 9, concluding remarks are presented.
∂ γ v1∗ ∂ γ v1 ∂ γ v2∗ ∂ γ v2
= + εη γ,t + O(ε2 ), = + εφγ,t + O(ε2 ),
∂tγ ∂tγ ∂tγ ∂tγ
∂ β v1∗ ∂ β v1 ∂ β v2∗ ∂ β v2
= + εη β,t + O(ε2 ), = + εφβ,t + O(ε2 ),
∂xβ ∂xβ ∂xβ ∂xβ
∂v1∗ ∂v1 ∂v2∗ ∂v2
= + εη x + O(ε2 ), = + εφx + O(ε2 ),
∂x ∂x ∂x ∂x
∂ 2 v1∗ ∂ 2 v1 ∂ 2 v2∗ ∂ 2 v2
2
= + εη xx + O(ε2 ), = + εφxx + O(ε2 ),
∂x ∂x2 ∂x 2 ∂x2
..
., (4)
where ξ, τ , η, φ are the infinitesimals; η γ,t and φγ,t are the extended infinitesimals of order γ; η β,x and φβ,x are the
extended infinitesimals of order β; η x , φx , η xx and φxx are the integer order extended infinitesimals, which leave the
system of eqs. (2) invariant. The infinitesimal transformation (4) are such that if v1 , v2 is the solution of (2) then so
is v1∗ , v2∗ . The associated infinitesimal generator is given by
∂ ∂ ∂ ∂
X = ξ(x, t, v1 , v2 ) + τ (x, t, v1 , v2 ) + η(x, t, v1 , v2 ) + φ(x, t, v1 , v2 ) . (5)
∂x ∂t ∂v1 ∂v2
Eur. Phys. J. Plus (2018) 133: 520 Page 3 of 19
pr(γ,β) X = X + η γ,t ∂∂tγ v1 + φγ,t ∂∂tγ v2 + η β,x ∂∂xβ v1 + φβ,x ∂∂xβ v2 + η x ∂v1,x + φx ∂v2,x + η xx ∂v1,xx + η xxx ∂v1,xxx + . . . , (6)
γ 2
where ∂xγ1 vi = ∂∂xvγi , ∂vi ,x1 = ∂x
∂vi
1
, ∂vi ,x1 x1 = ∂∂xv2i and so on, for i = 1, 2 and x1 = x, t.
1 1
Since the lower limit of the Riemann-Liouville fractional derivative (3) with respect to x, t is fixed, therefore the
invariance condition yields
ξ(x, t, v1 , v2 )|x=0 = 0, τ (x, t, v1 , v2 )|t=0 = 0. (7)
The infinitesimals are given as
dx∗ dt∗
ξ(x, t, v1 , v2 ) = , τ (x, t, v1 , v2 ) = ,
d =0 d =0
dv ∗ dv ∗
η(x, t, v1 , v2 ) = 1 , φ(x, t, v1 , v2 ) = 2 , (8)
d =0 d =0
also,
η x = Dx η − v1,x Dx ξ − v1,t Dx τ − v2,x Dx ξ − v2,t Dx τ,
η xx = Dx η x − v1,xx Dx ξ − v1,xt Dx τ − v2,xx Dx ξ − v2,xt Dx τ. (9)
In a similar way we can obtain η xxx , φx , φxx and so on, where Dx is the total derivative given by
d dv1 dv2 dv1,x dv2,x
Dx = + v1,x + v2,x + v1,xx + v2,xx + .... (10)
dx dx dx dx dx
The γ-th–order extended infinitesimal η γ,t related to the Riemann-Liouville fractional derivative is
η γ,t = Dtγ (η) + ξDtγ (v1,x ) − Dtγ (ξv1,x ) + τ Dtγ (v1,t ) − Dtγ (τ v1,t ). (11)
Now by applying the generalized Leibnitz rule [25, 26], eq. (11) can be expanded as
∞ ∞
γ ∂ γ v1
γ γ−n
γ
η = Dt (η) − γDt τ
γ,t
− Dt (ξ)Dt (v1 , x) −
n
Dtn+1 (τ )Dtγ−n (v1 ), (12)
∂tγ n=1
n n=1
n + 1
where Dt represents the total derivative operator. Now by using the generalized chain rule and the generalized Leibnitz
rule Dtγ (η) can be obtained as
∞ n
∂γ η ∂ γ v1 ∂ γ ηv1 ∂ γ v2 ∂ γ ηv2 γ ∂ ηv2 γ−n
Dtγ (η) = γ + ηv1 γ − v1 + η v2 − v 2 + μ η,γ,1 + μη,γ,2 + D (v2 )
∂t ∂t ∂tγ ∂tγ ∂tγ n=1
n ∂tn t
∞ n
γ ∂ ηv1 γ−n
+ D (v1 ), (13)
n=1
n ∂tn t
where
∞
n
γ n k 1
m k−1
tn−γ ∂m ∂ n−m+k η
μη,γ,i = (−vi )r m (vik−r ) n−m k , i = 1, 2 (14)
n=2 m=2 k=2 r=2
n m r k! Γ (n − γ + 1) ∂t ∂t ∂vi
∂η
and ηvi = ∂v i
.
After substituting the value of Dtγ (η) in (12), the γ-th–order extended infinitesimal η γ,t for the system of FPDEs
of the form (2) is
∞
∂γ η ∂ γ v1 ∂ γ ηv1 ∂ γ v2 ∂ γ ηv2 γ
η γ,t = γ + (ηv1 − γDt (τ )) γ
− v 1 γ
+ η v 2 γ
− v 2 γ
− Dtn (ξ)Dtγ−n (v1 , x)
∂t ∂t ∂t ∂t ∂t n=1
n
∞ n
∞ n
γ ∂ ηv2 γ−n γ ∂ ηv1 γ
+ n
D t (v 2 ) + μη,γ,1 + μη,γ,2 + n
− D n+1
t (τ ) Dtγ−n (v1 ). (15)
n=1
n ∂t n=1
n ∂t n + 1
Similar to this expression of η γ,t , the γ-th extended infinitesimal φγ,t , β-th extended infinitesimal η β,x and β-th
extended infinitesimal φβ,x are obtained easily.
Page 4 of 19 Eur. Phys. J. Plus (2018) 133: 520
(3 + β) (β − 3)
ξ = c1 x, τ= c1 t, η= c1 H, φ = (β − 3)c1 U, (16)
2γ 2
where c1 is an arbitrary constant and the Lie symmetry generator of the system (1) is obtained as
∂ (3 + β) ∂ (β − 3)H ∂ ∂
X=x + t + + (β − 3)U . (17)
∂x 2γ ∂t 2 ∂H ∂U
Proof. The invariance criteria of system (1) under the one parameter Lie group of transformations are obtained as
where η xxx is the extended infinitesimal of order 3, η γ,t , φγ,t are extended infinitesimals of order γ, and η β,x , φβ,x are
extended infinitesimals of order β.
Now substituting the values of the prolongations and equating the coefficient of various powers of partial derivatives
of H and U to zero, for 0 < γ, β ≤ 1 the determining equations are obtained as
ξt = ξH = ξU = 0,
τx = τH = τU = 0,
ηU = 0,
ηH − 3ξx − φU + γDt τ = 0,
η + H(γDt τ − βDx ξ) = 0,
ηH − φU − γDt γ + βDx ξ = 0,
∂tγ η − H∂tγ ηH = 0,
∂tγ φ − U ∂tγ φU + ηxxx = 0,
γ n γ
∂ ηH − Dtn+1 τ = 0, n ∈ N,
n t n+1
γ n γ
∂ φU − Dtn+1 τ = 0, n ∈ N,
n t n+1
β n β
∂ ηH − Dxn+1 ξ = 0, n ∈ N,
n x n+1
β n β
∂ φU − Dxn+1 ξ = 0, n ∈ N. (19)
n x n+1
Solving all PDEs and FPDEs together in the system (19), also using (7), we obtained the infinitesimals
(3 + β) (β − 3)
ξ = c1 x, τ= c1 t, η= c1 H, φ = (β − 3)c1 U, (20)
2γ 2
where c1 is the arbitrary constant.
From these infinitesimals (20), the symmetry generator (17) of system (1 is obtained. Thus, the auxiliary equations
are
dx dt dH dU
= (3+β) = (β−3) = . (21)
x t H (β − 3)U
2γ 2
The symmetry reduction of the system (1) is obtained in the following theorem.
Eur. Phys. J. Plus (2018) 133: 520 Page 5 of 19
2γ γ(β−3)
Theorem 2. The similarity transformations of the system with similarity variable y = xt− β+3 are H(x, t) = t β+3 f (y)
2γ(β−3)
and U (x, t) = t β+3h(y) and the corresponding symmetry reduction of the system of NLFPDEs (1) for γ, β > 0 into
the system of NLFODEs is given by
6γ
1− β+3 ,γ
P β+3 f (y) + y −β Q−β,β
1 h (y) + f (y) Q−β,β
1 f (y) = 0,
γ
γ(β−9)
1+ β+3 ,γ
P β+3 h (y) + y −β h(y) Q−β,β
1 f (y) + f (y) Q−β,β
1 h (y) + f (y) = 0, (22)
γ
r−1
1 d
Pϑ,γ f (y) = ϑ+k− y Mϑ+γ,r−γ
f (y), y > 0,
> 0, γ > 0,
dy
k=0
[γ] + 1 if γ ∈
/ N,
r= (23)
γ if γ ∈ N,
where ⎧ ∞
1
⎪
⎨
1
(ρ − 1)γ−1 ρ−(ϑ+α) f yρ dρ
ϑ,γ if γ > 0,
M f (y) = Γ (γ) 1
⎪
⎩
f (y) if γ = 0
is the left-hand–sided Erdèlyi-Kober fractional integral operator.
Also, (Qϑ,β
) is the right-hand–sided Erdèlyi-Kober fractional differential operator given as
r
1 d
Qϑ,β
f (y) = ϑ+k+ y Tϑ+β,r−β f (y), y > 0,
> 0, β > 0,
dy
k=1
[β] + 1 if β ∈
/ N,
r= (24)
β if β ∈ N,
where ⎧ 1
1
⎪
⎨ 1
ϑ,β (1 − ρ)β−1 ρϑ f yρ dρ if β > 0,
T f (y) = Γ (β) 0
⎪
⎩
f (y) if β = 0
is the right-hand–sided Erdèlyi-Kober fractional integral operator.
β β
Proof. First, we find the value of the fractional partial derivatives ∂∂xH ∂ U
β and ∂xβ of order β > 0.
For n − 1 < β < n; n ∈ N, then by the definition of the Riemann-Liouville fractional derivative (3), we have
x
2γ
∂β H ∂n 1 γ(β−3)
= (x − s)n−β−1 β+3
t f t− β+3 s ds. (25)
∂xβ ∂xn Γ (n − β) 0
2γ
Let ω = xs , and using similarity variable y = xt− β+3 , we get following equation:
γ(β−3) 1
∂β H ∂ n t β+3
= (1 − ω)n−β−1 xn−β f (yω)dω
∂xβ ∂xn Γ (n − β) 0
γ(β−3) ∂ n
= t β+3 n
xn−β T10,n−β f (y), (26)
∂x
where
1 1
T10,n−β f (y) = (1 − ω)n−β−1 f (yω)dω (27)
Γ (n − β) 0
2γ
Now for greater simplification, let ρ(y) be a continuously differential function for y = xt− β+3 , then we have the
following relation:
∂ d 2γ d
x ρ(y) = x ρ(y)t− β+3 = y ρ(y). (28)
∂x dy dy
Thus, it follows:
∂ n
n−β
0,n−β ∂ n−1 d
0,n−β
x T1 f (y) = xn−β−1
n−β+y T1 f (y) ,
∂xn ∂xn−1 dy
n
d
0,n−β
= . . . = x−β j−β+y T1 f (y). (29)
j=1
dy
∂ n
n−β
0,n−β
x T1 f (y) = x−β (Q−β,β
1 f )(y). (30)
∂xn
Therefore, by using (30) in (26), we obtain
∂β H γ(β−3)
β+3 x−β
−β,β
= t Q1 f (y). (31)
∂xβ
Similarly, we may have
∂β U 2γ(β−3)
−β −β,β
= t β+3 x Q1 h (y). (32)
∂xβ
γ γ
Also, in a similar way, the fractional partial derivatives ∂∂tH ∂ U
γ and ∂tγ are obtained of order γ > 0 as follows:
∂γ H −6γ 6γ
1− β+3 ,γ
=t β+3 P β+3 f (y),
∂tγ γ
∂γ U γ(β−9) γ(β−9)
1+ β+3 ,γ
=t β+3 P β+3 h (y), (33)
∂tγ γ
where (Pϑ,γ ) is the left-hand–sided Erdèlyi-Kober fractional differential operator given by (23).
Hence, by using (31), (32) and (33), the space-time fractional variant Boussinesq system (1), is reduced to the
NLFODE for 0 < γ, β < 1 written as
6γ
1− β+3 ,γ
P β+3 f (y) + y −β Q−β,β
1 h (y) + f (y) Q−β,β
1 f (y) = 0,
γ
γ(β−9)
1+ ,γ
P β+3 β+3 h (y) + y −β h(y) Q−β,β 1 f (y) + f (y) Q−β,β
1 h (y) + f (y) = 0. (34)
γ
⎛
⎞ ∞
Γ 1 + γ(β−3) n
β+3 − β+3
∞ nγ ∞
Γ (1 + n)
Γ (1 + k)
⎝
⎠ an y + y
n −β n
bn y + ak an−k y n
= 0,
n=0 Γ 1 − β+3
6γ
− β+3
nγ
n=0
Γ (1 − β + n) n=0 k=0
Γ (1 − β + k)
⎛
⎞ ∞ n
∞ Γ 1 + 2γ(β−3) − nγ
β+3 β+3 Γ (1 + n − k) n
⎝
⎠
bn y + y
n −β
bk an−k + ak bn−k y
Γ 1 + γ(β−9) Γ (1 − β + n − k)
β+3 − β+3
nγ
n=0 n=0 k=0
∞
⎧⎛
⎞ ⎫
⎨ Γ 1 + 2γ(β−3) ⎬
1 ⎝ β+3 1
a3 = −
⎠ b0 + 2y −β a 0 b0 ,
6 ⎩ Γ 1 + γ(β−9) Γ (1 − β) ⎭
β+3
⎧⎛
⎞ ⎫
⎨ Γ (1 − β)Γ 1 + γ(β−3)β+3
⎬
b0 = − ⎝
⎠ a0 y β + a20 . (38)
⎩ Γ 1 − β+36γ ⎭
Now comparing coefficient for n ≥ 1, in (37), we have the following recursion relation:
⎧⎛
⎞
1 ⎨ Γ 1 + 2γ(β−3)
β+3 − nγ
β+3
an+3 = − ⎝
⎠ bn
(n + 1)(n + 2)(n + 3) ⎩ γ(β−9)
Γ 1 + β+3 − β+3 nγ
n ⎫
Γ (1 + n − k) ⎬
+ y −β (an−k bk + ak bn−k ) ,
Γ (1 + n − k − β) ⎭
k=0
⎧⎛
⎞ ⎫
n
Γ (1 + n − β) ⎨⎝ Γ 1 + β+3 − β+3 ⎠ ⎬
γ(β−3) nγ
Γ (1 + k)
bn = −
an y β + ak an−k . (39)
Γ (1 + n) ⎩ Γ 1 − β+36γ
− β+3
nγ Γ (1 + k − β) ⎭
k=0
In view of (39), all coefficients an (n ≥ 3) and bn (n ≥ 0) of the power series (35) can be found by choosing arbitrary
constants ai (i = 0, 1, 2).
Hence, (1) have the exact power series solution having coefficients depending on (39). Thus, the power series
solution for (35) is described as
⎛⎛
⎞ ⎞
Γ 1 + 2γ(β−3)
∞
1 β+3 1 1
f (y) = a0 + a1 y + a2 y 2 − ⎝⎝
⎠ b0 + 2y −β a 0 b0 ⎠ y 3 + −
6 Γ 1 + β+3 γ(β−9) Γ (1 − β) n=1
(n + 1)(n + 2)(n + 3)
⎧⎛
⎞ n ⎫
⎨ Γ 1 + 2γ(β−3) − β+3
nγ
⎬
β+3 Γ (1 + n − k)
× ⎝
⎠ bn + y −β (an−k bk + ak bn−k ) y n+3 ,
⎩ Γ 1 + γ(β−9) − nγ Γ (1 + n − k − β) ⎭
β+3 β+3 k=0
⎧⎛
⎞ ⎫
∞
⎨ Γ 1 + γ(β−3)
− nγ
n ⎬
Γ (1 + n − β) ⎝ β+3 β+3 Γ (1 + k)
h(y) = b0 + −
⎠ an y β + ak an−k y n . (40)
n=1
Γ (1 + n) ⎩ Γ 1 − 6γ − nγ Γ (1 + k − β) ⎭
β+3 β+3 k=0
Page 8 of 19 Eur. Phys. J. Plus (2018) 133: 520
Fig. 1. Panels (a) and (b) show the behavior of the exact solution of H(x, t) and U (x, t), respectively, in (41) w.r.t x and t,
when γ = .25, β = .75, a0 = a1 = a2 = 1, n = 45.
Fig. 2. Panels (a) and (b) show the behavior of the exact solution of H(x, t) and U (x, t), respectively, in (41) w.r.t x and t,
when γ = .25, β = .75, a0 = a1 = a2 = 1, n = 115.
From figs. 1–4, one can see that the terms of the power series solution are decreasing as the value of n increases.
Eur. Phys. J. Plus (2018) 133: 520 Page 9 of 19
Fig. 3. Panels (a) and (b) show the behavior of the exact solution of H(x, t) and U (x, t), respectively, in (41) w.r.t x and t,
when γ = .25, β = .75, a0 = a1 = a2 = 1, n = 445.
Fig. 4. Panels (a) and (b) show the behavior of the exact solution of H(x, t) and U (x, t), respectively, in (41) w.r.t x and t,
when γ = .25, β = .75, a0 = a1 = a2 = 1, n = 1000.
5 Convergence analysis
The present section contains the analysis of convergence [24, 29] of the power series solution (41), obtained in the
previous section. Now from (39), we have
n
|an+3 | ≤ M |bn | + |an−k ||bk | + |ak ||bn−k | , (42)
k=0 k=0
Γ (1+ 2γ(β−3) −
nγ
β+3 )
where M = max β+3
γ(β−9) nγ , y −β Γ (1+n−k)
Γ (1+n−k−β) , for all k = 0, 1, 2, . . . , n,
Γ (1+ β+3 − β+3 )
n
|bn | ≤ N |an | + |ak ||an−k | , (43)
k=0
Γ (1+ γ(β−3) − nγ
β+3 )
where N = max β+3
6γ
Γ (1− β+3 nγ
− β+3 )
y β , Γ Γ(1+k−β)
(1+k)
, for all k = 0, 1, 2, . . . , n.
We define two power series:
∞
∞
such that
pi = |ai |, qj = |bj |, for i = 0, 1, 2, 3, and j = 0,
Page 10 of 19 Eur. Phys. J. Plus (2018) 133: 520
and
n
pn+3 = M qn + pn−k qk + pk qn−k ,
k=0 k=0
k
qn = N pn + pk pn−k , (45)
k=0
P = P (y) = pn y n , Q = Q(y) = qn y n
n=0 n=0
is the majorant series of the series f (y) and g(y) in (35), respectively. Further, it is demonstrated that the series
P = P (y) and Q = Q(y) are convergent:
∞
P = p0 + p1 y + p2 y 2 + p3 y 3 + pn+3 y n+3
n=1
∞
n
= p0 + p1 y + p2 y 2 + p3 y 3 + M qn + pn−k qk + pk qn−k y n+3
n=1 k=0 k=0
= p0 + p1 y + p2 y 2 + p3 y 3 + M [y 3 (Q − q0 ) + y 3 (P Q − p0 q0 ) + y 3 (P Q − p0 q0 )],
and
∞ ∞
n
n
Q = q0 + qn y = q0 + N pn + pk pn−k
n=1 n=1 k=0
= q0 + N (P − p0 ) + (P − p20 ) . 2
Consider the implicit functional system with respect to the independent variable y,
F (y, P, Q) = P − p0 − p1 y − p2 y 2 − p3 y 3 − M [y 3 (Q − q0 ) + 2y 3 (P Q − p0 q0 )],
G(y, P, Q) = Q − q0 − N [(P − p0 ) + (P 2 − p20 )].
Since F , G are analytic in the neighbourhood of (0, p0 , q0 ) and F (0, p0 , q0 ) = 0, G(0, p0 , q0 ) = 0.
Furthermore, the Jacobian determinant
∂(F, G)
= 1 = 0.
∂(P, Q) (0,p0 ,q0 )
Therefore, using the implicit function theorem [30], the series P = P (y) and Q = Q(y) are analytic in a neighbourhood
of the point (0, p0 , q0 ). This implies that the two power series (35) are convergent and converge in a neighbourhood of
the point (0, p0 , q0 ).
∞
δ ∂ ∂ ∂ ∂
j
= j
+ (Dtγ )∗ β ∗
γ j + (Dx ) β
+ (−1)k Di1 Di2 , . . . , Dik j
, (50)
δH ∂H ∂(Dt H ) j
∂(Dx H ) ∂(H )i1 ,i2 ,...,ik
k=1
where Dik is the total derivative operator with respect to the ik -th variable. Also (Dtγ )∗ and (Dxβ )∗ are the adjoint
operators [25, 32] of the Riemann-Liouville fractional derivative operators Dtγ and Dxβ , respectively.
From (48) and (49), the adjoint equations are obtained as
δL
= F1∗ = (Dtγ )∗ u + (Dxβ )∗ (uH) + (Dxβ )∗ (vU ) + u∂xβ H + v∂xβ U − vxxx = 0,
δH
δL
= F2∗ = (Dtγ )∗ v + (Dxβ )∗ u + (Dxβ )∗ (vH) + u∂xβ H = 0. (51)
δU
The system (1) is nonlinear self-adjoint if by substituting the value of new dependent variables
eq. (51) must satisfy, with at least one of the introduced new dependent variables as non zero. Now, the derivatives
of v = ψ(x, t, H, U ) with respect to x are
vx = ψx + ψH Hx + ψU Ux ,
vxx = ψxx + 2ψxH Hx + 2ψxU Ux + ψHH Hx2 + 2ψHU Hx Ux + ψU U Ux2 + ψH Hxx + ψU Uxx ,
vxxx = ψxxx + 6ψxHU Hx Ux + 3ψHHU Hx2 Ux + 3ψHH Hx Uxx + 3ψHU U Hx Ux2 + 3ψxH Hxx
+ 3ψHU (Hx Uxx + Ux Hxx ) + 3ψU U Ux Uxx + 3ψxxU Ux + 3ψxxH Hx + 3ψxU Uxx
+ ψH Hxxx + ψU Uxxx + 3ψxHH Hx2 + 3ψxU U Ux2 + ψHHH Hx3 + ψU U U Ux3 . (53)
(Dtγ )∗ ϕ + (Dxβ )∗ (ϕH) + (Dxβ )∗ (ψU ) + ϕ∂xβ H + ψ∂xβ U − ψxxx − 6ψxHU Hx Ux − 3ψHH Hx Uxx
− 3ψxxU Ux − 3ψxH Hxx − 3ψHHU Hx2 Ux − 3ψHU U Hx Ux2 − 3ψxxH Hx − 3ψHU (Hx Uxx − Ux Hxx )
− 3ψU U Ux Uxx − 3ψxU Uxx − ψH Hxxx − ψU Uxxx − 3ψxHH Hx2 − 3ψxU U Ux2 − ψHHH Hx3 − ψU U U Ux3 =
γ γ
∂ H ∂β H ∂β U ∂ U ∂β U ∂β H ∂3H
λ1 +H + + λ2 +H β +U + ,
∂tγ ∂xβ ∂xβ ∂tγ ∂x ∂xβ ∂x3
(Dtγ )∗ ψ + (Dxβ )∗ ϕ + (Dxβ )∗ (ψH) + ϕ∂xβ H =
γ γ
∂ H ∂β H ∂β U ∂ U ∂β U ∂β H ∂3H
λ3 + H + + λ 4 + H + U + . (55)
∂tγ ∂xβ ∂xβ ∂tγ ∂xβ ∂xβ ∂x3
Equating the coefficients of various powers of H and U and their partial derivatives in both sides of (55), then solving
them simultaneously, we get λi = 0, i = 1, 2, 3, 4. and
ϕ = ϕ(x, t, H, U ),
ψ = x2 A1 (t) + xA2 (t) + A3 (t), (56)
With the help of the new conservation theorem, the Noether operators to construct conservation laws of systems
FPDEs are given as follows.
The Lie characteristic functions W 1 and W 2 corresponding to symmetry generators are defined by
2
∂L ∂L
t
C = (−1) k
Dtγ−1−k (W j )Dtk − (−1) J1 W
m j
, Dtm , (58)
j=1
∂(Dtγ Hj ) ∂(Dtγ Hj )
k=0
where m = [γ] + 1, and W j , (j = 1, 2) are given by (57) and Hj , (j = 1, 2) are dependent variables, i.e., H1 = H and
H2 = U , also J1 (h1 , h2 ) is the integral defined as
t q
1 h1 (x, s)h2 (x, r)
J1 (h1 , h2 ) = dr ds
Γ (m − γ) 0 t (r − s)γ+1−m
2
∂L ∂L
x
C = k
(−1) Dxβ−1−k (W j )Dxk − (−1) J2n
W j
, Dxn , (59)
j=1 k=0 ∂(Dxβ Hj ) ∂(Dxβ Hj )
where n = [β] + 1, and W j , (j = 1, 2) given by (57) and Hj , (j = 1, 2) are dependent variables, also J2 (h1 , h2 ) is the
integral defined as
x p
1 h1 (s, t)h2 (r, t)
J2 (h1 , h2 ) = dr ds,
Γ (n − β) 0 x (r − s)β+1−n
for any two functions h1 (x, t) and h2 (x, t).
By using (57) and symmetry generator (17), the Lie characteristic functions are
(β − 3) (3 + β)
W1 = H − xHx − tHt ,
2 2γ
(3 + β)
W 2 = (β − 3)U − xUx − tUt . (60)
2γ
Now, usingthe basic definitions presented above and the adjoint variables (56), the conserved vectors of system (1),
are obtained as follows.
Considering the following cases:
Case 1. For 0 < γ < 1, the t-component of the conserved vector is obtained as
C t = I 1−γ (W 1 )ϕ + I 1−γ (W 2 )ψ + J1 (W 1 , Dt ϕ) + J1 (W 2 , ψt ).
Case 3. For 0 < β < 1, we have the value of ψ = a (constant), and the x- component of the conserved vector is as
Case 4. For 1 < β < 2, we have the value of ψ = a1 x + a2 , where a1 , a2 are arbitrary constants and the x-component
of the conserved vector is obtained as
C x = Dxβ−1 (W 1 )(Hϕ + (a1 x + a2 )U ) + Dxβ−1 (W 2 )(ϕ + (a1 x + a2 )H) − Ix2−β (W 1 )Dx (Hϕ + (a1 x + a2 )U )
− Ix2−β (W 2 )Dx (ϕ + (a1 x + a2 )H) − J2 (W 1 , Dx2 (Hϕ + (a1 x + a2 )U )) − J2 (W 2 , Dx2 (ϕ + (a1 x + a2 )H)).
Eur. Phys. J. Plus (2018) 133: 520 Page 13 of 19
H(x, 0) = h(x),
U (x, 0) = φ(x). (61)
Suppose that the solution of (1) is expressed in the form of the fractional power series expansion about initial point
t = 0 as given below
∞
tkγ
H(x, t) = hk (x) , (62)
Γ (kγ + 1)
k=0
∞
tk γ
U (x, t) = φK (x) , x ∈ I, t > 0. (63)
Γ (kγ + 1)
k=0
The analytic approximate solution for the system (1) with initial condition (61) is in the form of an infinite fractional
power series given by RPSM.
In order to obtain numerical values from these series, Hm (x, t) and Um (x, t) denote the m-th truncated series of
H(x, t) and U (x, t) respectively. That is,
m
tkγ
Hm (x, t) = hk (x) , (64)
Γ (kγ + 1)
k=0
m
tk γ
Um (x, t) = φK (x) , x ∈ I, t > 0. (65)
Γ (kγ + 1)
k=0
For m = 0 by using the initial condition, the 0-th residual power series approximate solutions of H(x, t) and U (x, t)
are written in the following form:
m
tkγ
Hm (x, t) = h(x) + hk (x) , (68)
Γ (kγ + 1)
k=1
m
tk γ
Um (x, t) = φ(x) + φK (x) , x ∈ I, t > 0. (69)
Γ (kγ + 1)
k=1
In this way, the m-th residual power series approximate solution Hm (x, t) and Um (x, t) can be obtained if hk (x) and
φk (x) are known for k = 1, 2, 3, . . . , m.
Now let us define the residual function for (1) and (61) as follows:
Here we state some important results of ResH,m (x, t) and ResU,m (x, t), which are essential in the residual power
solution:
1)
ResH (x, t) = 0, ResU (x, t) = 0; (72)
Page 14 of 19 Eur. Phys. J. Plus (2018) 133: 520
2)
3)
Substituting the m-th truncated series (68) and (69) of H(x, t) and U (x, t), respectively, into (71) and calculating the
(m−1)γ
fractional derivative Dt of RH,m (x, t) and RU,m (x, t), for m = 1, 2, 3, . . .. at t = 0 together with (74), we obtain
the following algebraic system
(m−1)γ
Dt ResH,m (x, 0) = 0,
(m−1)γ
Dt ResU,m (x, 0) = 0, 0 < γ ≤ 1, m = 1, 2, 3, . . . . (75)
The values of hm (x) and φm (x) are obtained by solving system (75). Therefore, the m-th residual power series
approximate solution is derived.
In the below discussion, we determined in detail the 1st and 2nd residual power series approximate solution.
For m = 1, the first RPS solution can be written in the form of
tγ
H1 (x, t) = h(x) + h1 (x) , (76)
Γ (γ + 1)
tγ
U1 (x, t) = φ(x) + φ1 (x) . (77)
Γ (γ + 1)
The 1st residual function can be written as follows:
For m = 2, the 2nd residual power series solution can be obtained as follows:
tγ t2γ
H2 (x, t) = h(x) + h1 (x) + h2 (x) , (85)
Γ (γ + 1) Γ (2γ + 1)
tγ t2γ
U2 (x, t) = φ(x) + φ1 (x) + φ2 (x) . (86)
Γ (γ + 1) Γ (2γ + 1)
φ2 (x) = − h(x)∂xβ φ1 (x) + φ1 (x)∂xβ h(x) + h1 (x)∂xβ φ(x) + φ(x)∂xβ h1 (x) + ∂x3 h1 (x) . (91)
tγ t2γ
H2 (x, t) = h(x) − h(x)∂xβ h(x) + ∂xβ φ(x) − h(x)∂xβ h1 (x) + h1 (x)∂xβ h(x) + ∂xβ φ1 (x) , (92)
Γ (γ + 1) Γ (2γ + 1)
tγ
U2 (x, t) = φ(x) − φ(x)∂xβ h(x) + h(x)∂xβ φ(x) + ∂x3 φ(x) − h(x)∂xβ φ1 (x) + φ1 (x)∂xβ h(x)
Γ (γ + 1)
t2γ
+ h1 (x)∂xβ φ(x) + φ(x)∂xβ h1 (x) + ∂x3 h1 (x) . (93)
Γ (2γ + 1)
In the same way, we can find the remaining approximate solution of order 3, 4 and so on of system (1). Hence, we have
m
tkγ
Hm (x, t) = hk (x) ,
Γ (kγ + 1)
k=0
m
tk γ
Um (x, t) = φK (x) , x ∈ I, t > 0. (94)
Γ (kγ + 1)
k=0
Page 16 of 19 Eur. Phys. J. Plus (2018) 133: 520
Fig. 5. Behavior of the first RPS approximate solution (83) for different values: (a) γ = .1 and β = .3; (b) γ = .4 and β = .6;
(c) γ = .9 and β = .9.
By using the concept of the generalized Taylor series, the RPSM is a technique to find the analytic approximate
solution of physical phenomena in the form of convergent series. By utilizing the terms of RPS approximations, higher
terms will reduce the error of approximation. Here we present some graphical explanations that are reported according
to the obtained first and second approximation with respect to h(x) and φ(X) chosen as below:
2(sin(x) + cos(x))
h(x) = 1 − ,
sin(x) − cos(x)
2
sin(x) + cos(x)
φ(x) = −2 − 2 . (95)
sin(x) − cos(x)
This section contains graphical explanations of the numerical simulations of RPS of system (1). Figures 5 and 6 show
the behavior of the first RPS approximate solution of (83) and (84), respectively, for different values of γ = .1 and
β = .3, γ = .4 and β = .6 and γ = .9 and β = .9 and β = .25, respectively. Similarly, figs. 7 and 8 show the behavior
of the first RPS approximate solution of (92) and (93), respectively, for different values of γ = .1 and β = .3, γ = .4
and β = .6 and γ = .9 and β = .9, respectively. Here, one can see that if γ and β are increasing with different values,
these figures are almost equal and in good agreement with one another, with respect to the correctness, and also the
absolute error is very small. Therefore, it can be stated that the wave solution is smoother around the 0 as γ and β
approaches 1.
Eur. Phys. J. Plus (2018) 133: 520 Page 17 of 19
Fig. 6. Behavior of the first RPS approximate solution (84) for different values: (a) γ = .1 and β = .3; (b) γ = .4 and β = .6;
(c) γ = .9 and β = .9.
Fig. 7. Behavior of the first RPS approximate solution (92) for different values: (a) γ = .1 and β = .3; (b) γ = .4 and β = .6;
(c) γ = .9 and β = .9.
Page 18 of 19 Eur. Phys. J. Plus (2018) 133: 520
Fig. 8. Behavior of the first RPS approximate solution (93) for different values: (a) γ = .1 and β = .3; (b) γ = .4 and β = .6;
(c) γ = .9 and β = .9.
9 Conclusion
In the present study, the symmetry variables and symmetry transformations of the third order space-time fractional
variant Boussinesq system have been derived. We have introduced the Lie symmetry analysis for investigating systems
of space-time NLFPDEs and successfully reduced the system of NLFPDEs into a system of NLFODEs. The explicit
power series solution of the considered system has been derived by using the reduced NLFODEs. Also, the obtained
power series solution has been analyzed for convergence. The conservation laws for the space-time fractional governing
system are constructed by using the generalized Noether’s theorem and new conservation theorem. Some authors have
studied the same system for the integer order derivative and for the time fractional order derivative. In comparison
with them, our result recovers the symmetries of the system of the integer order derivative for γ = β = 1 studied in
ref. [20]. The approximate solutions of the governing system subject to the given initial conditions are constructed with
accuracy using the RPSM, which is the modern analytical iterative technique. The graphical results demonstrated
that the nearly similar and coinciding behavior of the RPS approximate solution for γ = .1 and β = .3, γ = .4 and
β = .6 and γ = .9 and β = .9 in terms of accuracy.
Support of CSIR Research Grant to one of the authors, BK, for carrying out the research work is fully acknowledged.
References
1. M. Dehghan, J. Manafian, A. Saadatmandi, Numer. Methods Partial Differ. Equ. 26, 448 (2010).
2. B. Zheng, Commun. Theor. Phys. (Beijing) 58, 623 (2012).
3. S. Zhang, H.-Q. Zhang, Phys. Lett. A 375, 1069 (2011).
4. J.-H. He, X.-H. Wu, Chaos, Solitons Fractals 30, 700 (2006).
5. C. Chen, Y.-L. Jiang, Commun. Nonlinear Sci. Numer. Simul. 26, 24 (2015).
6. R.A. Leo, G. Sicuro, P. Tempesta, Fract. Calc. Appl. Anal. 20, 212 (2017).
7. G. Wang, A.H. Kara, K. Fakhar, Nonlinear Dyn. 82, 281 (2015).
8. G.W. Bluman, S.C. Anco, Symmetry and Integration Methods for Differential Equations (Springer-Verlag, New York, 2002).
Eur. Phys. J. Plus (2018) 133: 520 Page 19 of 19