Home Search Collections Journals About Contact us My IOPscience
Hidden and contact symmetries of ordinary differential equations
This content has been downloaded from IOPscience. Please scroll down to see the full text.
1995 J. Phys. A: Math. Gen. 28 6707
(http://iopscience.iop.org/0305-4470/28/23/020)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 193.140.28.22
This content was downloaded on 13/11/2014 at 12:44
Please note that terms and conditions apply.
I. Phys. A. Math. Gem.28 (1995) 6707-6716. Printed in the UK
Hidden and contact symmetries of ordinary differential
equations
Barbara Abraham-Shraunert, P G L Leachtlly, K S Govindertll and Gail
Ratcliff§
t Department of Electrical Eogineering, Washington University, St Louis, MO 63130,USA
J Department of Mathematics, University of the Aegean, Karlovassi 83200. Sa", Greece
5 Department of Mathematics and Computer Science, University of Missouri-St Louis, St Louis,
MO 631214499. USA .
Received 3 July 1995
Abstract. The Lie algebra for the maximal contact symmetries of third-order ordinary
differential equations (ODES)is examined for type I and I1 hidden symmetries where- the analysis
of hidden symmetries for point sy"e(ries is extended ta contact symmetries. ODES invariant
under the group associated with the ten-dimensional (maximal) Lie algebra may produce type I
hidden symmetries for two-parameter subgroups and type I1 hidden symmetries for certain
solvable non-Abelian three-parameter subgroups in the Ihird-order ODESwhen they are reduced
in order A new class of type I1 hidden symmetries is recognized in which contact symmetries
uansform to point symmetries in some reduction paths. Two examples of ODES hvariant under
subgroups of the ten-parameter group under which yJ" = 0 is invariant demonstrate the new
class of type I1 hidden symmetries.
1. Introduction
Hidden symmetries of ordinary differential equations have recently been introduced [14]
and analysed for various physical and mathematical equations [5-16]. These are symmetries
not found by the Lie classical method for point symmetries of differential equations [l, 17-
201. Type I @) hidden symmetries of ordinary differential equations (ODES) have been
defined as point symmetries that are lost (are gained) when the order of the ODE is reduced by
the invariants of a Lie group symmetry. Hidden symmehies of partial differential equations
(PDES) also exist but are not discussed here [10,21].The invariants of symmetries of ODES
have been found from the characteristic equations of the local @oint) group generators but
invariants have also been found from exponential non-local group generators [6,9,11,16].
For ODES these invariants are usually called differential invariants but the two lowest order
invariants are the path curve, a function of the independent and dependent variable, and the
first differential invariant, a function of the independent variable; the dependent variable
hnd~thederivative of the dependent variable with respect to the independent variable.
.' me inves'tigations of hidden symmetries have focused on the reduction of order of
ODES by additional and longer reduction paths and on^ understanding the origin of these
11 Permanent address: Lkpytment of Mathematics and Applied Mathematics, University of Natal, Private XiO,
Dalbridge 4014. South Africa.
7 Member of the hire for T h e o m i d and Computational Chemistry, University,of Natal. Dukan and associate
member of the Centre for Nonlinear Shdies, University of the Witwatersrand. Johannesburg. :. .
03054470/95/236707+10$19.50 @ 1995 IOP Publishing Ltd 6707
6708 B Abraham-Shrauneret a1
hidden symmetries. A motivation for studying these hidden symmetries arose from the
nonlinear characteristic equations of the Vlasov equation for plasmas for which additional
solutions would be advantageous [221. The existence of type I hidden symmehies allows
the increase in the order of an ODE to a new ODE of one higher order which has more
than one additional point symmetry. This ad hoc approach requires guessing the variable
transformation. In a more systematic approach an ODE can be reduced in order by non-
normal subgroup invariants such that one or more point symmetries of the ODE are lost in
addition to the symmeby used in the reduction of the ODE. The lower-order reduced ODE
has a type I hidden symmetry and tables can be prepared of these ODES [2,4]. From the
tables the order of the ODES in the table can be increased to the higher-order ODE with
the additional symmetries. For both the ad hoc and systematic methods an ODE can be
reduced in order more than would be expected by the point symmetries determined by the
Lie classical method. Type II hidden symmetries also extend the reduction path to an ODE
of lower-order than expected by the symmetries determined of the original ODE by the Lie
classical method since one or more symmetries are gained when the order of this ODE is
reduced.
The origin of type I hidden symmetries is known. These arise if the order of an ODE
is reduced by the invariants of a non-normal subgroup.. The existence of a non-normal
subgroup can be identified from the commutation relations of the Lie algebra of the group
generators which represent the symmetry under which the ODE is invariant [l, 191. A normal
subgroup has the infinitesimal (group) generator Ut which satisfies the commutator
[Ui, u j ] = c h u k (1)
for j # i where either k = i or Ct.= 0. If j # i, k # i and C$ # 0, the point symmetries
of CJj vanish when the order of the ODE is reduced by the non-normal subgroup invariants
(path curve and first differential invariant) of Vi.
The origin of type I1 hidden symmetries has until recently been obscure. One class
of type II hidden symmetries arises when a third-order ODE, invariant under a solvable,
non-Abelian three-parameter Lie group with at least one commutator that involves the three
group generators, is reduced in order by the appropriate non-normal subgroup invariants
[9,14]. The two distinct Lie algebras are:
Case 1
[uj,uj]=c $ f / k [~j,L/x]=c~kuj [(Ik.ui]=o (2)
Case 2
[vi,uj]= Ckuk [uj,U,] =cfkuk [uk, uj]= o (3)
where the repeated indices do not indicate summation. The generator basis of case 2 may be
modified such that two commutators are zero. It is assumed that the order of the third-order
ODE is reduced by the invariants of Ui and that the order of the second-order ODE is reduced
by the invariants of Gk (the reduced form of u k ) for these two cases. Then ch is always
non-zero but Cjk or Cfk may be zero. The reduction of the third-order ODE transforms Uj
to a linear non-local group generator and to a local group generator. This has been
shown by assuming the normal forms for & and [14] and by solving for the general
form of U,.The reduction of the second-order ODE by the invariants of the reduced form
of transforms the reduced form of V j to a local group generator. For case 1 we could
just as well reduce the order of the third-order ODE by u k . For Case 2 reduction by u k
first gives only local group generators in the subsequent reductions. Reduction by Uj first
gives non-local group generators in both cases for Cli and (Ik if both structure constants are
Hidden and contact symmetries of differential equaiions 6709
non-zero. We note that, if both structure constants are non-zero in case 1, that there is no
path that has local group generators in all reductions, but the group is nevertheless solvable.
Solvable structures have been applied to type U hidden symmetries [X, 121 that appeared
in two examples [1,3] of second-order ODES. The solvable structures approach in these
references substitutes a linear operator for the second-order ODE as has been done in an
alternate approach to symmetry analysis of ODES [19,23]. In addition to the point group
generator, that represents the symmetries of this ODE, it introduces a group generator of
the reduced ODE which reveals the existence of the hidden symmetry. The PDE for the
coordinate function of this latter group generator is the determining equation found by
applying the Lie classical method to the first-order ODE. The difficulties usually associated
with solving this determining equation for the invariance of a first-order ODE are avoided
by the choice of a single coordinate function. More general forms of the group generator
of the first-order ODE that represent the type 11 hidden symmetry are possible but can be
difficult to calculate. Type I1 hidden symmetries are not analysed by the method of solvable
structures in this paper.
Third-order ODES and contact symmetries are discussed in section 2. The commutation
table of the Lie algebra for a y”’ = 0 found previously [15] is given in section 3 and
the reduction of ODES by use of these tables is discussed.. A new class~oftype 11 hidden
symmetries is identified in section 3. In section 4 a second-order ODE with the new type I1
hidden symmetry is analysed. In section 5 a third-order ODE which has two classes of
type II hidden symmetries is analysed. Section 6 presents the conclusions.
2. Third-order ODES and contact symmetries
Contact transformations are transformations in which the transformed variables 2,j , and 7
are functions of the original variables x , y and y’ [19]. The infinitesimal transformation is
represented by the group generator which has the form
where
These relations are explained by Stephani I191 and in other references. Customarily the two
conditions in equation (5) are then replaced by the coordinate functions defined in terms of
a characteristic (generating) function Q ( x , y , y‘) which are
If the coordinate functions Cj and qj, depend on x and y only, the contact symmetries are
identical to point symmetries. To differentiate between contact symmetries that are identical
with point symmetries and those that are not we define a contact symmetry where at least
one of ej and qj depends on y’ as an intrinsic contact symmetry.
Computing the contact symmetries for an ODE by a direct method is done in principle
in the same fashion as for the Lie classical method for point symmetries. The contact
symmetries for first-order ODES are essentially the same as point symmetries but for second-
order ODES there are, in principle, an infinite number of contact symmetries where the
determining equation is a single PDE in x, y and y’. This single PDE is even more complicated
to solve than the single determining equation for point symmetries of a first-order ODE. The
6710 B Abraham-Shrauner et a1
determining equation for a third-order ODE does separate into a set of differential equations.
The method employed here for finding the contact symmetries of a third-order ODE avoids
the introduction of the generating function nj(x,y. y') and uses the symbolic program LIE
[24]. The third-order ODE is replaced by three lower-order ODEs where y' is replaced by a
third variable, U. The point symmetries which are determined by LIE for this set of ODES are
the contact symmetries. All contact and generalized symmetries may be found in principle
if the hiid-order ODE is replaced by three first-order ODES [l] but finding the symmetries of
a set of first-order ODES is not simple. Usually an ansatz of the dependence ofthe coordinate
functions on the variables must be made [251. For contact symmetries of a third-order ODE
y"' = f ( x , Y , y', Y") 6
we let U = y' such that
U') = f ( x , y , U , U')
U = y'
U' = y"
where the prime denotes differentiation with respect to x and equation (Sc) is necessary so
that the computer program does not treat U' and p" as independent variables. The coordinate
functions as found by the symbolic program LIE for the point symmetries of equations ( 8 4 -
(Sc) are now functions of (x, y , U). These coordinate functions can be expressed as functions
of (x. y. y') for the contact symmetries of the original third-order ODE (7).
3. Lie algebra of contact symmetries for third-order ODES
The commutation tables for two third-order ODEs have been reported [15]. The commutation
table for the ODE
y!" = 0 (9)
is given in table 1. The group generators for the ten contact symmetries in different form
given by Ibragimov [20] are
Gq =
a
- a a a
G9=2(xy'-y)-+xyn-+y"7
ax ax ay ay
a a a
G5=x-+y-
ax ay
ClO = (x2y' - 2xy)-ax + (;-x y * - 2y2 ) $
The contact symmetries of this Lie algebra sp(4) [15] can be further divided into seven
point symmetries, G1-G7. and three intrinsic contact symmetries, Cs-Glo, where all point
symmetries are also contact symmetries. The symmeiries, Gs-Glo, can be verified to be
intrinsic contact symmetries by finding the characteristic function from equation (6).
Hidden and confact symmetries of differential equations 6711
Table 1. Commutation table for the ten contact symmetries of y"' = 0.
The Kummer-Schwan equation
2y'y''' - 3y"* = 0
is also invariant under the ten-dimensional Lie algebra sp(4) for the contact symmetries.
The ten contact symmetries for the Kummer-Schwarz equation further divide into six point
symmetries, and four intrinsic contact symmetries. The commutation table for the Kummer-
Schwarz equation has the same form as that in table 1 but the group generators in tabIe 1
are replaced by other group generators that are given elsewhere [15].
The conditions for type I and type Jl hidden symmetries as given for the Lie point
symmetries apply also to contact symmetries. That can be seen since the contact symmetries
of the thud-order ODE (7) are point symmetries of the equivalent set of one second-order
and two first-order ODES in equations ( 8 4 4 8 4
The reduction of the equation (9) to quadratures can be done by using the symmetries of
the Abelian subalgebras of G I , G I and G3 or Gs, G g and Glo in any order. The symmetries
of the group generators Gq. Gs and G6 cannot be used to reduce equation (9) to quadratures
since the subgroup represented by these group generators is not solvable. Other reduction
paths may take various symmetries. Let us examine one,case. Reduce equation (9), y"' = 0,
by the invariants, U = x , U = y'. of G I to find
ii=O - ' ' (12)
where the overdot denotes differentiation with respect to U. The conditions for the group
generators of the reduced ODE to become non-local are given in equation (1). In this
reduction the symmetries of G6, Gg and Glo become non-local. The reduced group
generators'of Gz, G3, G4, Gg, G7 and Gs are local. In addition we note that the contact
group generator G s transforms to a point group generator when reduced. The reduced group
generators can be found from the relation
r5. = Gj(u)-a + G ~ ( u )aG
au
for j = 2 to 10 where we already~have the once extended group generators of the point
symmetries since the group generators in equation (10) represent contact symmetries. The
reduced group generators in the invariants, U and U, are
a
v, = -
av
v, = u- a
au
6712 B Abraham-Shrauner et al
v, = 2u-aua v, = u- a
a
v, = -
au
v, = u-aua KO= (11% - 2 u / udu); + (U? - 2uJ udu);.
The third-order ODE (9) has the maximum number of point symmetries which is seven.
One of those point symmetries is used to reduce the order of the ODE, one point symmetry
becomes non-local and one contact symmetry becomes a point symmetry which leaves
six point symmetries of the second-order ODE (12). However Lie [19,23] proved that the
second-order ODE, U = 0,has eight point symmetries. The other two point symmetries are
type I1 hidden symmetries, i.e. their origins lie in non-local symmetries of (9). (In the case
of a non-Cartan symmetry its origin lies in a non-local contact symmetry.)
The transformation of Gs from an intrinsic contact group generator to a point group
generator upon the reduction of order of the thud-order'ODE (9) introduces a new class
of type Il hidden symmetries. The other two contact symmetries as represented by Gg
and Glo transform to non-local symmetries. The intrinsic contact symmetries all reduce
to symmetries that are something other than intrinsic contact symmetries. This is expected
since as y' is the new dependent variable, the coordinate functions of the reduced group
generators depend on the new variables U and U and possibly the non-local variable x .
Many examples of intrinsic contact symmetries transforming to point symmetries upon the
reduction of order of the ODE by symmetry invariants can be found by using the symmetries
of the group Sp(4). A complete investigation of all the reduction paths that are possible
of equation (9) with the symmetries of Sp(4) is clearly formidable. Nevertheless, if that
were done, a number of lower-order ODES with varying number of symmetries could be
found. The reduction of the Kummer-Schwarz equation proceeds in the same manner. The
significant result in comparing the symmetries of equations (9) and (1 1) is that the point
symmetries differ but the Lie algebra of the ten contact symmetries is the same. As a
consequence the reduction in order by symmetries of these two equations should be similar.
4. Reduction of ODE invariant under one point and one contact symmetry
For the first example we consider a second-order ODE that is invariant under one Lie point
symmetry and one contact symmetry. The group generators are G4 (point) and Gs (contact)
from equation (10) and they commute. The general form of a second-order ODE invariant
under the two groups represented by these group generators is
where g(y') is an arbitrary function of y'. The ODE (15) has a type I1 hidden symmetry
since the Lie classical method finds only one Lie point symmetry. The other symmeq
can be verified as an intrinsic contact symmetry by finding the generating function from
equation (6) for Gg. The reduction is done by the invariants of the point symmetries of Gq
first although we could do the reduction by Gs first since the two-parameter subgroup is
Hidden and contact symmetries of differential equations 6713
Abelian. The invariants are U = y , v = y’. The reduced first-order ODE is
U2
VU = (16)
2u.+ uZg(v)
where U = dv/du. The reduced group generator V8 is found to be
where the contact group generator Ga has been transformed upon reduction to a local
(point) group generator vg. That occurs since the differential invariants of Gg define new
coordinates u ( x , y ) and v ( x , y, y’) such that the derivative y‘ is folded into the dependent
variable.
The first-order ODE (16) is invariant under the first prolongation (extension) of &. The
canonical coordinates calculated from V8 are U and U f v2. In these coordinates the first-order
ODE is separable. Also the ODE is linear if U is the dependent variable. The result for U is
Further integration is possible only if equation (18) can be inverted for v in terms of U
which produces y‘ = h ( y ) from which an implicit function of x is found as an integral.
Parametric expressions for x and y in terms of integrals (one is double) over the parameter
v are also possible.
The significant result of this section is that a second-order ODE invariant under only
one point symmetry has a new type I1 hidden symmetry that is a contact symmetry. With
this type II hidden symmetry the second-order ODE reduces to a first-order ODE which has
an additional point symmetry over that predicted by the Lie classical method for point
symmetries of the second-order ODE. As a result the second-order ODE can be reduced to
quadratures or parametric relations for the two variables x and y can be found. The direct
determination of the contact symmetries of a second-order ODE is difficult since a single
determining equation, a PDE, is found. However, the realization that contact symmetries
may be present and can transform to point symmetries in the reduced first-order ODE is
important. , .
5. Reduction of ODE invariant under one point and two contact symmetries
The third-order ODE, invariant under a threeparameter subgroup with a three-dimensional
subalgebra of the three group generators: G4 (point), Gg (contact) and G9 (contact) in
equation (lo), has the solved form
y”
Y”’ = jp (5R
-2Y)
for g an arbitrary function of its argument.. By the Lie classical method this ODE has
only one point symmetry, that of Gq. Nevertheless, if.this third-order ODE is reduced
by the invariants of G4, the resultant second-order ODE has a point symmetry from the
transformation of the contact symmetry of G8 to a point symmetry of the reduced ODE. We
know that the reduced form of Gg is local as G4 and Gs commute. On the other hand
G9 transforms to a non-local group generator since the commutation relations for the three
group generators are applicable to either case in equations (2) or (3) with CjKor Cjk = 0
respectively. As the reduced form of G8 is local and the subalgebra of G4,138 and G9 is
of the form in equations (2) or (3), then reduction of the second-order ODE to a first-order
6714 B Abraham-Shrauner et a1
ODE by the invariants of the reduced G8 gives a local reduced group generator for the twice
reduced Gg.
The first reduction of equation (19) is done with the invariants of G4 which are U = y
and U = y'. The reduced ODE is
u2ij + U L P = -g53v (Tu -,2u).
U
The reduced group generators are
where as predicted from the Lie algebra V, and V, are point and linear non-local group
generators respectively. The second-order ODE is invariant under the second prolongations
of ,Vsand V9.
The second reduction is done by the invariants of V8 which are
1 2u
z=u w=_--.
u t )
The first-order ODE is
- +
z3dw z2w = -g(zw).
dz
The first extension of V9 is
The reduced form of V,'" is U,,
which is now a local group generator as predicted. The first-order ODE is invariant under
the first prolongation of (19. This example is particularly illuminating since two classes
of type II hidden symmetries occur: the transformation of a contact symmetry to a point
symmetry and the transformation of a linear non-local symmetry to a point symmetry.
A physically motivated equation which fits into the context of the examples considered
here is found in the Langevin equation, which describes the non-relativistic one-dimensional
motion of a panicle in the presence of radiation, given by [26,7]
d2x ...
m- = mx = eE -k F-+ m r x
dt2
where z = 2e2/3mc3, m is the mass of the particle, e is its charge, F = F ( x , t ) . a given
external force and E is the electric field of the radiation. If the external force is linear in
the displacement, i.e. F =,--Kx, [27] shows that the equation
possesses five Lie point symmetries in general. The Lie algebra of these five symmetries is
3A1 @ A I @ A I [28]. In the case that 2m = -27r2K the number of symmetries increases
to ten contact symmetries. Consequently the methods of reduction as discussed above can
be used for (26). The presence or not of the forcing term does not affcct the number of
symmetries although it does increase the complexity of their expressions.
Hidden and contact symmetries of differential equations 6715
6. Conclusions
The usefuhess of contact transformations for reducing the order of ODES has been
demonstrated. The contact transformations have frequently be& dismissed as curiosities
but they are fundamental in evaluating the symmetries of third-order ODES and reducing the
order of third-order ODES. The evaluation of the contact symmetries of third-order ODES
is now especially easy with the realization that the third-order ODE may be replaced by
an equivalent set of lower-order ODES in the independent variable, the dependent variable
and the first derivative for which the point symmetries are the contact symmetries of the
third-order ODE. The advent of symbolic computations of point symmetries adds to the ease
of determination of the contact symmetries.
The reduction of order of the third-order ODE, y"' = 0, by a combination of contact
and point symmetries is similar to that for the reduction of ODES by point symmetries. The
higher-dimensional Lie algebra (ten-dimensional) for the contact symmetries than for the
point symmetries (seven-dimensional) offers more possibilities for reduction of the ODE.
Type I and type Il hidden symmetries as were found in the reduction of ODES by point
symmetries occur with contact symmetries. The significant new result is the identification
of a new variety of type I1 hidden symmetry where the contact symmetry transforms to a
point symmetry upon reduction of order of the ODE.
Since this new type II hidden symmetry may occur in other third-order and also second-
order ODES, we have analysed two examples of ODES invariant under a subgroup of the ten-
parameter Lie group. The second-order ODE is invariant under one point and one contact
symmetry which has an Abelian Lie algebra. It has the new type Il hidden symmeuy since
the contact symmetry transforms to a point symmetry upon reduction of the ODE to first
order by the invariants of the point symmetry. The significance of this example is that
second-order ODES may possess contact symmetries which transform to point symmetries
upon reduction of order of the ODE but the existence of the contact symmetries may be
difficult to detect.
The third-order ODE is invariant under one point symmetry and two contact symmetries.
Again the Lie algebra is solvable and the contact symmetries successively transform to
point symmetries. The third-order ODE has the new type II hidden symmetry which is a
contact symmetry and the second-order ODE has the type I1 hidden symmetry which is a
linear non-local symmetry. The latter is a double type I1 hidden symmetry since one contact
symmetry upon the first reduction becomes a linear non-local symmetry and then transforms
to a point symmetry upon the second reduction.
Acknowledgments
This work was partially funded by the Southwestem Bell Foundation Grant and B Abraham-
Shrauner thanks the Foundation for its support. P G L Leach and K S Govinder thank the
Foundation for Research Development of South Africa and the University of Natal for their
continuing support.
References
[I] Olver P J 1986 Applicutinm ,$Lie Croups to Differential Equationc (Berlin: Springer) pp 73, 148, 1.74,
184-5
[21 Abraham-Shrauner B and Guo A 1992 Hidden symmeuies associated with the projective group of nonlinear
first-order ardinnry differential equations J. Phys. A: Math. Gen. 25 5597408
6116 B Abraham-Shrauner et a1
[3] Guo A and Abraham-Shrauner B 1993 Hidden symmetries of energy canserving differential equations IMA
J. Appl. Moth. 51 147-53
[4] Abraham-Shrauner B and Guo A 1994 Hidden symmetries of differential equations Cont. Math. 160
(Providence, RI:American Mnthematica! Society) pp 1-13
I51 Abraham-Shrauner B 1993 Hidden symmetries and 1ineariw.tion of the modified Painlev&Ince equation
I. Math. Phys. 34 4809-16
[6] Abraham-Shrauner B and Guo A 1993 Hidden and nonlocal symmetries of nonlinear differential equations
Advanced Analytical and Computational Methodr in Marhemdical Physics ed
si and A Valenti (Dordrecht: Khwer) pp 1-5
[7] Abraham-Shrauner B and Leach P G L 1993 Hidden symmetries of nonlinear ordinary differential equations
&hitingSymmetry in AppliedmdNumerical Analysis (Lectures in AppliedMathemdics 29) (Providence,
RI: American Mathematical Society) pp 1-10
[SI Hartl T and Athome C 1994 Solvable structures and hidden symmetries J. Phys. A: Math. Gen. 27 3463-74
191 Abraham-Shrauner B 1994 Hidden symmetriesof ordinary differential equations Proc. IMACS World Congress
vol 1, ed W F Ames (Atlanla, G A lntematiord Association for Mathemtics and Computers in Simulation)
PP 1-1
1101 Coggeshall S V, Abraham-Shrauner B and Knapp C 1994 Hidden symmetries of partial differential equations
Pmc. of IMACS World Congress vol 1, ed W F Ames (Atlanta, G A International Association for
Mathematics and Computers in Simulation) pp 102-7
[l 11 Abraham-Shrauner B 1994 Nonlocal group generators Preprint
LlZ] Mansfield E L 1994 Geometric Inlegrution of Differentid Eqrrati!,ns (Notes) Depmmenl of Mathematics,
University of Exeter
[13] Govinder K S and Leach P G L 1995 On the determination of nonlocal symmetries J. Phys. A: Math. Gen.
28 5349
[I41 Abraham-Shrauner B 1995 Hidden symmetries and nonlocal group generators for ordinary differential
equations IMA J. AppL Math. (submitted for publication)
1151 Abraham-Shrauner B, Ratcliff G, Govinder K S and Leach F G L 1995 Linearization of mird order ordin&
differential equations by conmet transformations J. Muth AnaL Appl. (submitted for publication)
1161 Abraham-Shnuner B, Govinder K S and Leach P G L 1995 Integration of second order ordinary differential
equations not possessing Lie point symmetries Phys. Lett. 203A 169-74
[I71 Ovsiannikov L V 1982 Croup Analysk rfDifferential Eyuations (New York Academic)
[18] Bluman G W and Kumei S 1989 S y m t r k s and Diferentiui Eyuutiunr (AppL Math. Sci. 81) (Berlin:
Springer) pp 129-30
[i9] Stephani H 1989 Differentid Eyuations (Cambridge: Cambridge University Press) pp 3 1 - 2 , 6 1 4 83, 105-9
1201 lbragimov N H (ed) 1994 CRC Handbook ($Lie Gnup Ann1ysi.s of DifferentialEquations I (Baton Roca.
F L CRC) p 28
[21] Clarkson P A 1994 (private commuuication)
[22] Abraham-Shrauner B 1985 Lie wansformation group Solutions of the nonlinear one-dimensional V1;c;ov
equation 1. Muth. Phy.7. 26 1428-35 =
[23] Cohen A 1911 An Intmduction to the Lie Theory of One-Purumeter Groups (Boston, MA: Heath)
[241 Head A K 1993 LIE 8 PC program for Lie analysis of differential equations Comput. Phys. O!mm 77 241-8
[25] Abraham-Shrauner B 1990 Lie gmup symmetries and invariance of the Henon-HeiJes equations J. Math.
Phy.s. 31 1627-31
[26] de la PeRa-Auehch L and Cello A M 1976 The harmonic oscillator in a mndom elecmmagnelic field
SchrBdinger’s equation and radiative corrections Rev. Ma,Fk. 25 1-21
[271 Soam Net0 1 J and Vianna 1 D M 1988 lnfinircsimal symmetry transformation of the Langevin equation
J. Phy,s. A: Math. Gen. 21 2487-90
[281 Mahomed F and Leach P G L 1990 Symmetry Lie algebns of nih order ordinary differential equations
J. Math. Anal. Appl. 151 80-107
