EDUCATION Revista Mexicana de Fı́sica E 59 (2013) 140–147 JULY–DECEMBER 2013

Variational symmetries of Lagrangians
G.F. Torres del Castillo
Departamento de Fı́sica Matemática, Instituto de Ciencias
Universidad Autónoma de Puebla, 72570 Puebla, Pue., México.
C. Andrade Mirón, and R.I. Bravo Rojas
Facultad de Ciencias Fı́sico Matemáticas Universidad Autónoma de Puebla,
Apartado postal 165, 72001 Puebla, Pue., México.
Received 25 June 2013; accepted 8 November 2013

We present an elementary derivation of the equation for the infinitesimal generators of variational symmetries of a Lagrangian for a system
with a finite number of degrees of freedom. We also give a simple proof of the existence of an infinite number of Lagrangians for a given
second-order ordinary differential equation.

Keywords: Lagrangians; symmetries; constants of motion; ordinary differential equations.

Presentamos una derivación elemental de la ecuación para los generadores infinitesimales de simetrı́as variacionales de una lagrangiana
para un sistema con un número finito de grados de libertad. Damos también una prueba simple de la existencia de un número infinito de
lagrangianas para una ecuación diferencial ordinaria de segundo orden dada.

Descriptores: Lagrangianas; simetrı́as; constantes de movimiento; ecuaciones diferenciales ordinarias.

PACS: 45.20.Jj; 02.30.Hq; 02.20.Sv

1. Introduction that even though, at least since the nineteenth century, it is
known that any second-order ODE possesses an infinite num-
In classical mechanics, the Lagrangian of a given mechanical ber of Lagrangians, this result is not presented in the standard
system leads to its equations of motion, and the identification textbooks on classical mechanics (see, however, Ref. [4]). In
of the continuous symmetries of the Lagrangian allows one Sec. 3 the formulas applicable to the case with an arbitrary
to find constants of motion (or first integrals). However, the number of degrees of freedom are given. Throughout this pa-
identification of such symmetries is often based on the exis- per various examples are given, illustrating the concepts and
tence of ignorable coordinates, which depends on the coordi- methods introduced here.
nates chosen, and the symmetries commonly considered are
restricted to rotations and translations (see, e.g., Refs. [1–3]).
The Lagrangian formalism is also useful in many other ar- 2. Systems with one degree of freedom
eas. Any second-order ordinary differential equation (ODE)
can be seen as the Euler–Lagrange equation for some La- In order to present the ideas in a simple way, it is convenient
grangian (in fact, for an infinite number of Lagrangians) and to consider firstly the case where there is only one degree of
many systems of second-order ODEs can be derived from a freedom, or we have a single second-order ODE.
Lagrangian.
It turns out that a Lagrangian may possess many non- 2.1. Variational symmetries of a Lagrangian
trivial continuous symmetries and, what is more relevant, in
many cases, some of them can be readily found by solving We shall consider one-parameter families of transformations
an equation applicable in any coordinate system. The aim of
this paper is to present an elementary derivation of the equa- x0 = x0 (x, t, s), t0 = t0 (x, t, s), (1)
tion that determines the so-called variational symmetries of a
Lagrangian, and of the expression for the constants of motion where s is a parameter that takes values in some neighbor-
associated with these symmetries. The results presented here hood of zero, and we shall assume that the transformation (1)
are applicable to any system of second-order ODEs derivable reduces to the identity for s = 0; that is x0 (x, t, 0) = x and
from a Lagrangian, and to any second-order ODE, not neces- t0 (x, t, 0) = t. For a fixed value of s, Eqs. (1) give a trans-
sarily related to classical mechanics. formation from the plane (x, t) into the plane (x0 , t0 ). Such
In Sec. 2 we consider systems with one degree of freedom transformations are called point transformations (see, e.g.,
(or a single second-order ODE), deriving the basic equations Refs. [5–8]; more general transformations are also useful,
that determine the variational symmetries of a Lagrangian see, e.g., Refs. [5, 6, 9]). Some examples of one-parameter
and the corresponding first integrals. In order to apply these families of point transformations are
results to any second-order ODE, we show how to find a La- 1
grangian for a given second-order ODE. It may be remarked x0 = xes − gt2 (e3s − es ), t0 = te3s/2 , (2)
2

t) = ẋ2 − . t). dt dt 2 3 t00 t0 showing that Eq. t)) = equivalently. t0 = L x. Fis. te2u ) = ((xe−u ) e−s . (6) 2 dt dt dt dt · ¸ ¾ 1 ¡ ¢ − g 2 xes − gt2 e3s − es te3s/2 e3s/2 for all values of s for which the transformation is defined. Mex. 8]. each one-parameter family of point transformations sat. (7) is 1 − ts 2(1 − ts)2 1 − ts equal to the difference of the values of F at the endpoints. = L x.. . which follows from the fact that 2 3 the Euler–Lagrange equation determines the local extrema of Indeed. + g t 1 − 6e + 5e . using the chain rule). t) if dt dt 6 1 £ ³ ´ ¤ µ ¶ 0 µ ¶ 2 dx0 dt dx d + g e−s/2 ẋ − e3s/2 − e−s/2 gt te3s/2 L x0 . as we shall see. which means that. ẋ. t. mapped into a curve that minimizes (or maximizes) the in- fined for all s ∈ R. t) ≡ (x0 . . t. into another solution. + F (x. te 2s+2u ) = ϕs+u (x. t0 ). t1 ) [1–3. and the point transformations satisfying 1 3 4¡ 4s 6s ¢ (6) with dF/dt 6= 0 are sometimes called Noether symme. 0 . We shall say that the one-parameter family of transfor. from Eqs. that does not depend on t. (3) tion (1). where t00 .t = (e ẋ) − e 0 dt dt 2 3 Zt1 µ 0 ¶ Zt1 µ ¶ µ ¶ 0 dx 0 0 dx t2 x6 L x . a curve that minimizes (or max- imizes) the first term on the right-hand side of Eq. t dt = ẋ2 − . s) is some function. t + F (x. t0 ). ẋ. [5–8]). the family of transformations (2) is a varia- parameter groups of transformations.t = e ẋ − e3s/2 − e−s/2 gt L(x. How- dt t0 ever. Condition (6) µ 0 ¶ 0 · ¸ amounts to 0 dx 0 dt (te2s )2 −3s 2 (xe−s )6 2s L x . ϕs (ϕu (x. 24 tries [5] or divergence symmetries [6]. t. ϕs (x. hence. te2s ). (x1 . Hence.) isfying Eq. (6) maps µ ¶ t2 x6 any solution of the Euler–Lagrange equation corresponding L(x. therefore. t1 ). L x. treating the derivative as a quotient of differentials (or. 0 . a transformation satisfying Eq. (te2u ) e2s ) = dx0 es dx − (e3s − es )gtdt −s/2 (xe −s−u . dt 0 e dt dt t0 and. For instance. The last term on the right-hand side of Eq. with fixed endpoints (x0 . t dt = L x. t01 are the values of t0 corresponding to the points (x0 . but (3) is defined only for s 6= 1/t. this problem is simplified if we start Rev. t dt. (7) making use of the definition (6). (x1 . . E 59 (2013) 140–147 . then 1 3 1 2 L(x. (4) the same endpoints. the transformations (4) are variational by the Lagrangian L. fact. Some authors reserve the µ ¶ · dx d 1 2 2 ¡ ¢ name variational symmetry for the point transformations sat. In are defined. if tional symmetry of the Lagrangian we define ϕs (x. These one-parameter = =e ẋ − (e3s/2 − e−s/2 )gt. t) = ẋ + g ẋ t − g 2 xt. Zt1 d Finding the variational symmetries of a given Lagrangian. tegral on the left-hand side. 0 . t)) = ϕs+u (x. Refs. dt0 e3s/2 dt groups of transformations arise in a natural way in the solu- tion of systems of first-order ODEs (see the examples below). ẋ. Eqs. (9) to L.t + g t x 1 − e4s isfying Eq. in the case of the transforma.g. is not an easy task. (6) without the last term of the right-hand side dt dt 2 ¸ (e. 2 where F (x. (4) we have the integral Zt1 µ ¶ dx0 e−s dx dx = 2s = e−3s ẋ. respectively. For instance. (7) is It may be noticed that the transformations (2) and (4) are de. s) dt. t0 ). according to the transforma- x gt3 s t x0 = − . t0 = te2s . (5) 6 2 for all values of s and u such that both sides of the equation where g is a constant [the constant appearing in Eqs. As we shall show be- low. (8) ϕs (ϕu (x. VARIATIONAL SYMMETRIES OF LAGRANGIANS 141 where g is a constant. µ ¶ 0 ½ ³ ´ ¤ 0 dx 0 0 dt 1 £ −s/2 3 mations (1) is a variational symmetry of a given Lagrangian L x . s). t). t) = (xe−s . from Eqs. t0 = . (6) yields a constant of motion for the ODE given On the other hand. Actually. (2) we have ϕs (xe−u . symmetries for the Lagrangian More precisely. hence. . (6) holds with F = 0. and and it is therefore a constant when one considers curves with x0 = xe−s . (Note that s is a parameter. (2)]. (2)–(4) are examples of local one. 0 . tions (4).

µ ¶µ ¶ 1 ∂ξ ∂ξ ∂G ∂G also Sec.25) of Ref. s) ¯¯ As pointed out above. (9. t. from the Euler–Lagrange equation ∂x ∂ ẋ ∂s dt0 s=0 ∂t µ ¶ µ ¶ d ∂L ∂L ∂ dt0 ∂ dF = . [5] and Eq. [10]). of solutions of Eq. (12) gives rise to a constant of motion. L(x. here x.31) ∂s ∂t ∂s ∂x s=0 dt ∂s s=0 dt of Ref. making use of the chain rule and the definitions Eq. we see that dt ∂x ∂ ẋ ∂t µ ¶ ¯ d ∂L ∂L µ ¶ 0 ∂ 0 0 ∂ 0 ¯¯ = ẋ + ∂ dx0 (dt ) dx − (dx ) dt dt ∂ ẋ ∂t = ∂s ∂s ¯¯ ∂s dt0 s=0 (dt0 )2 ¯ therefore. (9.. and G depend on (x. ẋ. (When G = const. (12) is a partial differential equation for the tion. When G = const.g. Mex. [8]. dη/dt = ∂η/∂t + ẋ ∂η/∂x). the condi.38) of Ref. µ ¶ µ ¶¯ · µ ¶ ¸ ∂ dt0 ∂ ∂t0 ∂t0 ¯¯ d ∂L ∂L = + ẋ η+ξ L− ẋ − G = 0. the only way in For a given Lagrangian.g.. t) = mẋ2 − mgx. [7] and Eq. (12) is also a solution of this η(x. (12) can be written as ¯ s=0 µ ¶ µ ¶ µ 0¶ µ 0 ¶¯ d ∂L ∂L dη dξ ∂x ∂t ¯ η+ − ẋ (dt0 ) d − (dx0 ) d ¯ dt ∂ ẋ ∂ ẋ dt dt ∂s ∂s ¯¯ dη dξ µ ¶ = ¯ = − ẋ . (10) metries of L comes from the fact that. [8]. (10. (13. s) ¯¯ efficients.31) of Ref. t) ≡ ¯ . (11) dt ∂ ẋ ∂x ∂s dt s=0 ∂s dt s=0 and the chain rule..27) of Ref. below). (14) reduces to Eq. with constant co- ∂x0 (x. t) = . As we shall see below.I. For a given function coincides (recall that we are not using the equations of mo- L(x.) Since η. (10. ẋ and t are independent variables). we have Treating the derivative dx0 /dt0 as a quotient of differen- tials. t). each pair of functions ∂s s=0 ξ. In ∂L ∂L ∂ dx0 ∂L η+ + ξ fact.1. C. Eq. ANDRADE MIRÓN. ∂s s=0 equation (see. (12) is a linear operator acting on the functions ξ and ¯ η and. the interest in the variational sym- ξ(x. Indeed. (12) we obtain ∂F ¯¯ G≡ µ ¶ ∂s ¯s=0 ∂η ∂η ∂ξ 2 ∂ξ − mgη + mẋ + ẋ − ẋ − ẋ is some function of (x. 2 ∂t ∂x ∂t ∂x tion G = const. Eqs. The left-hand side of at s = 0. by Rev. t). (22). µ ¶ ¯ ∂ dF d ∂F ¯¯ = . ¯ ∂t0 (x. 13. (6) with respect to s. therefore.158) of Ref. TORRES DEL CASTILLO.142 G. t). AND R. if the Euler–Lagrange equations hold. d ∂L dξ dG (dt0 )2 ¯ dt dt + L− ẋ ξ + L = . e. is a very strong restriction on the symmetries of a given Lagrangian. t) only. that is. ẋ. (12) 1 ∂x ∂ ẋ dt dt ∂t dt dt L(x. (13) + L(x. we obtain that is. ẋ..F. Thus. Eq. (12) determines which this last equation can be identically satisfied is that the the infinitesimal generators (represented by the functions ξ coefficient of each power of ẋ on each side of the equation and η) of variational symmetries of L. to a function with a total derivative with respect to the µ ¶ time equal to zero. two functions ξ and η. Eq. (14) ∂s dt s=0 ∂s ∂t ∂x ¯s=0 dt ∂ ẋ ∂ ẋ µ ¶¯ ¯ ∂ ∂t0 ∂ ∂t0 ¯¯ d ∂t0 ¯¯ dξ thus showing that the expression inside the brackets is a con- = + ẋ ¯ = ¯ = stant of motion (cf. cf. BRAVO ROJAS looking for the infinitesimal generators of such symmetries. Fis. Example ∂s dt s=0 dt ∂s ¯s=0 As a first example.1. we shall consider the standard Lagrangian Thus. 2. similarly. + mẋ2 − mgx + ẋ = + ẋ . t) ≡ ¯ . condition (12) ∂t ∂x ∂t ∂x reduces to Eq. any linear combination. Eq. (15) 2 where ¯ Substituting this last expression into Eq. ξ. µ ¶ ∂L ∂L dη dξ ∂L dξ dG η+ − ẋ + ξ+L = . Eq. e. t. (11) amounts to for a particle of mass m in a uniform gravitational field. ¯ dt ∂ ẋ dt dt s=0 On the other hand. that satisfies Eq. differentiating both sides of Eq. dL ∂L ∂L ∂L = ẋ + ẍ + initions (10). ẋ.7 of Ref. η. Eq. which depend on (x. [3] and Eq. using the elementary rules of differentiation and the def. and. (4. E 59 (2013) 140–147 . t) (recall that. [3].

up to an irrelevant constant term. ∂ξ and. 1 1 1 1 1 ∂t ∂t (19) c1 m xẋt − g ẋt3 − ẋ2 t2 − x2 + gxt2 − g 2 t4 2 2 2 2 8 Equation (16) implies that ξ is some function of t only. (23) X3 ≡ .g. ϕ2 ≡ x − ẋt − gt2 . 2 4 ∂t 2 4 ∂x µ ¶ 1 ∂ ∂ ∂ + c4 mx − mgt2 + c5 (−mgt). d A3 2 d B 3 dA respectively. each of the func- and (19). there are only two functionally independent first in- m = −mg − mg . and ẋ0 on both sides of It may be remarked that. (14)] ∂ξ ∂G µ ¶ −mgη − mgx = . t0 = t + s). one is left with just a one-parameter group of ∂x variational symmetries of the Lagrangian (15). In this case. ∂t ∂x ∂x find that. ∂t 2 ∂x µ ¶ µ ¶ 3 1 ∂ 1 3 2 ∂ + c2 − mgtx + mg 2 t3 X2 ≡ t + x − gt . [6. e. X5 ≡ . e.. which is the ∂η 1 ∂ξ obvious one (x0 = x. (26) we obtain the vector field stituting these expressions into the previous results we obtain · µ ¶ ¸ 2 ∂ 1 3 ∂ ξ = c1 t2 + c2 t + c3 . (25) 2 dt3 dt 2 dt 2 Since A and B are functions of t only. if one assumes that G is equal to the equation we obtain the four equations zero. It may be noticed that the constant of motion = 0. ẋ2 . (18) ∂t ∂x ∂x generator (22) is [see Eq.g. where c4 and c5 are two additional arbitrary constants. (27) 2 ∂t ∂x ∂x Rev. (21) + c5 m(ẋ + gt). ẋ. c3 are arbitrary constants. X = c1 t + xt − gt ∂t 2 ∂x µ ¶ µ ¶ · µ ¶ ¸ 1 3 1 3 2 ∂ 1 3 ∂ η = c1 xt − gt + c2 x − gt + c4 t + c5 (22) + c2 t + x − gt2 2 2 4 ∂t 2 4 ∂x showing that in this case the solution of Eq. ∂ ∂ ξ +η . (22) into Eq. (16) tained above. we have The values of ϕ1 and ϕ2 are the values of ẋ and x at t = 0. (17) ∂x 2 ∂t that L does not depend explicitly on the time. (17) it follows that 1 2 1 2 + c3 m − ẋ − gx + c4 m ẋt − x + gt 2 2 1 dA η= x + B(t). and =− g . Fis. (18) Since the constants c1 . in the terminology of the 2 4 theory of differentiable manifolds (see. Making use Eqs. Using now Eqs. that is. multiplying c3 is minus the total energy. (12) contains ∂ ∂ ∂ five arbitrary constants. or a trivial constant. Refs. though and ∂ 2 G/∂x∂t gives they cannot be functionally independent. and ∂t ∂x This combination is invariant under coordinate transforma- 1 3 B(t) = − c1 gt3 − c2 gt2 + c4 t + c5 . . tegrals. 11]). any constant of motion must be some ∂t2 ∂x ∂t function of. . Mex. X4 ≡ t . for a second-order ∂2η ∂η ∂ξ ODE. µ ¶ 1 3 2 1 2 1 1 2 3 + c2 m xẋ − g ẋt − ẋ t + gxt − g t ξ = A(t) (20) 2 4 2 2 4 µ ¶ µ ¶ and from Eq. (18) and (19) we + c3 + c4 t + c5 . . ϕ1 ≡ ẋ + gt. 1 d3 A d2 B 3 dA 1 x + 2 =− g . VARIATIONAL SYMMETRIES OF LAGRANGIANS 143 equating the coefficients of ẋ3 . ∂η ∂ξ ∂G The constant of motion associated with the infinitesimal m − mgx = . . dt3 dt2 2 dt The functions ξ and η can be conveniently combined in These equations imply that the linear partial differential operator A(t) = c1 t2 + c2 t + c3 . then c1 = c2 = c4 = c5 = 0.. tions and constitutes a vector field. E 59 (2013) 140–147 . Sub. c5 are arbitrary. c2 . (24) 2 dt where B(t) is another function of t only. which is related to the fact − = 0. (26) where c1 . Substituting Eqs. which is a linear combination of the five vector fields µ ¶ µ ¶ 1 3 1 G = c1 mx2 − mgt2 x + mg 2 t4 2 ∂ 1 3 ∂ 2 2 8 X1 ≡ t + xt − gt . instead of the five-parameter family of symmetries ob- = 0. the equality of the partial derivatives ∂ 2 G/∂t∂x tions inside the parentheses is a constant of motion.

(33) can also be written as Inserting this expression into the first equation in (29) we ob- tain the linear equation dM ∂f = −M . then d(M1 /M2 )/dt = 0. ẋ. X3 generates the time variational symmetry of a Lagrangian. t. which means dition). the Lagrangian is determined up to 1 − ts 2(1 − ts)2 Rev. t). t). (30). Substituting the expression for L thus x0 = − . the system of equations (28) takes the form holds for all values of x. treating x and t as parameters (that specify ∂ ∂L ∂ ∂L ∂ ∂L ∂L the initial conditions). How- 1 ever. which is the local group of point transformations given in tors of the variational symmetries of the Lagrangian (15) and Eqs. of the group of point transformations mations corresponding to that infinitesimal generator. ẋ. hence. (13) amounts to η(x0 (x. TORRES DEL CASTILLO. with. ¯ such that the Euler–Lagrange equation (13) be equivalent to ¯ ∂x0 (x. in principle. t. x0 = x. Making use of Eq. t0 (x. and t. that is. Eq.3. C. (31). s=0 ∂s s=0 and therefore we are looking for a function L(x. For example. ant t with a total derivative with respect to (1 − ts)x0 = − + const. s) ¯¯ ¯ Eq. t) such demanding that the equalities hold for all values of s (not that only for s = 0). s=0 ∂s s=0 ¯ ∂ ∂L ∂ ∂L ∂ ∂L ∂L ¯ + ẋ + ẍ = ¯ ∂t0 (x.144 G. in the case of the vector + ẋ + f (x. tions x0 = x + s. t0 ). one-parameter group of transformations whose infinitesimal 2 generator is defined by a given pair of functions η(x. t). (34) µ ¶3 dt ∂ ẋ dx0 t 0 1 t − x =− g This last equation shows that M is defined up to a multi- ds 1 − ts 2 1 − ts plicative constant of motion.. t). x0 = x + st. (27)]. (30) since M1 /M2 can be a trivial constant (i. = ξ(x0 . s) ¯¯ ∂t ∂ ẋ ∂x ∂ ẋ ∂ ẋ ∂ ẋ ∂x ξ(x0 (x. t0 = t. Equation (28) follows from which may correspond to a mechanical system or may have the definition (10) written as some other origin. Fis. it is always possible. t0 ). if M1 and M2 are two (recall that here t is a parameter determining the initial con. (30) (see the example in Sec. X5 generates the transla- rectly to find a first integral [by means of Eq. 2(1 − ts) the time equal to zero as a consequence of Eq. (29) ds 2 ds the partial derivatives of L commute and letting From the second of these last equations (separating variables) we obtain ∂2L 1 M≡ . Taking the partial deriva- dx0 1 dt0 tive with respect to ẋ on both sides of Eq. To this end. a real number) or gt2 a function of x. L(x. (14)]. obtained into Eq. = t02 . s))¯ = ¯ . (28) (30) ds ds with the initial condition (x. ẋ. AND R. we note that Eq. E 59 (2013) 140–147 . t) = 0. (3). also form a basis of a Lie algebra. t. and X2 is the infinitesimal generator out the need to know a one-parameter family of transfor. assuming that = x0 t0 − gt03 . one finds that the one-parameter group the commutator.e. from Eq. which is a first-order linear partial differential equation for 1 − ts M . s). displacements t0 = t + s. below). = η(x0 .F.2. ANDRADE MIRÓN. (33) ∂t ∂x ∂ ẋ ∂ ẋ t t0 = . t) = (31) ∂t ∂ ẋ ∂x ∂ ẋ ∂ ẋ ∂ ẋ ∂x field X1 [see Eq. t) and 2. but this fact will not be elaborated here. ẋ. (34). ẋ. we can use it di. t. ∂ ẋ2 t one obtains and the integration constant is determined by the condition that t0 = t at s = 0. with the bracket given by In a similar way. to find a unique local x0 = xes/2 + gt2 (es/2 − e2s ). The integration constant has to be chosen in such a way that Once we have a solution of Eq. The existence of an infinite number of Lagrangians ξ(x. t. Mex. (32) − 0 = s + const. t0 (x.I.2. −1/t0 = s − 1/t. can find an expression for L. 2. t. ẋ. ẋ. (34). BRAVO ROJAS These vector fields form a basis for the infinitesimal genera. (31). containing two indeterminate x gt3 s functions of x and t. Thus that there is an infinite number of Lagrangians for Eq. t0 = tes . s). (30). All we have to do is to find the solution of the system of first-order ODEs Given a second-order ODE 0 0 dx dt ẍ = f (x. we want to find some function. of transformations generated by X4 is that of the Galilean Once we have obtained the infinitesimal generator of a transformations t0 = t. solutions of Eq. that is ∂M ∂M ∂f ∂M + ẋ +M + f (x. Hence. (32) we x0 = x for s = 0. s))¯ = ¯ .

2 3 dt 2. Starting from the variational symmetries of the Lagrangian (9) are given by the equation of motion ẍ = −g. A nonstandard Lagrangian which reduces to Eq.3. where g and h are some functions of two variables. 2. ∂t ∂x tuting this expression for L and f (x. with Some recent applications of this relationship can be found in. (38) ẋ3 g ẋ2 t 6 L= + + ẋg(x. r du 1 4 As pointed out in Ref. A damped harmonic oscillator 2. M must be a trivial constant [as ξ = 2ct. (This new variable arises in a natural manner dt = = . trary function of the constants of motion (25). function of (x.e. (33) is the equation for the =± u2 − u4 + 4ku. Equation (34) takes the form dM/dt = Eq. t) parameter group of transformations generated by (37). ∂ 2 L/∂ ẋ2 = t2 . t)ẋ + h(x. c = 1 is the one given by Eqs. and the references cited therein. E 59 (2013) 140–147 .g.e. 2. from Eq. η = −γx/2. the Thus. t) is an arbitrary function of two variables and. L= e (ẋ − ω 2 x2 ).) (Cf. (9) if Φ = const. (34) we have dM/dt = 0. also Ref. and where g(x.. (39) 2 ẍ + ẋ + x5 = 0. [see Eq. t2 2 x6 dΦ(x. in place of the second-order ODE (35).. [4]. t) + h(x. Example. therefore. Mex. Choosing M = ẋ + gt we find 1 − c(3t2 xẋ + t3 x6 + 3t3 ẋ2 ) = const. t). VARIATIONAL SYMMETRIES OF LAGRANGIANS 145 the total derivative with respect to the time of an arbitrary where k is a constant. (34) takes the form M γ and we can choose M = eγt . Eq. t) are some functions of two variables. Example. t) L= ẋ − + . we can choose g = 0 = . t). ẋ. (4). (35) which corresponds to a damped harmonic oscillator (here γ t and ω are constants). t). t) = −2ẋ/t − x5 into Choosing g = 0 and h = −eγt ω 2 x2 /2. we can choose M = t2 . 1 2 2 The Lagrangian (40) reproduces the ODE (39) if and only if L= t ẋ + g(x. L= e ẋ + g(x. dt 2t3 Rev. we find the Lagrangian µ ¶ which leaves the Lagrangian (41) invariant. x0 = xe−γs/2 . in order dx −t 2 x ± t4 x2 − 43 t6 x6 + 4kt3 to reproduce the equation of motion.. (14)]. (37) in the case of the standard Lagrangian (15)]. Eq. (31) we obtain known Lagrangian ∂g ∂h 1 γt 2 − x5 t2 = . we find that all a particle in a uniform gravitational field.1. t) x6 t2 ∂Φ(x.1.2. h+ = . Jacobi last multiplier of the system of equations dt t 3 dx dẋ where u ≡ x2 t. [12–14]. As a final example. Therefore.2. vector field ∂ x ∂ X= −γ ∂Φ(x. One of these ∂t ∂x 6 is given by ξ = 1. (36) 2 ∂g ∂h −eγt ω 2 x + = . ẋ.2. In fact. [15]. and h = −g 2 xt. t0 = t + s. t)ẋ + h(x. Substitut- q ing L into the Euler–Lagrange equation we see that. Hence dM 2 1 γt 2 =M . (36). equivalently. substituting into Eq. Example.. Thus. with G = const. t) and h(x. (41) ∂t ∂x 2 which can be written as One finds that the infinitesimal generators of the varia- µ ¶ tional symmetries of this Lagrangian [i. 6 2 we have the first-order ODE (38). which amounts to where g and h are two functions to be determined. Refs. we find a nonstandard Lagrangian for Following the steps presented in Sec. in this way we obtain the Lagrangian (8). and G = const. i. from (ξ ∂/∂t+η ∂/∂x)u = 0. or an arbi- where c is an arbitrary constant. η = −cx. Fis.) The one- ẋ f (x. The Emden–Fowler equation Another illustrative example is given by the equation In the case of the Emden–Fowler equation ẍ + γ ẋ + ω 2 x = 0. the solutions of ∂g ∂ x6 t2 = h+ . or. ∂x 6 ∂t generates the one-parameter group of point transformations where Φ(x. t) ∂t 2 ∂x g= . we obtain the well- Eq. (40) dt t 2 hence.2. e. Substi. (12)] form a five-dimensional vector space. t).

ẋ.t Eq. is a constant of motion. in this section we only give the definitions and main e. not every system of two or more Since the derivations are almost identical to those presented second-order ODEs can be derived from a Lagrangian (see. q̇i . ẏ 2 . is a variational symmetry of the Lagrangian L(qi . . t0 ). we have a system of n second-order ODEs derivable 3. 0) = t. Fis. . (49) ∂t dt dt ∂t ∂x 2 ∂y for some function G(qi . (45) X8 ≡ t + xt + yt − gt . (47) into dq 0 dt dqi L qi0 . dt dt dt ẋẏ.I. 2. t0 = t0 (q1 . ODEs.1. t. Equation (45) is a partial differential equation for n + 1 functions of n + 1 variables. Final remarks X n ∂ ∂ ξ + ηi . have different variational symmetries. qn . on both d sides of the equation. or ∂t i=1 ∂qi (42)] that map any solution of an ODE. This group is determined by the are variational symmetries of the Lagrangian leading to that system of first-order ODEs ODE or system of ODEs. t0 = L qi . ẏ. s). X3 ≡ . Example from a Lagrangian L(qi . and X3 generate translations along the x. which reduces to a component of the angu- lar momentum when g = 0. grees of freedom By contrast with the case of a single second-order ODE. (43) one finds that the functions (44) correspond to µ ¶ 1 2 ∂ ∂ the infinitesimal generator of a variational symmetry of L if X6 ≡ gt + y −x . . i=1 ∂qi ∂ q̇i dt dt ∂t 2 ∂x 2 4 ∂y µ ¶ ∂L dξ dG 2 ∂ ∂ 1 3 ∂ + ξ+L = .. X2 ≡ . y. ∂ ∂ ∂s s=0 X4 ≡ t . and the terms that do not contain ẋ or ẏ. Systems with an arbitrary number of de. t) if which is the standard Lagrangian for a particle of mass m µ ¶ 0 µ ¶ in a uniform gravitational field. ẋ2 ẏ. (45) it follows that is especially interesting because in the limit g = 0 it gener- n à n ! ates rotations about the origin in the xy plane. 0i . For instance. (48) ¯ ∂x ∂y ∂t0 (qi . C. [8] and the references cited therein). . of degrees of freedom or. . n. t0 (0) = t. .146 G. t). considered in Sec. The one-parameter family of point transformations A simple example is given by the Lagrangian qi0 = qi0 (q1 . ¯ ∂x ∂y ∂t ∂qi0 (qj . qn . 2 ∂x ∂y Xn · µ ¶¸ µ ¶ ∂L ∂L dηi dξ ∂ x ∂ y 3 2 ∂ ηi + − q̇i X7 ≡ t + + − gt . X6 from Eq. one finds that the infinitesimal genera- + F (qi . . the ODE ds ds Rev. t) ≡ ¯ . whose solution yields a vector field 4. Making use of the Euler–Lagrange equations The vector fields X1 . If one looks for all the point transformations [Eqs. results applicable to the case of a mechanical system with an arbitrary number. . t). qn . in Sec. ANDRADE MIRÓN. t. or of a system of which is the infinitesimal generator of a local group of vari. Moreover. . t. . X5 ≡ t . with the initial condition qi0 (0) = qi . Assuming that bination of the eight vector fields qi0 (q1 . q̇i . . s) ¯¯ ηi (qj . . t0 ). t) ≡ ¯ . ẋẏ 2 . respectively. = ξ(qj0 . t. (43) tor of any variational symmetry of (47) must be a linear com- dt where F is some function. more gener- ally. . (44) ∂s s=0 and from Eq. TORRES DEL CASTILLO.2. and t axes. X2 . 2. s). s). (42) m 2 L= (ẋ + ẏ 2 ) − mgy. t. . qn . t. 0) = qi and t0 (q1 . different Lagrangians dqi0 dt0 corresponding to the same ODE or system of ODEs may = ηi (qj0 . (47) 2 i = 1. t. into another solution.g. E 59 (2013) 140–147 . . X4 and X5 generate d ∂L ∂L Galilean transformations. dt ∂ q̇i ∂qi in (49) correspond to symmetries that are not obvious. Substituting Eq. . . 2. ẋ2 . BRAVO ROJAS 3. AND R. equating the coefficients of ẋ3 . . for all s. one finds that not all of them ational symmetries of L. while the other three vector fields − = 0. (1). . ẏ 3 . . . with ∂ ∂ ∂ the aid of the definitions X1 ≡ . the constant X ∂L X ∂L of motion associated with this symmetry is m(y ẋ − xẏ) − ηi + ξ L − q̇i − G (46) i=1 ∂ q̇i i=1 ∂ q˙i mg(tx − t2 ẋ/2).F. Mex. Ref. s) ¯¯ ξ(qi . (45). n.

(Springer-Verlag. (1998) 245. New York.M. Acknowledgment ists a first integral associated with each of them. 2012). eral point symmetries of the equation or system of equations.W. 18. where g is a constant.P. 9. 1989). 2000). Bluman and S. 4th ed. differentiable manifolds. Goldstein. San Francisco. metries (Cambridge University Press. E.B.) wish to thank functionally independent. [17]. P. Stephani. Kumei.R. 1993). Arnold. London. (14) and (46)].C. Symmetries and Differential Equa- 2nd ed.I. 44 (2003) 2111. the variational symmetries may not be the more gen. V. J. use of Lie group analysis. Phys. Calkin.F. A Treatise on the Analytical Dynamics of Par. cal Physics Approach (Birkhäuser Science. Torres del Castillo. 7. 17. W. 6. Symmetry Methods for Differential Equations: A 16. Nucci.T. 11.L. the Vicerrectorı́a de Investigación y Estudios de Posgrado of grals are associated with variational symmetries. 4.A. and J. 116 (2011) 5. making use of the language of ODEs. 3rd ed. which lead to this equation. Phys. M. Math. (2009) 431. Lagrangian and Hamiltonian Mechanics (World 10. P. not all first inte. Cambridge. Guha and A.I.C. New York.C. C. Classical Mechan. 179. as we have The authors wish to thank the referee for helpful comments. B. Mathematical Methods of Classical Mechanics. vector fields and differential forms. On the other hand. P. VARIATIONAL SYMMETRIES OF LAGRANGIANS 147 ẍ = −g. 2. 2. re. Vijayakumar. 1990). SIAM Rev. Differentiable Manifolds: A Theoreti- ics. H. van Brunt. Austral. Applications of Lie Groups to Differential Equations.L. though.P. J. Acta Appl.E. In Ref. New York. Phys. 1996). Ghose Choudhury. admit five-parameter equations.G. making and three-parameter groups of variational symmetries. Cantrijn. H. Nucci. (Addison-Wesley. the first integrals obtained in this manner need not be Two of the authors (C. 15. Jnl of Applied Mathematics 9 Beginner’s Guide (Cambridge University Press. 8. spectively. The Hamilton–Jacobi Theory in the Calculus of Vari- Scientific. 380 (2012) 012008. while the Lagrangians (15) the aid of the so-called adoint symmetries of the system of and (8). M. [16]. Chap. Jr. M.G. G. (Springer-Verlag. 16 ticles and Rigid Bodies. 2004). possesses an eight-parameter systems of EDOs. Hydon. Anco and G. the variational symmetries are very useful because there ex. Whittaker. J. Leach. Bluman. Soc. 13. without making use of a Lagrangian. Chap. 1989). Ser. 14. Math. tions (Springer-Verlag. M. J.: Conf. Sarlet and F. Differential Equations: Their Solution Using Sym. 2000). B 32 2nd ed. Marcelli and M. Bhutani and K. Nonlinear Math.C. Cambridge. (1991) 457. Safko. ations (Van Nostrand. 2002). Rund. the Universidad Autónoma de Puebla for financial support it is shown that it is possible to find first integrals of EDOs or through the program “Jóvenes Investigadores. VII. New York. S. for a given ODE or system of the basic results given here. Cambridge. 23 (1981) 467. 3. Olver. Poole. Euro. and R. Rev. G. Nucci and P. which can be readily calculated [Eqs. Singapore. The review paper [18] presents various generalizations of In spite of the fact that. with group of point symmetries [7]. E 59 (2013) 140–147 .J. Chap. Mex. (Cambridge University Press. The Calculus of Variations (Springer-Verlag. A similar result is presented in Ref. H. Fis. O.” 1. New York. seen. 12. Math. X. 1966).