Академический Документы
Профессиональный Документы
Культура Документы
Sergio A. Hojman Citation: J. Math. Phys. 34, 2968 (1993); doi: 10.1063/1.530109 View online: http://dx.doi.org/10.1063/1.530109 View Table of Contents: http://jmp.aip.org/resource/1/JMAPAQ/v34/i7 Published by the American Institute of Physics.
Related Articles
Classical and quantum probabilities as truth values J. Math. Phys. 53, 032101 (2012) Time-independent description of rapidly driven systems in the presence of friction: Multiple scale perturbation approach Chaos 22, 013131 (2012) Limitations on cloning in classical mechanics J. Math. Phys. 53, 012902 (2012) Equivalence problem for the orthogonal webs on the 3-sphere J. Math. Phys. 52, 053509 (2011) A Foucaults pendulum design Rev. Sci. Instrum. 81, 115102 (2010)
Downloaded 26 Oct 2012 to 132.248.7.10. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
based
Departamento de Fisica, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile
19 February 1993)
A new (non-Noetherian) symmetry transformation for the small oscillations problem is exhibited. A novel way to solve this classical problem, based on conservation laws constructed using the new symmetry, is presented. Applications of these results to other fields of physics are outlined.
I. INTRODUCTION A new way of solving the classical small oscillations problem is presented. The approach is based in the construction of new conserved quantities besides the total energy. This construction relies on the discovery of a non-Noetherian symmetry for the system.2V3Loosely speaking, a non-Noetherian symmetry is a symmetry of the equations of motion which does not leave the variational principle invariant, i.e., it does not satisfy Noether theorem. The method s presented seems especially appropriate to deal with systems with a large number of degrees of freedom as well as to provide new conservation laws in field theory. Let us now briefly consider the standard solution of the problem. The small oscillations problem is described by the regular Lagrangian
L=$Tij4 $-fV&q , i,j=l,...,n (1.1)
with
Tij = Tji 9 Vij = Vii )
(1.2)
and
The usual solution, described in detail in many textbooks, considers the eigenvalue problem for the frequencies det(--02Tij+
Vii) ~0,
(1.4)
as a starting point. Then, the solution to the eigenvector problem, ( -fOiTij+ Vij)A~=O (1.5)
for each eigenfrequency w,,, provides all the information needed to construct the normal coordinates which completely decouple the original problem. To simplify the discussion we assume that the frequencies are not degenerate. The normal coordinates Q, which may be defined by Qo=AiTijqj U= l,...,n (1.6)
0022-2488/93/34(7)/2968/7/$6.00 @ 1993 American institute of Physics
2968
Downloaded 26 Oct 2012 to 132.248.7.10. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
symmetries in mechanics
2969
i.e., n uncoupled harmonic oscillator equations, one for each eigenfrequency o,, completing in this way the solution of the problem.
PROBLEM He (2.1)
The Lagrangian ( 1.1) is time independent and, therefore, the Hamiltonian H,=$( T- )ijpipj+;vijq is conserved. The momenta pi are defined in the usual fashion
and ( T- ) is the inverse of Tij [which exists because the Lagrangian ( 1.1) is regular, Eq. j (1.3)]. Of course, Ho may also be written in terms of the velocities 0 as (2.3) The equations of motion for the Lagrangian ( 1.1) are T,,+j + Vii&o or
qf=Fi(qi,gj,t), Fi= - ( T-l)iivjk4k.
(2.4)
(2.5)
transformation
t =t
q i(t)=qi(t)+6q (t),
(2.6)
with ~qi=E(T-l)iiv.Mk
J
(2.7)
maps solutions qi( t) of Eqs. (2.4) in solutions q t) of the same equations. In other words, the i( transformation (2.7) is a symmetry transformation for Eqs. (2.4). Nevertheless, transformations (2.7) are non-Noetherian for the Lagrangian ( 1.1) in the sense that in spite of mapping the space of solutions of Eqs. (2.4) into itself, they do not satisfy Noether theorem for s Lagrangian ( 1.1) . Given a non-Noetherian symmetry, several conservation laws may be constructed.2T3 In particular, if a constant of motion C= C(q ,$,t) for the problem in question is known, then its deformation SC along the symmetry transformation Sqi,
ac
=~,sd+@;rt I
acJ
(Sq ),
(2.8)
with
2970
symmetries in mechanics
(2.9) is also a constant of the motion.3 This result may be considered a generalization of Poisson s theorem for the case of non-Noetherian symmetry transformations. For the small oscillations problem, we may deform Ho along Sq given by Fq. (2.7). We get that
Hl=tiViJ#+ Vij( T- )jkVklQi&
is conserved as it can be readily checked. We may now consider expression (2.8) for H, with the same transformation
H2
Hz=f[
(VT- V)iJ@+
(2.12)
for any positive integer s is conserved. Of course, at most n of these constants of motion are functionally independent due to Cayley-Hamilton theorem. It is straightforward to show that they are in involution, that is, that their Poisson brackets vanish, i.e.,
[H,,H,,] =0 tfs,s +
(2.13)
Therefore, the system (2.4) is completely integrable according to Liouville theorem.4 Coms plete integrability is, of course, a weaker result than the general solution outlined in the Introduction. Fortunately, the general solution may also be achieved with the results obtained here. Complete integrability may, on the other hand, be a most welcome result in field theory, for instance, which we will discuss briefly in the last section of this article.
Ill. COMPLETE SOLUTION OF THE PROBLEM
To get the general solution of the problem amounts essentially to display the expressions for the normal coordinates Q, in terms of the original coordinates q once the eigenfrequencies 0, are known. We start then by solving the eigenvalue problem for the frequencies det( --W Tij+
Vii) =O (3.1)
as it is done in the classical solution. We proceed by computing the n conserved quantities Ho and H,
Ho =$ ( TiJ# H,=i[(V( T-*V)-)ijQ #+(V( + VijQIq ) 9 T- V)s)ijg#], s= l,...,n-
(3.2) 1.
(3.3)
Let us now take a shortcut to get the answer. Consider the energies E, associated to each of the normal coordinates Q, E,=~(&)+~coiw ,e, (no sum on a).
(3.4)
Downloaded 26 Oct 2012 to 132.248.7.10. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
symmetries in mechanics
2971
It is a straightforward
(3.6)
where t,, are defined as the diagonal components (in the eigenvectors basis) of the kinetic energy matrix
t$~(r~=A~TijA$ e
(3.7)
(3.8.1)
(3.8.2)
(W~)-1t,EI+(W~)-1t2E2+...+(O~)- E,=H,_I,
(3.8n)
1
co2 2 (w;)-l
...
*** ... 4 (wy
Ho HI
: FI, _,
(3.9)
The matrix V,, which appears in Eq. (3.9) is a Vandermonde matrix in n dimensions. Its determinant is well known to be A,,= fl
j>i
(c+w).
(3.10)
(3.11)
where the matrix C is described in what follows. Let us construct the last column C/n of the cofactors matrix C. One gets
Ci,g=In+l-iAn-l(wi),
(3.12)
where A,-, (ai) is the determinant of an (n - 1) dimensional Vandermonde matrix with wi excluded and li is defined as the coefficient of a in the (n - 1)th degree polynomial - P,+ 1(d,wz> where wi is not present (3.13)
2972
symmetries in mechanics
This result is easily verified by first constructing C,,, (which is just the Vandermonde determinant defined by excluding the last column and the last row of V,) proceeding with C,- tn and so on. The expressions for the rest of the columns of the matrix C can now be easily computed. Therefore, the energies 17&for each mode can be easily obtained by solving Eq. (3.9), i.e.,
(3.14)
with ( Vn)- given by Eqs. (3.11), (3.12), and (3.13). Knowledge of En is equivalent to having the expressions for Q, and therefore the complete solution to the problem. We will illustrate the method with an example.
IV. EXAMPLE
Consider a system of two particles of mass m described by the Lagrangian, L=fm(&+&) so that
Tij=m
-%k[(qf+&
+K(q,-q2)21
(4.1)
(4.2)
Vij =
(4.3)
has solutions
k CO:=-, m 2 k+2K fO2=m (4.5)
(4.6)
Downloaded 26 Oct 2012 to 132.248.7.10. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
symmetries in mechanics
2973
(4.8)
(4.9)
t2E2 =
--;&+HI
o;-w:
(4.10)
Ql=ql+a,
and
(4.13)
Qz=ql-qz,
with t1=t2=;.
(4.14)
(4.15)
We have completely solved the small oscillations problem using the results described in Sets. II and III.
V. CONCLUSIONS AND OUTLOOK
We have found a new way of solving the classical small oscillations problem. This new solution seems to be especially appropriate to handle problems with a large number of degrees of freedom because the solution can be readily expressed in general by means of Eqs. (3.11)) (3.12), (3.13), and (3.14) once the eigenfrequencies w, are known. The usual approach which requires the solution of n eigenvector problems, Eqs. ( 1.5), may become too involved for large n. Our method requires the inversion of a single matrix Tij , which is usually a very simple task. All other operations needed to solve the problem (matrix multiplication, polynomial computation) seem simpler than solving Eq. ( 1.5) for each eigenfrequency w, for large n. Once the energies E, are constructed, the normal coordinates Q, can be easily determined. The method may, of course, also be tried for the mathematical problem of simultaneous diagonalization of two symmetric matrices provided one of them be regular. The application of the approach to field theory may prove useful, in particular to provide infinitely many conservation laws for the dynamics of scalar fields in vacuum as well as in the presence of background gravitational fields. The method has also been successfully applied to getting new solutions of the problem of the motion of strings in a gravitational field background.6
2974
symmetries in mechanics
ACKNOWLEDGMENTS
The author would like to thank Fundaci6n Andes (Chile) for a fellowship. Partial support for this work has been provided for by Grant No. 9 l-0857 (Fond0 Nacioml de Ciencia y Tecnologia, Chile).
D. Landau and E. M. Lifschitz, Mechanics (Pergamon, Oxford, 1976); H. Goldstein, Classical Mechanics L. (Addison-Wesley, Reading, 1980). 2S. Hojman, J. Phys. A 17, 2399 (1984). Hojman, L. Nfiiiez, A. Patiiio, and H. Rago, J. Math. Phys. 27, 281 (1986); D. Del Castillo-Negrete and S. S. Hojman, ibid. 31, 2211 ( 1990), and references therein. 4E. T. Whittaker, A Treatise on the Analytical Dynamics of Particles and Rigid Bodies (Cambridge University, Cambridge, 1937). A. Hojman, J. de la Jara, and L. Peiia (in preparation). S. 6S. A. Hojman and J. de la Jara (in preparation).