Hamilton-Jacobi theory for one dimensional 5 0 0 autonomous systems using parametric 2 n a transformations J 2 2 1 v G. Gonz´alez ∗, 1 6 Departamento de Matem´aticas y F´ısica 0 1 0 I.T.E.S.O. 5 0 / h Perif´erico Sur # 8585 C.P. 45090 p - h Guadalajara, Jalisco, M´exico. t a m : v i Abstract X r a A necessary and sufficient condition for a parameter transformation that leaves invariant the energy of a one dimensional autonomous system is ob- tained. Using a parameter transformation the Hamilton-Jacobi equation is solved by a quadrature. An example of this approach is given. Keywords: Parameter transformation, Lagrangian, Hamiltonian, Hamilton- Jacobi equation, autonomous system. PACS:45.20.Jj ∗E-mail: gabriel−[email protected] 1 1 Introduction It is well known that the problem of integrating Hamilton’s canonical equations of motion, which are expressed as a system of ordinary differential equations of the first order, can be replaced by an equivalent problem of finding a complete solu- tion of a single non linear partial differential equation of the first order, called the Hamilton-Jacobi equation [4]. If we succeed in finding a complete solution of the Hamilton-Jacobi partial differential equation, the motion of the dynamical system can be easily found directly from this complete solution by using only elementary algebraic operations [9]. The Hamilton-Jacobi equation can be considered the most elegant and powerful method known for finding the general solution to the mechanical equations of mo- tion and gives an important physical example of the deep connection between first order partial differential equations and first order ordinary differential equations [8]. Each complete integral of the Hamilton-Jacobi equation gives rise to a family of solutions of Hamilton’s equations and is the generating function of the canonical transformation which maps the dynamical system to a trivial one with vanishing Hamiltonian. Apart from its practical aspects, the Hamilton-Jacobi theory leads to a geometric picture of dynamics relating the dynamics to wave motion and has been the starting point for Schr¨odinger to state the wave equation in quantum mechanics [2]. The applicability of using the Hamilton-Jacobi procedure depends in our ability to solve a first order non linear partial differential equation which can be very compli- cated to solve in some cases [7]. The main purpose of this article is to show that, under certain conditions, it is possible to introduce a parameter transformation 2 which allow us to obtain a complete integral for the Hamilton-Jacobi equation for one dimensional autonomous systems without having to solve the partial differential equation given by the Hamilton-Jacobi method. 2 Parameter transformations and the Hamilton- Jacobi equation Newton’sequationofmotionforonedimensionalautonomoussystemscanbewritten as the following dynamical system dx dv = v, = F(x,v), (1) dt dt where x is the position of the particle, v is the velocity and F(x,v) is the force divided by the mass of the particle. It is well known that the Lagrangian L(x,v) for a one dimensional system always exist [3] thus we may obtain (1) from the Euler-Lagrange equation [4] d ∂L ∂L = , (2) dt ∂v! ∂x and the associated constant of motion for (1) from the Legendre transformation [4] ∂L K(x,v) = v L, (3) ∂v − the numerical value of (3) will be referred as the energy of the system. Knowing the constant of motion of (1) we can obtain the associated Hamiltonian by expressing the constant of motion in terms of the canonical variables H(x,p) = K(x,v(x,p)) and using this Hamiltonian we obtain the Hamitlon-Jacobi equation ∂S ∂S H x, + = 0, (4) ∂x! ∂t 3 where S is known as Hamitlon’s principal function [4]. Let us now consider the following parameter transformation of the form τ = τ(t), x˜ = x(t), (5) which does not affect the position variable and it will be assumed that the functions τ(t) are of class C2 such that dτ/dt > 0. The problem is to find a parameter transformation which leaves invariant the energy of the system when subject to this type of transformation. To do this let L(x,v) and L˜(x˜,v˜) be the Lagrangians for (1) before and after the parameter transformation respectively. Assume that ∂2L/∂v2 and ∂2L˜/∂v˜2 do not vanish or become infinite in some region R of the dynamical space then we have the following Theorem 1 A necessary and sufficient condition for a parameter transformation to leave invariant the energy of a one dimensional autonomous system is that ∂2Ldv ∂2L˜ dv˜ = . ∂v2 dt ∂v˜2 dτ Proof 1 The Euler-Lagrange equation for L(x,v) may be written as ∂2L ∂2Ldv ∂L v + = ∂x∂v ∂v2 dt ∂x similarly the Euler-Lagrange equation for L˜(x˜,v˜) is given by ∂2L˜ ∂2L˜ dv˜ ∂L˜ v˜+ = , ∂x˜∂v˜ ∂v˜2 dτ ∂x˜ by assumption of the theorem we have ∂L ∂2L ∂L˜ ∂2L˜ v = v˜ ∂x − ∂x∂v ∂x˜ − ∂x˜∂v˜ 4 therefore ∂ ∂L ∂ ∂L˜ ˜ L v = L v˜ , ∂x − ∂v! ∂x˜ − ∂v˜! which implies ∂K(x,v) ∂K˜(x˜,v˜) = . ∂x ∂x˜ Taking the total time derivative of K˜(x˜,v˜) we obtain dK˜ ∂K˜ dx˜dτ ∂K˜ dv˜dτ = + , dt ∂x˜ dτ dt ∂v˜ dτ dt using the fact that ∂K˜/∂v˜ = (∂2L˜/∂v˜2)v˜ then dK˜ dx˜dτ ∂K˜ ∂2L˜ dv˜ = + dt dτ dt ∂x˜ ∂v˜2 dτ! but dx/dt = (dx˜/dτ)(dτ/dt), ∂K/∂x = ∂K˜/∂x˜ and∂2L/∂v2(dv/dt) = ∂2L˜/∂v˜2(dv˜/dτ) then dK˜ dK = dt dt which implies that K(x,v) = K˜(x˜,v˜) and completes the proof of the theorem. The main advantage of using this type of parameter transformation is that neither the position or the energy of the system are affected, using this fact and taking into account that we are dealing with an autonomous system we can express the solution to the Hamilton-Jacobi equation as [4] S = W(x) Et, S˜ = W˜ (x) Eτ, (6) − − ˜ where S and S are Hamilton’s principal functions for (1) before and after the parameter transformation respectively and W and W˜ represents the solution to a nonlinear first order partial differential equation given by [4] dW dW˜ H x, = E, H˜ x, = E, (7) dx ! dx ! 5 which is usually known as Hamilton’s characteristic function. If we know how the generalized linear momentum transforms under (5) and recalling that dW dW˜ = p, = p˜, (8) dx dx we may express Hamilton’s principal function for (1) as S = p(x,p˜(x,E))dx Et. (9) − Z Therefore, we obtain the solution for the Hamilton-Jacobi equation by a quadrature. This way of solving the Hamilton-Jacobi equation is useful to solve complicated problems as it would be shown in the following example. 3 Example Consider a relativistic particle of mass at rest m under the action of a constant force λ > 0 and immersed in a medium that exerts some type of friction which is proportional to the square of the velocity. The classical equation of motion for this system is given by dv m = (λ γv2)(1 v2/c2)3/2, (10) dt − − where γ is a positive real parameter and c represents the speed of light. Writing (10) at first order of approximation in v2/c2 we have dv m = (λ γv2)(1 α2v2), (11) dt − − where α2 = 3 .The Lagrangian associated to this system is [5] 2c2 mvtanh 1(αv) mλ L(x,v) = − e 2x(λα2 γ)/m e 2x(λα2 γ)/m 1 , (12) − − − − 2α − 2(λα2 γ) − − (cid:16) (cid:17) 6 and the Hamiltonian of this system is given by [5] m mλ H(x,p) = e 2x(λα2 γ)/m ∞ α2nv2n+2(x,p)+ e 2x(λα2 γ)/m 1 , − − − − 2 2(λα2 γ) − nX=0 − (cid:16) (cid:17) (13) where v2n+2(x,p) is given by n (2n+2)/(2n+1) p 2n+1 2λx v2n+2(x,p) = e 2γx/m . (14) − m (n+1)! m ! ! The Hamiltonian (13) is valid for the case αv < 1 and has physical meaning only | | when p > 0 and x > 0. Once knowing the Hamiltonian we can obtain the Hamilton- Jacobi equation for the system n (2n+2)/(2n+1) m ∂S 2n+1 2λx e 2x(λα2 γ)/m ∞ α2n e 2γx/m + − − − 2 ∂x m(n+1)! m ! ! n=0 X mλ ∂S e 2x(λα2 γ)/m 1 + = 0, (15) − − 2(λα2 γ) − ∂t − (cid:16) (cid:17) therefore one has to solve (15) to obtain Hamilton’s principal function, which at first sight may seem like a formidable task, but it can be done applying the approach described in the last section. Consider the following parameter transformation t τ = √1 α2v2dt, x˜ = x, (16) − Z0 therefore equation (11) transforms into dv˜ m = λ+γ˜v˜2, (17) dτ where γ˜ = λα2 γ and the transformation equations between one set of dynamical − variables (x,v) to the other set of dynamical variables (x˜,v˜) are given by x = x˜ x˜ = x (18) v = v˜ v˜ = v √1+α2v˜2 √1 α2v2 − 7 The Lagrangian and the Hamiltonian for (17) can be obtained, and are given by [6] m mλ L˜(x,v˜) = v˜2e 2γ˜x/m e 2γ˜x/m 1 , (19) − − 2 − 2γ˜ − (cid:16) (cid:17) p˜2 mλ H˜(x,p˜) = e2γ˜x/m + e 2γ˜x/m 1 , (20) − 2m 2γ˜ − (cid:16) (cid:17) where we have dropped the tilde for the position variable for the sake of simplicity. Using (10),(12), (17) and (19) it is easy to convince oneself that theorem 1 is fulfill, therefore the energy is invariant under transformation (16). What we need now is to find how the generalized linear momentum transforms under (16), to do that we express p˜= mv˜exp( 2γ˜x/m) in terms of v using (18), which gives − mve 2γ˜x/m (2n)!α2nv2n+1 p˜= mv˜e 2γ˜x/m = − = me 2γ˜x/m ∞ , (21) − − √1 α2v2 22n(n!)2 − nX=0 where we have used the fact that αv < 1 in the last step. Substituting (14) into | | (21) we have the way the generalized linear momentum transforms under (16) (2n+1)(2n)! λα2x n p˜= pe 2λα2x/m ∞ = pe λα2x/m I (λα2x/m)+I (λα2x/m) , − 2n(n+1)!(n!)2 m ! − 0 1 nX=0 (cid:16) (cid:17) (22) where I (z) is the modified Bessel function of the first kind [1] . Using the fact that n m2λ p˜= e γ˜x/m 2mE (e 2γ˜x/m 1), (23) − − s − γ˜ − and substituting (23) into (22) we have eγx/m 2mE m2λ (e 2(λα2 γ)x/m 1) p(x,p˜(x,E)) = − λα2−γ − − − , (24) qI (λα2x/m)+I (λα2x/m) 0 1 substituting (24) into (9) we obtain Hamilton’s principal function for x > 0 eγx/m 2mE m2λ (e 2(λα2 γ)x/m 1) S = − λα2−γ − − − dx Et. (25) qI (λα2x/m)+I (λα2x/m) − Z 0 1 All theexpressions derived inthis paper have the right limit when γ 0 andα 0. → → 8 4 Conclusions A necessary and sufficient condition for a parameter transformation that leaves in- varianttheenergyofaonedimensionalautonomoussystem wasobtained. Amethod for solving the Hamilton-Jacobi equation using a parameter transformation was de- duced. All the expressions obtained in this paper converge to the conservative case when the dissipation parameter goes to zero. 9 References [1] Andrews C. Larry, (1998). Special functions of mathematics for engineers, Ox- ford Univeristy Press. [2] Cook B. David, (1988). Schr¨odinger’s Mechanics, World Scientific Lecture Notes in Physics, Vol. 28. [3] Darboux,G.(1984).Lecons sur la th´eorieg´en´eraldes surfaces etles applications g´eometriques du calcul infinit´esimal, Gauthier-Villars, Paris, IVi´eme partie. [4] Goldstein H., (1980). Classical Mechanics, Addison-Wesley. [5] Gonz´alez, G. (2004). To appear in the Internation Journal of Theoretical Physics. [6] L´opez, G., Gonz´alez, G. (2003). Il Nuovo Cimento B 118, 1. [7] L´opez, Gustavo, (1999).Partial Differential Equations of First Order and Their Applications to Physics, World Scientific. [8] Vilasi, Gaetano,(2001). Hamiltonian Dynamics, World Scientific. [9] Vujanovic, B. D., Jones, S. E. (1989). Variational methods in nonconservative phenomena, Academic Press, Inc. 10