Open Access
Issue
Wuhan Univ. J. Nat. Sci.
Volume 29, Number 6, December 2024
Page(s) 547 - 557
DOI https://doi.org/10.1051/wujns/2024296547
Published online 07 January 2025

© Wuhan University 2024

Licence Creative CommonsThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

0 Introduction

With the continuous development of mathematics and physics, fractional calculus has been widely discussed in recent years. Fractional-order model enables us to describe complex dynamics and physical behaviors better. Fractional calculus has important applications in fluid mechanics, nuclear magnetic resonance imaging, mechanics of complex viscoelastic materials, and so on[1-5]. Riewe[6,7] introduced fractional derivatives to deal with dissipative forces. After that, research on fractional derivatives has mainly focused on the left and right Riemann-Liouville, the left and right Caputo, the Riesz-Riemann-Liouville and the Riesz-Caputo fractional derivatives. Recently, a more general definition of fractional derivative, which is called combined fractional derivative, was introduced. For example, Malinowska and Torres[8] studied fractional calculus of variations based on a combined Caputo fractional derivative. Zhang[9] established the fractional differential equations of motion using a combined Riemann-Liouville fractional derivative. In 2015, Luo et al[10] presented the dynamics of a Birkhoffian system with both combined Caputo fractional derivatives and combined Riemann-Liouville fractional derivatives. Specifically, the four derivatives mentioned above are all special cases of the combined fractional derivative. Therefore, the combined fractional derivative is general.

The singular system refers to the procedure described by the singular Lagrangian. The singular system plays a vital role in modern quantum field theory, such as gravity theory, string (membrane) field theory, Yang-Mills theory, supersymmetry, and supergravity[11]. When a singular system is described by the canonical variables, there are several inherent constraints. In this case, the system is called the constrained Hamiltonian system, which is essential in condensed matter theory, gauge field theory and many other aspects[12-14].

Noether symmetry and Lie symmetry are two commonly used methods to study constrained mechanics systems. Noether symmetry was put forward by Noether[15]. Lie symmetry was introduced by Lutzky[16]. Then Noether and Lie symmetry were studied further[17,18]. Particularly, for singular systems, Lie symmetry and conserved quantity based on Riemann-Liouville fractional derivative[19], Noether symmetry and conserved quantity based on mixed order derivative and Caputo fractional derivative[20] are studied. Besides, based on the combined fractional derivative, Noether symmetry and conserved quantity for the Birkhoffian system are studied[21]. In this paper, we intend to study Noether symmetry and Lie symmetry for the singular system based on the combined fractional derivative.

Section 1 gives the preliminaries. In Section 2, the fractional Lagrange equation is analyzed using the combined fractional derivative. In Section 3 and Section 4, the fractional primary constraint is established and the Hamilton equation of the fractional constraint is given. In Section 5 and Section 6, the fractional Noether symmetry, Lie symmetry, and conserved quantities are studied, respectively. An example is given in Section 7 and the conclusion is given in Section 8.

1 Preliminaries

The following preliminaries are about the combined fractional derivatives, which can be seen in Ref. [8], Ref. [22] and Ref. [23]. The combined fractional derivative includes the combined Riemann-Liouville fractional derivative and the combined Caputo fractional derivative.

The combined Riemann-Liouville fractional derivative and combined Caputo fractional derivative are

R L D γ α , β f ( t ) = γ t 1 R L D t α f ( t ) + ( - 1 ) n ( 1 - γ ) t R L D t 2 β f ( t ) (1)

C D γ α , β f ( t ) = γ t 1 C D t α f ( t ) + ( - 1 ) n ( 1 - γ ) t C D t 2 β f ( t ) (2)

where t1RLDtαf(t), tRLDt2βf(t), t1CDtαf(t) and tCDt2βf(t) are the left and right Riemann-Liouville and Caputo fractional derivatives, t[t1,t2], 0γ1, n-1α,β<n.

Under the condition 0<α,β<1, we have

t 1 R L D t α f ( t ) = D t 1 C t α f ( t ) - 1 Γ ( 1 - α ) f ( t 1 ) ( t - t 1 ) α (3)

t R L D t 2 β f ( t ) = D t C t 2 β f ( t ) + 1 Γ ( 1 - β ) f ( t 2 ) ( t 2 - t ) β (4)

Additionally, the fractional partial integral formulae are

t 1 t 2 [ * ] D t 1 R L t α η d t = t 1 t 2 η D t C D t 2 α [ * ] d t - j = 0 n - 1 ( - 1 ) n + j D t 1 R L D t α + j - n η ( t ) D n - 1 - j [ * ] | t 1 t 2 (5)

t 1 t 2 [ * ] D t R L t 2 β η d t = t 1 t 2 η D t 1 C D t β [ * ] d t - j = 0 n - 1 t R L D t 2 β + j - n η ( t ) D n - 1 - j [ * ] | t 1 t 2 (6)

t 1 t 2 [ * ] D t 1 C t α η d t = t 1 t 2 η D t R L D t 2 α [ * ] d t + j = 0 n - 1 t R L D t 2 α + j - n [ * ] D n - 1 - j η ( t ) | t 1 t 2 (7)

t 1 t 2 [ * ] D t 1 C t β η d t = t 1 t 2 η D t 1 R L D t β [ * ] d t + j = 0 n - 1 ( - 1 ) n + j D t 1 R L D t β + j - n [ * ] D n - 1 - j η ( t ) | t 1 t 2 (8)

t 1 t 2 [ * ] D t 1 R t 2 α η d t = ( - 1 ) n t 1 t 2 η D t 1 R C D t 2 α [ * ] d t - j = 0 n - 1 ( - 1 ) n + j D t 1 R D t 2 α + j - n η ( t ) D n - 1 - j [ * ] | t 1 t 2 (9)

t 1 t 2 [ * ] D t 1 R C t 2 α η d t = ( - 1 ) n t 1 t 2 η D t 1 R D t 2 α [ * ] d t + j = 0 n - 1 ( - 1 ) j D t 1 R D t 2 α + j - n [ * ] D n - 1 - j η ( t ) | t 1 t 2 (10)

where D=d/dt represents the integer order derivative.

It should be noted that in this paper, we set 0<α,β<1.

2 Fractional Lagrange Equation

In this section, the fractional variational problems under combined fractional derivatives are studied.

We give a function

I R L [ q R L ( ) ] = t 1 t 2 L R L ( t , q R L , D R L D γ α , β q R L ) d t ,   q R L ( t 1 ) = q R L 1 ,   q R L ( t 2 ) = q R L 2 , (11)

where qRL=(qRL1,qRL2,,qRLn), RLγα,βqRL=(RLDγα,βqRL1,DRLDγα,βqRL2,,RLDγα,βqRLn), qRL1=(qRL11,qRL12,,qRL1n), qRL2=(qRL21,qRL22,,qRL2n), and the Lagrangian LRL is assumed to be a C2 function. If qRL() is an extremal of Eq. (11), we obtain

d d ε R L I R L [ q R L + ε R L h R L ] | ε R L = 0 = 0 , (12)

where εRL is a small parameter, hRL=(hRL1,hRL2,,hRLn), hRL(t1)=hRL(t2)=0. Further, we get

t 1 t 2 ( L R L q R L k h R L k + L R L D R L D γ α , β q R L k D R L D γ α , β h R L k ) d t = 0 ,   k = 1,2 , , n . (13)

Using fractional partial integral formulae (Eqs. (5) and (6)) in the second terms of Eq. (13), we obtain

t 1 t 2 L R L D R L D γ α , β q R L k D R L D γ α , β h R L k d t = - t 1 t 2 ( h R L k D C D 1 - γ β , α L R L R L D γ α , β q R L k ) d t + γ Γ ( 1 - α ) t 1 t 2 ( t 2 - t ) - α h R L k d t L R L ( t 2 ) R L D γ α , β q R L k - 1 - γ Γ ( 1 - β ) t 1 t 2 ( t - t 1 ) - β h R L k   d t L R L ( t 1 ) R L D γ α , β q R L k (14)

where LRL(t1)RLDγα,βqRLk=LRL(t1,qRL(t1),RLDγα,βqRL(t1))RLDγα,βqRLk, LRL(t2)RLDγα,βqRLk=LRLRLDγα,βqRLk(t2,qRL(t2),RLDγα,βqRL(t2)). Substituting Eq. (14) into Eq. (13), we obtain

t 1 t 2 [ L R L q R L k - D C D 1 - γ β , α L R L R L D γ α , β q R L k + γ ( t 2 - t ) - α Γ ( 1 - α ) L R L ( t 2 ) R L D γ α , β q R L k - ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) L R L ( t 1 ) R L D γ α , β q R L k ] h R L k d t = 0 . (15)

According to the fundamental lemma of variational calculation[24], we can deduce that

L R L q R L k - D C 1 - γ β , α L R L R L D γ α , β q R L k + γ ( t 2 - t ) - α Γ ( 1 - α ) L R L ( t 2 ) R L D γ α , β q R L k - ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) L R L ( t 1 ) R L D γ α , β q R L k = 0 , k = 1,2 , , n . (16)

Eq. (16) is called the fractional Lagrange equation within the combined Riemann-Liouville fractional derivative.

The fractional Lagrange equation has another form, which can be obtained from Eqs. (3), (4), (7), (8) and (13) as

L R L q R L k - D R L 1 - γ β , α L R L D R L D γ α , β q R L k = 0 , k = 1,2 , n . (17)

Similarly, the fractional Lagrange equation within the combined Caputo fractional derivative can also be obtained as

L C q C k - D R L 1 - γ β , α L C C D γ α , β q C k = 0 ,   k = 1,2 , , n , (18)

where the Lagrangian LC=LC(t,qC,DCDγα,βqC), LC is assumed to be a C2 function,qC=(qC1,qC2,,qCn), Cγα,βqC=(CDγα,βqC1,DCDγα,βqC2,,DCDγα,βqCn).

Remark 1   Eqs. (16) and (18) are general and universal, so we can select different values of γ to get different results.

3 Fractional Primary Constraint

This section presents fractional primary constraints within the combined Riemann-Liouville fractional derivative and combined Caputo fractional derivative.

Define fractional generalized momentum and Hamiltonian as

p R L k = L R L ( t , q R L , D R L D γ α , β q R L ) D R L D γ α , β q R L k , (19)

H R L = p R L k D R L γ α , β q R L k - L R L ( t , q R L , D R L D γ α , β q R L ) ,   k = 1,2 , , n . (20)

Here, we consider that the LRL(t,qRL,DRLDγα,βqRL) is singular, meaning that only a portion of the terms RLγα,βqRLk, k=1,2,,n can be solved. In this case, we assume the number is R, 0R<n.

Firstly, when 1R<n, we have

R L D γ α , β q R L σ = f R L σ ( t , q R L , p R L E , D R L D γ α , β q R L ρ ) ,   σ , E = 1,2 , , R ,   ρ = R + 1 , , n , (21)

where pRLE=(pRL1,pRL2,,pRLR), RLDγα,βqRLρ=(RLDγα,βqRL(R+1),RLDγα,βqRL(R+2),,RLDγα,βqRLn), 1R<n. Substituting Eq. (21) into Eq. (19), we have

p R L k = g R L k ( t , q R L , f R L σ ( t , q R L , p R L E , D R L D γ α , β q R L ρ ) , D R L D γ α , β q R L ρ ) = g R L k ( t , q R L , p R L E , D R L D γ α , β q R L ρ ) , k = 1,2 , , n . (22)

If k=1,2,,R, Eq. (22) obviously holds. If k=R+1,,n, we have

p R L ρ = g R L ρ ( t , q R L , p R L E )   o r   p R L F = g R L F ( t , q R L , p R L E ) ,   1 R < n , (23)

where pRLF=(pRL(R+1),,pRLn), gRLF=(gRL(R+1),,gRLn), ρ=R+1,,n. Another form of Eq. (23) is

ϕ R L ( t , q R L , p R L ) = p R L F - g R L F ( t , q R L , p R L E ) = 0 , (24)

where ϕRL=(ϕRL1,ϕRL2,,ϕRL(n-R)), pRL=(pRL1,pRL2,,pRLn), 1R<n.

Secondly, when R=0, RLγα,βqRLk cannot be solved. Therefore, from Eq. (19), we have

p R L k = g R L k ( t , q R L ) ,   k = 1,2 , , n . (25)

ϕ R L a ( t , q R L , p R L ) = p R L a - g R L a ( t , q R L ) = 0 ,   a = 1,2 , , n . (26)

From Eq. (24) and Eq. (26), we can obtain

ϕ R L a ( t , q R L , p R L ) = 0 ,   0 R < n ,   a = 1,2 , , n - R . (27)

Eq. (27) is called fractional primary constraint within a combined Riemann-Liouville fractional derivative.

Similarly, we can define

p C k = L C ( t , q C , D C D γ α , β q C ) D C D γ α , β q C k ,   k = 1,2 , , n , (28)

H C = p C k D C γ α , β q C k - L C ( t , q C , D C D γ α , β q C ) ,   k = 1,2 , , n . (29)

The fractional primary constraint within the combined Caputo fractional derivative can be obtained as

ϕ C a ( t , q C , p C ) = 0 ,   0 R < n ,   a = 1,2 , , n - R ,   p C = ( p C 1 , p C 2 , , p C n ) . (30)

Remark 2   According to the definition of the fractional generalized momenta (Eqs. (19) and (28)) rather than the fractional Euler-Lagrange equations (Eqs. (16) and (18)), the fractional primary constraints (Eqs. (27) and (30)) are obtained.

4 Fractional Constrained Hamilton Equation

Firstly, the fractional constrained Hamilton equation within the combined Riemann-Liouville fractional derivative is studied.

From Eq. (20), we obtain

δ H R L = δ p R L k D R L γ α , β q R L k - L R L q R L k δ q R L k ,   k = 1,2 , , n . (31)

Besides, because of the Hamiltonian HRL=HRL(t,qRL,pRL), we have

δ H R L = H R L q R L k δ q R L k + H R L p R L k δ p R L k ,   k = 1,2 , , n . (32)

It follows from Eqs. (31) and (32) that

( R L D γ α , β q R L k - H R L p R L k ) δ p R L k - ( L R L q R L k + H R L q R L k ) δ q R L k = 0 . (33)

Making use of Eqs. (16) and (19), the term LRL/qRLk in Eq. (33) can be replaced by C1-γβ,αpRLk-γ(t2-t)-αΓ(1-α)pRLk(t2)+(1-γ)(t-t1)-βΓ(1-β)pRLk(t1), thus we have

( R L D γ α , β q R L k - H R L p R L k ) δ p R L k - [ C D 1 - γ β , α p R L k - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L k ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L k ( t 1 ) + H R L q R L k ] δ q R L k = 0 ,   k = 1,2 , , n , (34)

where pRLk(t1)=LRL(t1)RLDγα,βqRLk, pRLk(t2)=LRL(t2)RLDγα,βqRLk. When the system (Eq. (16)) is singular, we have

λ R L a ϕ R L a q R L k δ q R L k + λ R L a ϕ R L a p R L k δ p R L k = 0 . (35)

It follows from Eqs. (34) and (35) that

C 1 - γ β , a p R L k = - H R L q R L k + γ ( t 2 - t ) - α Γ ( 1 - α ) p R L k ( t 2 ) - ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L k ( t 1 ) - λ R L a ϕ R L a q R L k ,

R L γ α , β q R L k = H R L p R L k + λ R L a ϕ R L a p R L k ,   a = 1,2 , , n - R ,   0 R < n ,   k = 1,2 , , n . (36)

Eq. (36) is called fractional constrained Hamilton equation within a combined Riemann-Liouville fractional derivative.

Similarly, we can also get the Hamilton equation with fractional constraints within the combined Caputo fractional derivative

R L 1 - γ α , β p C k = - H C q C k - λ C a ϕ C a q C k , D C γ α , β q C k = H C p C k + λ C a ϕ C a p C k ,   k = 1,2 , , n ,   a = 1,2 , , n - R ,   0 R < n . (37)

Remark 3   Because of the various values of γ, different results can be obtained from Eqs. (36) and (37).

According to the above method of establishing the Hamilton equation with fractional constraints, it is very important to solve the Lagrange multiplier. Next, we will calculate the Lagrange multiplier by fractional Poisson bracket.

Let F=F(t,q,p), G=G(t,q,p), q=(q1,q2,,qn), p=(p1,p2,,pn), then we define

{ F , G } = F q k G p k - F p k G q k ,   k = 1,2 , , n . (38)

Then we have

( { ϕ R L a , H R L } + λ R L b { ϕ R L a , ϕ R L b } ) ϕ R L a q R L j q ˙ R L j + ( ϕ R L a t + ϕ R L a p R L j p ˙ R L j ) × ϕ R L a q R L k D R L γ α , β q R L k - ( ϕ R L a p R L k ϕ R L a q R L j q ˙ R L j ) [ C D 1 - γ β , α p R L k - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L k ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L k ( t 1 ) ] = 0 , a , b = 1,2 , , n - R ,   k , j = 1,2 , , n ,   0 R < n , (39)

and

( { ϕ C a , H C } + λ C b { ϕ C a , ϕ C b } ) ϕ C a q C j q ˙ C j + ( ϕ C a t + ϕ C a p C j p ˙ C j ) × ϕ C a q C k D C γ α , β q C k - ϕ C a p C k ϕ C a q C j q ˙ C j D R L 1 - γ β , α p C k = 0 , a , b = 1,2 , , n - R ,   k , j = 1,2 , , n ,   0 R < n . (40)

From Eqs. (39) and (40), the Lagrange multipliers can be calculated.

5 Noether Theorem

Definition 1   A quantity C is called a conserved quantity if and only if dC/dt=0.

5.1 Noether Theorem within the Combined Riemann-Liouville Fractional Derivative

The Hamilton action within the combined Riemann-Liouville fractional derivative is

I R L = t 1 t 2 [ p R L k D R L D γ α , β q R L k - H R L ( t , q R L , p R L ) ] d t . (41)

The infinitesimal transformations are

t ¯ = t + Δ t ,   q ¯ R L k ( t ¯ ) = q R L k ( t ) + Δ q R L k ,    p ¯ R L k ( t ¯ ) = p R L k ( t ) + Δ p R L k ,   k = 1,2 , , n . (42)

Namely,

t ¯ = t + θ R L ξ R L 0 ( t , q R L , p R L ) + o ( θ R L ) ,   q ¯ R L k ( t ¯ ) = q R L k ( t ) + θ R L ξ R L k ( t , q R L , p R L ) + o ( θ R L ) , p ¯ R L k ( t ¯ ) = p R L k ( t ) + θ R L η R L k ( t , q R L , p R L ) + o ( θ R L ) , (43)

where θRL is a small parameter, ξRL0, ξRLk, and ηRLk are infinitesimal generators, and o(θRL) means the higher order of θRL.

Letting ΔIRL be the linear part of I¯RL-IRL, and without considering the higher order of θRL, we obtain

Δ I R L = θ R L t 1 t 2 [ p R L k D R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ( p R L k D R L D γ α , β q R L k - H R L ) ξ ˙ R L 0 + ( p R L k d d t D R L D γ α , β q R L k - H R L t ) ξ R L 0 - H R L q R L k ξ R L k + λ R L a ϕ R L a p R L k η R L k + q R L k ( t 2 ) ξ R L 0 ( t 2 ) × ( 1 - γ ) p R L k Γ ( 1 - β ) d d t ( t 2 - t ) - β - q R L k ( t 1 ) ξ R L 0 ( t 1 ) γ p R L k Γ ( 1 - α ) d d t ( t - t 1 ) - α ] d t (44)

where ξRL0(t1)=ξRL0(t1,qRL(t1),pRL(t1)) and ξRL0(t2)=ξRL0(t2,qRL(t2),pRL(t2)). Let ΔIRL=0, then Eq. (44) gives

p R L k R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ( p R L k D R L D γ α , β q R L k - H R L ) ξ ˙ R L 0 + λ R L a ϕ R L a p R L k η R L k + ( p R L k d d t D R L D γ α , β q R L k - H R L t ) ξ R L 0 - H R L q R L k ξ R L k + q R L k ( t 2 ) ξ R L 0 ( t 2 ) × ( 1 - γ ) p R L k Γ ( 1 - β ) d d t ( t 2 - t ) - β - q R L k ( t 1 ) ξ R L 0 ( t 1 ) γ p R L k Γ ( 1 - α ) d d t ( t - t 1 ) - α = 0 (45)

Equation (45) is called the fractional Noether identity within the combined Riemann-Liouville fractional derivative.

Theorem 1   If ξRL0, ξRLk, and ηRLk satisfy Eq. (45), then there exists a conserved quantity

C R L = ( p R L k D R L D γ α , β q R L k - H R L ) ξ R L 0 + t 1 t { p R L k D R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ( ξ R L k - q ˙ R L k ξ R L 0 ) [ C D 1 - γ β , α p R L k - γ ( t 2 - τ ) - α Γ ( 1 - α ) p R L k ( t 2 ) + ( 1 - γ ) ( τ - t 1 ) - β Γ ( 1 - β ) p R L k ( t 1 ) ] } d τ + q R L k ( t 2 ) ξ R L 0 ( t 2 ) 1 - γ Γ ( 1 - β ) t 1 t p R L k d d τ ( t 2 - τ ) - β d τ - q R L k ( t 1 ) ξ R L 0 ( t 1 ) γ Γ ( 1 - α ) t 1 t p R L k d d τ ( τ - t 1 ) - α d τ (46)

for the system within the combined Riemann-Liouville fractional derivative.

Proof   Using Eqs. (27), (36), and (45), we have

d C R L d t = ( p R L k D R L D γ α , β q R L k - H R L ) ξ ˙ R L 0 + ξ R L 0 ( p ˙ R L k D R L D γ α , β q R L k + p R L k d d t D R L D γ α , β q R L k - H R L t - H R L q R L k q ˙ R L k - H R L p R L k p ˙ R L k )

+ p R L k R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ( ξ R L k - q ˙ R L k ξ R L 0 ) [ C D 1 - γ β , α p R L k - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L k ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L k ( t 1 ) ]

+ q R L k ( t 2 ) ξ R L 0 ( t 2 ) 1 - γ Γ ( 1 - β ) p R L k d d t ( t 2 - t ) - β - q R L k ( t 1 ) ξ R L 0 ( t 1 ) γ Γ ( 1 - α ) p R L k d d t ( t - t 1 ) - α

= H R L q R L k ξ R L k - λ R L a ϕ R L a p R L k η R L k + ξ R L 0 ( p ˙ R L k λ R L a ϕ R L a p R L k - H R L q R L k q ˙ R L k ) + ( ξ R L k - q ˙ R L k ξ R L 0 ) ( - H R L q R L k - λ R L a ϕ R L a q R L k ) = - λ R L a ϕ R L a p R L k ( η R L k - p ˙ R L k ξ R L 0 ) - λ R L a ϕ R L a q R L k ( ξ R L k - q ˙ R L k ξ R L 0 ) = - λ R L a δ ϕ R L a = 0 .

The proof is completed.

5.2 Noether Theorem within the Combined Caputo Fractional Derivative

The Hamilton action within the combined Caputo fractional derivative is

I C = t 1 t 2 [ p C k D C D γ α , β q C k - H C ( t , q C , p C ) ] d t . (47)

The infinitesimal transformations are

t ¯ = t + Δ t ,   q ¯ C k ( t ¯ ) = q C k ( t ) + Δ q C k , p ¯ C k ( t ¯ ) = p C k ( t ) + Δ p C k ,   k = 1,2 , , n . (48)

Namely,

t ¯ = t + θ C ξ C 0 ( t , q C , p C ) + o ( θ C ) ,   q ¯ C k ( t ¯ ) = q C k ( t ) + θ C ξ C k ( t , q C , p C ) + o ( θ C ) ,   p ¯ C k ( t ¯ ) = p C k ( t ) + θ C η C k ( t , q C , p C ) + o ( θ C ) (49)

where θC is a small parameter, ξC0, ξCk, and ηCk are infinitesimal generators. Letting ΔIC=I¯C-IC=0, we have

p C k D C γ α , β ( ξ C k - q ˙ C k ξ C 0 ) + ( p C k D C D γ α , β q C k - H C ) ξ ˙ C 0 + ( p C k d d t D C D γ α , β q C k - H C t ) ξ C 0 - H C q C k ξ C k + λ C a ϕ C a p C k η C k + q ˙ C k ( t 2 ) ξ C 0 ( t 2 ) ( 1 - γ ) p C k Γ ( 1 - β ) ( t 2 - t ) - β - q ˙ C k ( t 1 ) ξ C 0 ( t 1 ) γ p C k Γ ( 1 - α ) ( t - t 1 ) - α = 0 (50)

Equation (50) is called fractional Noether identity within a combined Caputo fractional derivative.

Theorem 2   If ξC0, ξCk, and ηCk satisfy Eq. (50), then there exists a conserved quantity

C C = ( p C k D C D γ α , β q C k - H C ) ξ C 0 + t 1 t [ p C k D C D γ α , β ( ξ C k - q C k ξ C 0 ) + ( ξ C k - q ˙ C k ξ C 0 ) D R L D 1 - γ β , α p C k ] d τ + q ˙ C k ( t 2 ) ξ C 0 ( t 2 ) 1 - γ Γ ( 1 - β ) t 1 t p C k ( t 2 - τ ) - β d τ - q ˙ C k ( t 1 ) ξ C 0 ( t 1 ) γ Γ ( 1 - α ) t 1 t p C k ( τ - t 1 ) - α d τ (51)

for the system within a combined Caputo fractional derivative.

Proof   Using Eqs. (30), (37), and (50), it is easy to obtain

d C C d t = ( p C k D C D γ α , β q C k - H C ) ξ ˙ C 0 + ξ C 0 ( p ˙ C k D C D γ α , β q C k + p C k d d t D C D γ α , β q C k - H C t - H C q C k q ˙ C k - H C p C k p ˙ C k ) + p C k D C γ α , β ( ξ C k - q ˙ C k ξ C 0 ) + ( ξ C k - q ˙ C k ξ C 0 ) D R L 1 - γ β , α p C k + q ˙ C k ( t 2 ) ξ C 0 ( t 2 ) ( 1 - γ ) p C k Γ ( 1 - β ) ( t 2 - t ) - β - q ˙ C k ( t 1 ) ξ C 0 ( t 1 ) γ p C k Γ ( 1 - α ) ( t - t 1 ) - α + q ˙ C k ( t 2 ) ξ C 0 ( t 2 ) ( 1 - γ ) p C k Γ ( 1 - β ) ( t 2 - t ) - β - q ˙ C k ( t 1 ) ξ C 0 ( t 1 ) γ p C k Γ ( 1 - α ) ( t - t 1 ) - α = H C q C k ξ C k - λ C a ϕ C a p C k η C k + ξ C 0 ( p ˙ C k λ C a ϕ C a p C k - H C q C k q ˙ C k ) - ( ξ C k - q C k ξ C 0 ) ( H C q C k + λ C a ϕ C a q C k ) = - λ C a ϕ C a p C k ( η C k - ξ C 0 p ˙ C k ) - λ C a ϕ C a q C k ( ξ C k - q ˙ C k ξ C 0 ) = - λ C a δ ϕ C a = 0

The proof is completed.

6 Lie Symmetry Conserved Quantity

6.1 Lie Symmetry Conserved Quantity within the Combined Riemann-Liouville Fractional Derivative

Eq. (36) can be expressed as

R L D γ α , β q R L k = h R L k ( t , q R L , p R L ) ,   k = 1,2 , , n , (52)

C 1 - γ β , α p R L k = f R L k ( t , q R L , p R L ) ,   k = 1,2 , , n . (53)

We then study Eqs. (52) and (53) under the infinitesimal transformations (Eq. (43)). For Eq. (52), we obtain

R L D γ ¯ α , β q ¯ R L k - h R L k ( t ¯ , q ¯ R L k , p ¯ R L k ) = D R L γ α , β q R L k - h R L k ( t , q R L , p R L ) + θ R L [ R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ξ R L 0 d d t D R L D γ α , β q R L k - X R L ( 0 ) ( h R L k ) - q R L k ( t 1 ) γ ξ R L 0 ( t 1 ) Γ ( 1 - α ) d d t ( t - t 1 ) - α + q R L k ( t 2 ) ξ R L 0 ( t 2 ) 1 - γ Γ ( 1 - β ) d d t ( t 2 - t ) - β ] (54)

where XRL(0)=ξRL0t+ξRLiqRLi+ηRLipRLi, i=1,2,,n. For Eq. (53), we have

C 1 - γ β , α p ¯ R L k = f R L k ( t ¯ , q ¯ R L , p ¯ R L ) = D C 1 - γ β , α p R L k - f R L k ( t , q R L , p R L ) + θ R L [ C D 1 - γ β , α ( η R L k - p ˙ R L k ξ R L 0 ) + ξ R L 0 d d t D C D 1 - γ β , α p R L k - X R L ( 0 ) ( f R L k ) - 1 - γ Γ ( 1 - β ) ( t - t 1 ) - β p ˙ R L k ( t 1 ) ξ R L 0 ( t 1 ) + γ Γ ( 1 - α ) ( t 2 - t ) - α p ˙ R L k ( t 2 ) ξ R L 0 ( t 2 ) ] (55)

The fractional primary constraint (Eq. (27)) gives

ϕ R L a ( t , q ¯ R L , p ¯ R L ) = ϕ R L a ( t , q R L , p R L ) + θ R L X R L ( 0 ) ( ϕ R L a ) . (56)

Therefore, we have

R L D γ α , β ( ξ R L k - q ˙ R L k ξ R L 0 ) + ξ R L 0 d d t D R L γ α , β q R L k - X R L ( 0 ) ( h R L k ) - q R L k ( t 1 ) γ ξ R L 0 ( t 1 ) Γ ( 1 - α ) d d t ( t - t 1 ) - α + q R L k ( t 2 ) ξ R L 0 ( t 2 ) 1 - γ Γ ( 1 - β ) d d t ( t 2 - t ) - β = 0 , (57)

C 1 - γ β , α ( η R L k - p ˙ R L k ξ R L 0 ) + ξ R L 0 d d t D C 1 - γ β , α p R L k - X R L ( 0 ) ( f R L k ) - 1 - γ Γ ( 1 - β ) ( t - t 1 ) - β p ˙ R L k ( t 1 ) ξ R L 0 ( t 1 ) + γ Γ ( 1 - α ) ( t 2 - t ) - α p ˙ R L k ( t 2 ) ξ R L 0 ( t 2 ) = 0 , (58)

and

X R L ( 0 ) ( ϕ R L a ) = 0 . (59)

Equations (57) and (58) are called the determined equations within the combined Riemann-Liouville fractional derivative, and Eq. (59) is called the limited equation within the combined Riemann-Liouville fractional derivative.

Besides, if we consider the deduction process of the system (Eq. (36)), an extra additional limited equation

ϕ R L a q R L i ( ξ R L i - q ˙ R L i ξ R L 0 ) + ϕ R L a p R L i ( η R L i - p ˙ R L i ξ R L 0 ) = 0 (60)

can be obtained. Then we have

Theorem 3   For the system within the combined Riemann-Liouville fractional derivative (Eq. (36)), when Eqs. (45), (57) and (58) are satisfied, Eq. (46) gives a Lie symmetry conserved quantity; when Eqs. (45), (57)-(59) are satisfied, Eq. (46) gives a weak Lie symmetry conserved quantity; when Eqs. (45), (57)-(60) are satisfied, Eq. (46) gives a strong Lie symmetry conserved quantity.

6.2 Lie Symmetry Conserved Quantity within the Combined Caputo Fractional Derivative

Eq. (37) can be expressed as

C γ α , β q C k = h C k ( t , q C , p C ) ,   k = 1,2 , , n , (61)

R L 1 - γ β , α p C k = f C k ( t , q C , p C ) ,   k = 1,2 , , n . (62)

Using the similar way, we can obtain the determined equations within the combined Caputo fractional derivative

C D γ α , β ( ξ C k - q ˙ C k ξ C 0 ) + ξ C 0 d d t D C γ α , β q C k - X C ( 0 ) ( h C k ) - q ˙ C k ( t 1 ) γ ξ C 0 ( t 1 ) Γ ( 1 - α ) ( t - t 1 ) - α + q ˙ C k ( t 2 ) ξ C 0 ( t 2 ) 1 - γ Γ ( 1 - β ) ( t 2 - t ) - β = 0 , (63)

R L 1 - γ β , α ( η C k - p ˙ C k ξ C 0 ) + ξ C 0 d d t D R L 1 - γ β , α p C k - X C ( 0 ) - 1 - γ Γ ( 1 - β ) p C k ( t 1 ) ξ C 0 ( t 1 ) d d t ( t - t 1 ) - β + γ Γ ( 1 - α ) p C k ( t 2 ) ξ C 0 ( t 2 ) d d t ( t 2 - t ) - α = 0 , (64)

the limited equation within the combined Caputo fractional derivative

X C ( 0 ) ( ϕ C a ) = 0 , (65)

and the additional limited equation within the combined Caputo fractional derivative

ϕ C a q C i ( ξ C i - q ˙ C i ξ C 0 ) + ϕ C a p C i ( η C i - p ˙ C i ξ C 0 ) = 0 , (66)

where XC(0)=ξC0t+ξCiqCi+ηCipCi, i=1,2,,n. Then we have

Theorem 4   For the system within the combined Caputo fractional derivative (Eq. (37)), when Eqs. (50), (63) and (64) are satisfied, Eq. (51) gives a Lie symmetry conserved quantity; when Eqs. (50), (63)-(65) are satisfied, Eq. (51) gives a weak Lie symmetry conserved quantity; when Eqs. (50), (63)-(66) are satisfied, Eq. (51) gives a strong Lie symmetry conserved quantity.

7 An Example

The singular Lagrangian is

L R L = - c 2 ( q R L 1 ) 2 + b 2 ( q R L 2 ) 2 + 1 2 ( q R L 2 R L D γ α , β q R L 1 - q R L 1 R L D γ α , β q R L 2 ) . (67)

Try to study its Noether symmetry and conserved quantity.

From Eqs. (19) and (20), we can obtain

p R L 1 = L R L R L D γ α , β q R L 1 = 1 2 q R L 2 ,   p R L 2 = L R L R L D γ α , β q R L 2 = - 1 2 q R L 1 , (68)

H R L = p R L 1 R L D γ α , β q R L 1 + p R L 2 R L D γ α , β q R L 2 - L R L = c 2 ( q R L 1 ) 2 - b 2 ( q R L 2 ) 2 . (69)

Then Eq. (27) gives two fractional primary constraints

ϕ R L 1 = p R L 1 - 1 2 q R L 2 = 0 ,   ϕ R L 2 = p R L 2 + 1 2 q R L 1 = 0 . (70)

From Eq. (39), we obtain

b q ˙ R L 1 q R L 2 + λ R L 1 q ˙ R L 1 + p ˙ R L 2 R L D γ α , β q R L 1 - q ˙ R L 1 [ C D 1 - γ β , α p R L 2 - ( t 2 - t ) - α Γ ( 1 - α ) p R L 2 ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L 2 ( t 1 ) ] = 0 , (71)

c q R L 1 q ˙ R L 2 + λ R L 2 q ˙ R L 2 - p ˙ R L 1 R L D γ α , β q R L 2 + q ˙ R L 2 [ C D 1 - γ β , α p R L 1 - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L 1 ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L 1 ( t 1 ) ] = 0 . (72)

Then, using Eq. (36), the fractional constrained Hamilton equation within the combined Riemann-Liouville fractional derivative is

R L D γ α , β q R L 1 = - 2 b q R L 2 + 2 [ C D 1 - γ β , α p R L 2 - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L 2 ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L 2 ( t 1 ) ] ,

R L D γ α , β q R L 2 = - 2 c q R L 1 - 2 [ C D 1 - γ β , α p R L 1 - γ ( t 2 - t ) - α Γ ( 1 - α ) p R L 1 ( t 2 ) + ( 1 - γ ) ( t - t 1 ) - β Γ ( 1 - β ) p R L 1 ( t 1 ) ] . (73)

From Noether identity (Eq. (45)), we have

p R L 1 D R L γ α , β ( ξ R L 1 - q ˙ R L 1 ξ R L 0 ) + p R L 2 R L D γ α , β ( ξ R L 2 - q ˙ R L 2 ξ R L 0 ) - c q R L 1 ξ R L 1 + b q R L 2 ξ R L 2 + ( p R L 1 d d t D R L D γ α , β q R L 1 + p R L 2 d d t D R L D γ α , β q R L 2 ) ξ R L 0 + λ R L 1 η R L 1 + λ R L 2 η R L 2 + ( p R L 1 R L D γ α , β q R L 1 + p R L 2 R L D γ α , β q R L 2 - H R L ) ξ ˙ R L 0 + q R L 1 ( t 2 ) ξ R L 0 ( t 2 ) × ( 1 - γ ) p R L 1 Γ ( 1 - β ) d d t ( t 2 - t ) - β - q R L 1 ( t 1 ) ξ R L 0 ( t 1 ) γ p R L 1 Γ ( 1 - α ) d d t ( t - t 1 ) - α + q R L 2 ( t 2 ) ξ R L 0 ( t 2 ) × ( 1 - γ ) p R L 2 Γ ( 1 - β ) d d t ( t 2 - t ) - β - q R L 2 ( t 1 ) ξ R L 0 ( t 1 ) γ p R L 2 Γ ( 1 - α ) d d t ( t - t 1 ) - α = 0 (74)

Then we can verify that

ξ R L 0 = - 1 ,   ξ R L 1 = ξ R L 2 = 0 ,   η R L 1 = η R L 2 = 0 (75)

satisfy Eq. (74). Therefore, we have

C R L = - ( p R L 1 R L D γ α , β q R L 1 + p R L 2 R L D γ α , β q R L 2 - c 2 q R L 1 2 + b 2 q R L 2 2 ) + t 1 t { p R L 1 d d τ D R L D γ α , β q R L 1 + p R L 2 d d τ D R L D γ α , β q R L 2 + q ˙ R L 1 [ C D 1 - γ β , α p R L 1 - γ ( t 2 - τ ) - α Γ ( 1 - α ) p R L 1 ( t 2 ) + ( 1 - γ ) ( τ - t 1 ) - β Γ ( 1 - β ) p R L 1 ( t 1 ) ] + q ˙ R L 2 [ C D 1 - γ β , α p R L 2 - γ ( t 2 - τ ) - α Γ ( 1 - α ) p R L 2 ( t 2 ) + ( 1 - γ ) ( τ - t 1 ) - β Γ ( 1 - β ) p R L 2 ( t 1 ) ] } d τ (76)

Specially, we degenerate the above fractional order to the integer order, namely α,β1, so we can get

C R L 0 = - c 2 q R L 1 2 + b 2 q R L 2 2 . (77)

If t[1,10], and let b=c=1,qRL1(1)=0,qRL2(1)=1, the trajectory of CRL0 can be drawn as follows.

It follows from Fig. 1 that CRL0 is a constant.

thumbnail Fig. 1 The trajectory of CRL0

8 Conclusion

Fractional calculus has important applications in various fields, and it is also a research hotspot. In this paper, based on the combined Riemann-Liouville fractional derivative and combined Caputo fractional derivative, we establish Noether theorem and Lie theorem for the fractional constrained Hamiltonian system. Besides, Noether identity, determined equation, limited equation and additional limited equation are presented.

In this paper, Noether-type conserved quantity is obtained from Lie symmetry, whether the Lie symmetry of the system can lead to another conserved quantity, called Hojman conserved quantity, remains to be studied. At the same time, Mei symmetry is also an effective method to solve the differential equation. Therefore, Mei symmetry is also a future research direction.

References

  1. Oldham K B, Spanier J. The Fractional Calculus [M]. New York: Academic Press, 1974. [MathSciNet] [Google Scholar]
  2. Luo S K. Analytical mechanics method of fractional dynamics and its applications [J]. Journal of Dynamics and Control, 2019, 17(5): 432-438. [Google Scholar]
  3. Yin Y D, Zhao D M, Liu J L, et al. Study on the rate dependency of acrylic elastomer based fractional viscoelastic model [J]. Chinese Journal of Theoretical and Applied Mechanics, 2022, 54(1): 155-163. [Google Scholar]
  4. Si H, Zheng Y A. Adaptive predictive synchronization of fractional order chaotic systems [J]. Journal of Dynamics and Control, 2021, 19(5): 8-12(Ch). [Google Scholar]
  5. Song C J. Generalized fractional forced Birkhoff equations [J]. Journal of Dynamics and Control, 2019, 17(5): 446-452. [Google Scholar]
  6. Riewe F. Nonconservative Lagrangian and Hamiltonian mechanics [J]. Phys Rev E, 1996, 53(2): 1890-1899. [CrossRef] [MathSciNet] [PubMed] [Google Scholar]
  7. Riewe F. Mechanics with fractional derivatives [J]. Phys Rev E, 1997, 55(3): 3581-3592. [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  8. Malinowska A B, Torres D F M. Fractional calculus of variations for a combined Caputo derivative [J]. Fract Calc Appl Anal, 2011, 14(4): 523-537. [Google Scholar]
  9. Zhang Y. Fractional differential equations of motion in terms of combined Riemann-Liouville derivatives [J]. Chin Phys B, 2012, 21(8): 84502. [Google Scholar]
  10. Luo S K, Xu Y L. Fractional Birkhoffian mechanics [J]. Acta Mech, 2015, 226(3): 829-844. [CrossRef] [MathSciNet] [Google Scholar]
  11. Li Z P. Classical and Quantal Dynamics of Constrained Systems and Their Symmetrical Properties [M]. Beijing: Beijing Polytechnic University Press, 1993(Ch). [Google Scholar]
  12. Dirac P A M. Lecture on Quantum Mechanics [M]. New York: Yeshiva University, 1964. [Google Scholar]
  13. Li Z P. Contrained Hamiltonian Systems and Their Symmetrical Properties [M]. Beijing: Beijing Polytechnic University Press, 1999. [Google Scholar]
  14. Li Z P, Jiang J H. Symmetries in Constrained Canonical Systems [M]. Beijing: Science Press, 2002. [Google Scholar]
  15. Noether A E. Invariante variationsprobleme [J]. Nachrichten von der Gesellschaft der Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-Physikalische Klasse, 1918, KI(2): 235-257. [Google Scholar]
  16. Lutzky M. Dynamical symmetries and conserved quantities [J]. Journal of Physics A: Mathematical and General, 1979, 12(7): 973-981. [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  17. Mei F X, Wu H B. Dynamics of Constrained Mechanical Systems [M]. Beijing: Beijing Institute of Technology Press, 2009. [Google Scholar]
  18. Mei F X. Analytical Mechanics (II) [M]. Beijing: Beijing Institute of Technology Press, 2013. [Google Scholar]
  19. Song C J, Wang J H. Conserved quantity for fractional constrained Hamiltonian system [J]. Wuhan University Journal of Natural Sciences, 2022, 27(3): 201-210. [CrossRef] [EDP Sciences] [Google Scholar]
  20. Song C J, Zhai X H. Noether theorem for fractional singular systems [J]. Wuhan University Journal of Natural Sciences, 2023, 28(3): 207-216. [CrossRef] [EDP Sciences] [Google Scholar]
  21. Song C J, Zhang Y. Noether symmetry and conserved quantity for fractional Birkhoffian mechanics and its applications [J]. Frac Cal Appl Anal, 2018, 21(2): 509-526. [CrossRef] [MathSciNet] [Google Scholar]
  22. Podlubny I. Fractional Differential Equations [M]. San Diego: Academic Press, 1999. [Google Scholar]
  23. Wu Q, Huang J H. Fractional Order Calculus [M]. Beijing: Tsinghua University Press, 2016(Ch). [Google Scholar]
  24. Gelfand I M, Fomin S V. Calculus of Variations [M]. London: Prentice-Hall, 1963. [Google Scholar]

All Figures

thumbnail Fig. 1 The trajectory of CRL0
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.