Вы находитесь на странице: 1из 13

Mathl. Comput.

Modelling

Pergamon

Vol. 25, No. 819, pp. 181-193, 1997


Copyright@1997
Elsevier Science Ltd

Printed in Great Britain. All rights reserved


0895-7177/97
$17.00 + 0.00

PII:SO8957177(97)00068-X

The Role of Symmetries


in Solving Differential Equations
M. C. NUCCI
di Matematica, Universitk di Perugia
06123 Perugia, Italy
nucci@gauss.dipmat.unipg.it

Dipartimento

review of the role of symmetries in solving differential equations is presented. After showing some recent results on the application of classical Lie point symmetries to problems
in fluid draining, meteorology, and epidemiology of AIDS, the nonclassical symmetries method is
presented. Finally, it is shown that iterations of the nonclassical symmetries method yield new nonlinear equations, which inherit the Lie symmetry algebra of the given equation. Invariant solutions
of these equations supply new solutions of the original equation. Furthermore, the equations yield
both partial symmetries 8s given by Vorobev, and differential constraints as given by Vorobev and
by Olver. Some examples are given. The importance of using ad hoc interactive REDUCE programs
is underlined.

Abstract-A

Keywords-Lie

symmetries,

Computer

algebra, Fluid mechanics,

Medical epidemiology.

1. INTRODUCTION
The most

famous and established


method for finding exact solutions of differential equations is
the classical symmetries
method (CSM), also called group analysis, which originated in 1881 from
the pioneering work of S. Lie [l]. Many good books have been dedicated to this subject and its
generalizations
[2-l 11.
Although
Lie point symmetries
represent a very powerful tool, they can yield very lengthy
calculations.
In fact, interest in them and their generalizations
has increased during the last
twenty years because of the availability
of symbolic computation
packages.
In this paper, we review some of our recent work in the area. First, we illustrate
Lie group
analysis with some examples of ordinary differential equations.
We show a third-order
ordinary
differential
equation
admitting
a three-dimensional
nonsolvable
Lie symmetry
algebra [12], a
third-order
ordinary differential equation admitting a two-dimensional
Lie symmetry algebra, and
a system of three first-order ordinary differential equations admitting a three-dimensional
solvable
Lie symmetry
algebra [13]. Each of these represents mathematical
models of fluid draining [14],
meteorology
[15], and epidemiology
of AIDS [16], respectively.
Second, we illustrate the nonclassical
symmetries
method introduced
by Bluman and Cole in
to the modified Korteweg-deVries
equation [18].
1969 [17] an d s h ow its application
Finally, we illustrate how to iterate the nonclassical symmetries method [19,20] for the purpose
of obtaining
new solutions and show some examples.
A reliable and user-friendly
symbolic computation
program is the key to the widespread
and
successful application
of Lies method and its generalizations.
Each given example has been
derived by using our own interactive
REDUCE programs
[21], which calculate
the Lie point
symmetries,
and the nonclassical
symmetries.
Supported

in part by Fondi M.U.R.S.T.

60% and 40%.


Typeset
181

by &,+5-&X

M. C. NUCCI

182

2. APPLICATIONS
OF LIE GROUP ANALYSIS
ORDINARY
DIFFERENTIAL
EQUATIONS
2.1. Fluid

TO

Draining

The equation
Wnt =w

-2

w =

w(s),

(1)

represents the small-w limit of an equation which is relevant to fluid draining problems on a dry
wall, and the large-w limit to draining over a wet wall [14]. Its general solution was given in 1221
by a lucky guess method. In [12] Lies method was applied to (1) and led to a three-dimensional
nonsolvable Lie algebra LB with basis
Xl

= 33,

x,

x.2 = sa, +w&,

Let us consider the two-dimensional


invariants of order I 2 is given by

= s=iJ, +

2swa,.

(2)

subalgebra spanned by X1, X2. A basis of its differential

w,

u =

v =

wwl.

(3)

Then, equation (1) is reduced to the following first-order equation:


dv
= 2Lv+ 1,
OdZL
which admits the nonlocal operator Xs in the space of variables u, v, i.e.,
x3 = w(8, + ?&).

(5)

We put (5) into its semicanonical form [12], i.e., X3 = wd,, by introducing the new variable
z=qJ---

2L=
2

which is obtained by solving


;++0.
Then, equation (4) becomes a Riccati equation in the variables z, 26,i.e.,
dzl
U=
dx=-2-+e*

(6)

Finally, the general solution of (6) is easily found in terms of Airy functions [22].
2.2. Meteorology
Let us consider the Lorenz system [15]

5 = o(y - x),

(7)

y=-0z-+Tz-~,

(3)

L = sy - bz,

(9)

where o, b, and T are parameters. This system can be reduced to a single third-order ordinary
differential equation for 2 (231,which admits a two-dimensional Lie symmetry algebra if q = l/2,
b = 1, and T = 0. Then, this third-order equation becomes
2xxett - 222 + 5x2 - 3xn + 2x32'+ 32x1 + x4 + x2 = 0,

00)

Solving Differential Equations

and admits a two-dimensional

183

Lie symmetry algebra Lz with basis


Xl

x2=et/z(at-%a,>.

4,

01)

A basis of its differential invariants of order I 2 is given by


9=(X+-2,

?f!?=

(12)

X+qXf+;)X-3.
(

Then, equation (10) is reduced to the following first-order equation:


(It, - 24) $
which can be easily integrated

= -2?&# - I#>

(13)

[24] to give
1+ 4ti - 4+2 = c1

(14)

Cl+ 2$)2
with an arbitrary constant cl
ordinary differential equation

Substituting

I and its derivatives into (14) yields a second-order

1 + 4 (2 + (312)s + z/2) x-~ - 4 (s -t ~/2)~X-* =I c1


(X3 + 2x + 3x + x)2 5-s

which admits the Lie symmetry algebra Lg. Lies classification of two-dimensional algebras into
four canonical types [25] allows us to integrate (15) by quadrature if we introduce the canonical
variables
2/=I --2eeti2,
transform equation

u=-,

eTt/2
X

(16)

into
l+4(~~z-4~~
2u$$

and the

(11) into
=

(F&)2 -

a,,

= C1

X2=va,+uau.

(18)

Then, the general solution of (17) can be easily derived [25] to be

F 2c2?.?

c2u 4>-112dU=f2)+Cg

(19)

with arbitrary constants ~2, cg. This solution has already been obtained by Sen and Tabor [23],
~though they used a lengthier analysis.
2.3.

Epidemiology

of AIDS

In [13], Lie group analysis has been applied to a model formulated by Anderson (161, which
describes HIV transmission in male homosexual/bisexual cohorts:

dul

-L=

dt

--P~1~2

Ul -i-u2

+u3

-WI,

(20)

(21)
dw

dt = vu2- Qu3*

cw

184

M. C. NUCCI

This compartmental model divides the population at time t into susceptibles (HIV negatives),
infecteds (HIV positives), and AIDS patients, represented by al(t), zlz(t), and ~sft), respectively.
HIV infecteds are individuals who test positive for specific antibodies to the virus [26]. AIDS
patients are persons exhibiting characteristic clinical manifestations of full-blown AIDS, the endstage of the disease [27]. The parameter /.Jis the per capita natural death rate (nonAIDS related)
of both susceptibles and infecteds, and cr is the AIDS-related death rate. The parameter @ is the
average probability that an infected individual will infect a susceptible partner over the duration
of their relationship [16,28,29], and c is the effective rate of partner change within the specified
risk category [16,29]. In the model, all infecteds are supposed to develop AIDS with an average
incubation period l/v [16,27].
It is well known that a first-order system of ordinary differential equations admits an infinitedimensional Lie symmetry algebra [25]. Lies theorem allows us to integrate the given system
by quadrature, if we find a thr~dimension~
solvable Lie algebra. Let the operator r be the
generator of a symmetry group

r = <(t,Ul,Z&,U3)&

+ r]l(t,%r~2,113)&,

+7)2(t,Ul,U2,U3)&

+~3(k%,U2~U3)f%a.

(23)

To find a thre~dimension~
solvable Lie algebra, we make an ansatz on the form of the operator I,
and then verify if such an operator is admitted by the system. We assume that the functions
5, r]r, and ~2 are polynomials of second degree on ~1, us, and us. After some lengthy analysis, we
obtain that (20)-(22) admits a three-dimensional solvable Lie algebra Ls with basis
Xl

= at _ Pld,,

x2 = Ul&*
x3

-_ ,-+++

- pu2avz

+ uza,,
a,,

- /Lu3&&,

+ u3&,

+ I!!a+uaus
262

(24)
(25)

)
>

(26)

if the death rate of AIDS patients is the sum of the death rate of HIV infecteds plus the probability
of transmi~ion per partner contact multiplied by the effective rate of partner change, i.e., ~3:=
p + PC. It is easy to show that X2 and Xs span a two-dimensional ideal Ls of La. If L1 denotes
the ideal spanned by X3, we get the following chain of inclusions:
L3 2 L2 3 Ll,

which means that the algebra we have found is solvable. By using a basis of differential invariants
of L~L, we can reduce the system to a first-order ordinary differential equation which can be
integrated by quadrature, because it admits the Lie algebra Ls/Ls spanned by Xr in the new
variables. A basis of differential invariants of Ls of order 0 is
t,

+-Jc

u1+

213

(27)

Then, the system (20)-(22) can be easily reduced to the following first-order equation:

which admits the operator Xl = &, i.e., the operator Xr in the variables (t, 6). Therefore,
can be easily integrated by quadrature. Its general solution is
e@ t (PC - u)

c= eBC%+eYt,i3cct. - evtclu'

(28)

Solving Differential Equations

185

where cl is an arbitrary constant. From (29) and (27) we obtain


u3 =

ePct (211 + ~2) v -

/3cu~ + evt (/3c -

eflct (PC -

which substituted into the system (20)-(22)


general solution of the system (20)-(22):

v) qu2

v)

and after some easy calculations yields the following

eVtc2
e@ [evt

(0 c - u) cl

(30)

eflctp C]

ept+vt
[ev(/?c-u)cl+e

pctpc]

[evt (p c - u) cl

eflct+pt+vt [evt (Pc - u) cl

,p

w-a

eflctu]

ePctPc] (PC - v)

Ct+Zt

(eP=t~C+eLBCC1-eYtC~~)~

eo Ct+pt+Yt [eVt (/3 c -

(31)

dtPcc2

v) cl + eP ctp c] (p c -

+ c3
V)

(32)

-evtc2
+

e@

[eyt (PC

v) cl +

eflctpc]

where c2,cs are arbitrary constants. In [13], the behaviour of this solution is compared with
epidemiologic data on the incidence of HIV-infection, and AIDS in male homosexual/bisexual
populations in U.S. areas most affected by the HIV epidemic: the Multicenter AIDS Cohort
Study [30], the San Francisco Mens Health Study [31], and the San Francisco City Clinic Cohort, [32].

3. NONCLASSICAL

SYMMETRIES

METHOD

The nonclassical symmetries method (NSM) was introduced in 1969 by Bluman and Cole [17]
to obtain
new exact solutions of the linear heat equation, i.e., solutions not deducible from the
classical symmetries method (CSM). The NSM consists of adding the invariant surface condition
to the given equation, and then applying the CSM. The main difficulty of this approach is that
the determining equations are no longer linear. On the other hand, the NSM may give more
solutions than the CSM. The NSM has been successfully applied to various equations [33--37],
for the purpose of finding new exact solutions.
3.1.

Modified

Korteweg-de

Vries Equation

In [18], the NSM was applied to the modified Korteweg-de Vries (mKdV) equation
ut = u,,,

- 6~~212,

(33)

with the invariant surface condition given by


u, = G(t, 2, u).

(34)

Then, G must satisfy


3GGzz, + 3GG,G,,
+3G2G,G,,

+ 3GG,G,,

+ 3G,G,,

+ G3G,w

- Gt + G,,,

1 Just to cite some of numerous papers on this subject.

+ 3G2G,,,

- 6u2G, - 12uG2 = 0.

(35)

M. C. NUCCI

186

If we apply the NSM to (35), we find that its invariant surface condition
Sl(t, 2, u, G)G.

+ 52(4 2, u, Wz

+ b(t,

r, u, G>G

- v(t, 5, u, G) = 0

(36)

is given by
G, = -221,

(37)

G = 4 - u2,

(38)

which implies

with an arbitrary

function q(z,t).

Then, substituting
qt =

into (35) yields

qxxx- Wm

(39)

which is the Kortweg-deVries (KdV) equation. In fact, (34) and (33)2 yield the Biicklund transformation between the mKdV and KdV equations [38], i.e.,
u, =q-u2,

(40)

Ut = qzz - 292 + 2u2q - 2uq,.

(41)

This means that the Backlund tr~sformation


(40),(41) between the mKdV and KdV equations
can be obtained by applying the NSM twice to the mKdV equation.
4.

ITERATING

THE NONCLASSICAL

SYMMETRIES

METHOD

Recently, Galaktionov [39] and King [40] have found exact solutions of certain evolution equa
tions which apparently do not seem to be derivable by either the CSM or NSM. However, we
have shown [19] that these solutions can be obtained by iterating the NSM (INSM). Successive
iterations generate new equations which, although more complex than the original equations, inherit the same Lie symmetry algebra. Invariant solutions of these equations are just the solutions
found by Galaktionov and by King.
The use of a symbolic computation program becomes imperative, because these equations can
be quite long: one more independent variable is added at each iteration.
4.1. Applications

to Evolution Equations

Let us consider an evolution equation in two independent variables and one dependent variable

The invariant surface condition is given by

Vi@,2, u)ut + Vi@,5, u)uz = G(t, z, u).

(43)

Let us take the case with VI = 0 and Vs = 1, so that (43) becomes


uz = G(t,

2,

u).

(44)

Applying the NSM leads to an equation for G (the G-equation [18]). Its invariant surface condition
is given by
&(t, 2, U, G)Gt + Ez(G G u, W,
+ 53@, 5, u, G)Gu = v(t, G u, G).
(45)
aAfter substituting us with C, and usziF with @,

respectively.

Solving Differential Equations


Let us consider

the case 51 = 0, <2 = 1, and <s = G, so that


G, + GG, =

Applying

the NSM leads to an equation

We could keep iterating

the p-equation,

to obtain

which corresponds

u,,

the Q-equation,

(46)
Clearly,

= q.

(47)

which corresponds

to

to

0, + GQ, + q% + RR, = u,,,,


and so on. Each
the appropriate
v-equation,
and
This iterating
Vorobev in [41].

(45) becomes

q(t, x, u, G).

for 77(the q-equation).

G, + GG,

187

(49)

= p(t, x, u, G, q, O),

of these equations inherits the symmetry algebra of the original equation, with
prolongation:
first prolongation
for the G-equation,
second prolongation
for the
so on.
method is strongly connected to the definition of partial symmetries
given by
To exemplify, let us consider the heat equation

with G-equation
2GG,,

i- G2G,,

- Gt +- G,,

= o,

(51)

2GqZU + qZs - qt + G2qUU = 0.

(52)

and v-equation
273%G -k 2GQ%G

i- 7j2qGG +

The G-equation
corresponds
to the zeroth-order
differential constraint
as given by Vorobev [42,
p. 76, formula (3.5)], while the q-equation
gives the partial symmetry of the heat equation as in
[41, p. 324, formula (14)], and in [42, p. 83, formula (4.10)].
Also, it, should be noticed that the uxx
-equation
of (42) is just one of many possible
+Y
r-extended
equations as defined by Guthrie ;n [43].
In [19], it has been shown that the solutions obtained in [39,40] are just t-independent
invariant
solutions,
which have x as similarity
variable.
In this way, ordinary differential
equations
are
obtained and their general solution will depend on arbitrary functions oft. Substituting
into the
original equation will give rise to ordinary differential equations to be satisfied by these functions
oft.
GALAKTIONOVS EXAMPLE. [39] Let us consider

the equation

Ut=uz,+U2fU2,

(53)

with G-equation

2GG,, -I- G2G,, + G2G, - u2G, -

Gt + G,,

+ 2GG,

+ zuG

0,

(54)

and q-equation
27777,G+ 2%uG

+ q2?,GG - 2UGqG + 2GqzU + qzl

+2Gqz - qt + G2qw, + G2qu - u2q, + zq2 + suq + 2G2 = 0.

(55)

M. C.Nuccr

188

The symmetry algebra of (53) is spanned by the two vector fields Xi = at and Xz = 8,.
Therefore, t-independent invariant solutions of (55) are given in the form n = 71(x,u,G). A
particular case is nU = 0, which implies 7) = L(z, G). Substituting this expression for q into (55)
leads to L = f(z)G where
-cl sin x + cs cos x
f(x) = c2sinz + cl cosx
(56)
and cl, c2 are arbitrary constants. If we take cr = 0, then
q = cot(z)G,

(57)

which is just the differential constraint for (53) given by Olver in [44], i.e.,
u

IX

(58)

cot(x)?&.

Integrating (58) with respect to z gives rise to


u = ml(t)

where 201,ws are arbitrary functions oft.


$1

cos(z)

+ wz(t),

(59)

Finally, the substitution of (59) into (53) leads to

=w:+wg,

?iQ = 2WlW2 _ w2.

(60)

This is the solution derived by Galaktionov for (53). In the same way, iterating the NSM twice
will generate the solutions of the other equations in [39].

KING'SEXAMPLE. [40]
Let us consider the equation
N-l
?Jt= vu,, + ml, + -vvp.
r

(61)

Its symmetry algebra is spanned by the vector fields


x2 = tat - vd,,

x1 = ra, + 2v&

x3

=a$.

(62)

More vector fields exist in the following cases:


4-N

a = 2(N - 3)

=+ rBsNd,. + 2(3 - N)r2-Nvt3,,

N=le&.,
N=:,cr=

-1 I

N = 1, cy = -;

&

(64)
rlogrd,

+ 2(1+ logr)v&,

r2& -t-47-v&,,

a,..

(65)
(66)

We derive the R-equation3 of (61) and look for t-independent invariant solutions which have T
as similarity variable. Such invariant ~lutio~
can be obtained by the subalgebra spanned by
(X2, X3). The third prolongation of X2 is given by

9 =tat -

v& - vr&+ - %+aJ,, - vm&lprr,

(67)

which implies that the corresponding invariant solutions of the R-equation have the following
form:
52 = %-r = vF(r,

E4, E5),

(68)

Solving Differential Equations

189

where <4 = G/u - v,./v, (5 = q/v E vTT/v, and F must satisfy a very long equation not reported
here [19]. Let us assume that F is linear with respect to & and [s, i.e.,
F

S1(T)55

+ S2(r)t4

+ S3(T),

(69)

with arbitrary functions ~1, sz, and sg. Then, we obtain five cases, two more than those considered
in [40]. In particular, we show the case N = 1, cr = -3/4 with

sl(~)Q

S;(T)

;s:(T)

1 [

2 s:(r)

J4 -

+ $(r)s:(T)

1,

(70)

i.e.,
V

where sl(r)

r~r=srv,,--

(s;+~+.-2(s;+~s+,

(71)

must satisfy the equation

=o,

9sl+9slsl+s;

(72)

which admits a Lie symmetry algebra of dimension eight, so it is linearizable. The linearizing
transformation [25] can be obtained by using a two-dimensional subalgebra Pi, l?z, such that
l?i V I72 = 0, e.g.4,
r3
Ii = -@pi
rz =

rsl - 6)(r2sq - 6rsi + 18)&,,


- 6)&. + 18(
(73)

r2Sl&

$(rS;

6~s~

-I i8)&,

The general solution of (72) is


-6(~2 + T)

sr =

6cr - c$ - 2czr - r2

(74

which transforms (71) into

V
=

18(2cr - c; - 2c2r - r2)

-6(~2 + r)
6cl-c;-2c2~-r2

r +

(6~ - c; - 2c2r - r2)2

24(c2 + r)
+ (6cr - cf - 2c2r - r2)2v*

(75)

A particular solution of (75) is


v = a&)
Substituting (76) into (61) imposes q,(t)

(6~1 - c; -

2~27 -

Application

(76)

to have the form

e = 24clt + c3
4.2.

T2).

(77)

to Prandtl Equation

In [20], INSM has been applied to the laminar boundary layer model introduced in 1904 by
Ludwig Prandtl [45]
Uy = w,
uus + VT&= WY,
215+vy =o,
(78)
4The commutator

of these two vector fields is [Il, rs] = -6Il.

M. C. Nuccr

190

which was called the Prandtl system in [46], and to an equivalent single equation for u
2 = 0)
%%YY - u21yuzy + uuxuyy - urJa,

(79)

which was called the Prandtl equation. Both the equation and the system were reduced to several
ordinary differential equations, and new solutions were found. Therefore, it was shown that INSM
can be applied to systems of partial differential equations, and to single equations which are not
of evolutive type.
The Prandtl system (78) admits an infinite-dimensional Lie symmetry algebra generated by
the operators [4]

where o(z) is an arbitrary function of 2. It is easy to verify that the NSM is equivalent to the
CSM for the Prandtl system (78). Details on INSM applied to the Prandtl system (78) can be
found in [20].
The Prandtl equation (79) admits the same infinite-dimensional Lie symmetry algebra of the
Prandtl system, i.e.,
x1 = za, -k Il&,

x2 = &

- a?.&&,

x3

x,

= &-,

= cy(z)d$/.

(81)

The NSM is equivalent to the CSM for the Prandtl equation also. Application of INSM to the
Prandtl equation (79) means that we consider the case with VI = 0 and Vs = 1, so that the
invariant surface condition

becomes
uy = G(z, M,u).
The NSM leads to an equation for G, the G-equation.
&(z:, Y, a, G)G

+ Cz(x, Y, utt,GE,

+ 53(~

(83)

Its invariant surface condition is given by


Y, 21, G)Gu

= q(x, Y, u, G).

(84)

If we consider the case & = 0, (2 = 1, and <s = G, then (84) becomes


G, + GG, = rl(x, 2/,21,G),

(85)

and the NSM leads to an equation for q, the q-equation. Clearly,


G, + GG,

uyy z q.

036)

Continuing with the iterations gives rise to the Q-equation, which corresponds to

and so on. Each of these equations inherits the Lie symmetry algebra of (79) with the appropriate
prolongation: first prolongation for the G-equation, second prolongation for the n-equation, third
prolongation for the Q-equation, and so on.
Let us consider the R-equation and find its z-independent invariant solutions, which have y as
similarity variable [20]. Such invariant solutions can be obtained by the subalgebra spanned by
{Xl, X3). The third prolongation of Xr is given by

191

Solving Differential Equations

while
xs zxs.

(89)

The corresponding

invariant solution of the R-equation is

with (1 = G/u. z ~~/~, & = 771%sz ~*~/~, which substituted into the R-equation leads to an
equation for @(<I, y,&). Let us assume that Qi is linear with respect to y and 53, i.e.,
* = ~I(&)53

with arbitrary

+ ~dJI)Y

(91)

+ S3(G),

functions si, se, and sg. Then, we obtain that Cpmust be of the following form:
Q=

4~5

+ 4~2 + 3c3[4 + 4c3t3 - 4Ef - 4M3

(92)

with arbitrary constants cl, cs, and ca. Substituting


equation for u. and its y-derivatives:

If ci = -6?3&1936,

and c2 = 15&704,

7744v,,u,u
U YY?/
=

(--

(92) into (90) yields the following third-order

then (93) becomes

+ 7744u,,csu2 - 77442~; + 5808~;~~~ - 2692u&u2


7744u2

which admits a three-dimensional


& = e-(Wll)Y~y

+ 165~;~~)
!

(94

solvable Lie symmetry algebra generated by the operators


+ 5C3,-WfVu~~,
44

23 = ua,

22 = a,,

195)

with the commutators


I&,Z2I

I&,

= ~21,

[Zz, 231 = 0.

231 = 0,

(96)

Let us show that (94) is a particular case of equation (2.7) as given in [12] with X = 0. In fact,
by introducing the change of variables

g=logu.-5c3

U=Ge

-Try7

l1

(2CSlll)Y

equation (94) transforms into


f&j

= iigj

(32

+3%-4

,
>

which corresponds to equation (2.7) in (121with X = 0 and f(*) = 3 * +3/ * -4. Operators
transform into
21 = a,,

z2 = a,,

23 = ,iz+j,

(98)
(95)
(99)

with commutators
[G,

Z2] = 0,

[&,

&]

= 0,

[.z2, .z3] = z2.

(100)

Therefore, 21 and 22 span the ideal L2 = (2,) 22). A basis of differential invariants of La of
order < 2 is given by
z = Gsig, v=Qj.
(101)

M. C. Nucct

192

Then, equation (94) is reduced to the following first-order equation:


dv
-=3;+3;-4,
dz

(102)

which admits the operator .!?a in the space of variables (2,~).


the general solution of (102) in the following implicit form:

Through this operator we obtain

(103)
with an arbitrary function hi(z).
_2JZarctan

Then, by introducing the original variables, we obtain

%_VRj

+ log 3% - 4iigQg + 2Gs _ 410gc,


a;

-z/Y

+ 4k = 0,

(194)

which is a second-order differential equation admitting the symmetry algebra L2. This means
that (104) can be integrated by using Lies approach [25] to obtain
a =

F(g +

hz(z),

h(z))

d@ -t- h3(2),

(105)

where F is a suitable function, while hsfzr) and h3(2)


are arbitrary functions of 3~. Finally,
substituting (105) with (97) into (79) yields a system of equations for h;(z) (i = 1,2,3). Therefore,
(105), (97)> and the system for hi(s) give rise to a new solution of the Prandtl equation (79).

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
II.
12.
13.
14.
15.
16.
17.
18.
19.
20.

S.Lie, fiber die Integration durch bestimmte Integrale von einer Klasse linearer partieller Differentialgleichungen, Arch. Math. (Kristiana) 6, 328 (1881).
W.F. Ames, Nonlinear Partial Difierential Equations in Engineering, Vol. 2, Academic Press, New York,
(1972).
G.W. Bluman and J.D. Cole, Similarity Methods for Differential Equations, Springer-Verlag, Berlin, (X974).
L.V. Ovsjannikov, Group Analysis of Di~e~ntial Equations, Academic Press, New York, (1982).
P.J. Olver, Appl~~tions of Lie Gwups to ~~~t~al
Equations, Springer-Verl~, Berlin, (1986).
G.W. Bluman and S. Kumei, Symmetry
and ~~e~ntial
Eq~atjo~, Springer-Verlag, Berlin, (1989).
C. Rogers and W.F. Ames, Nonlinear Bo~nda~
Value Problema in Science and ~ngin~ng,
Academic
Press, New York, (1989).
H. Stephani, Differential Equations. Their Solution Using Symmetries, Cambridge University Press, Cambridge, (1989).
J.M. Hill, Differential Equations and Group Methods for Scientists and Engineers, CRC Press, Boca Raton,
FL, (1992).
CRC Handbook of Lie Gwup Analysis of Diflerential Equations, Vol. I: Symmetries, Exact Solutions, and
Conservation hws, (Edited by N.H. Ibragimov), CRC Press, Boca Raton, FL, (1994).
CRC Handbook of Lie Group Analysis of Differential Equations, Vol. II: Applications in Engineering and
Physical Sciences, (Edited by N.H. Ibragimov), CRC Press, Boca Raton, FL, (1995).
N.H. Ibragimov and M.C. Nucci, Integration of third order ordinary differential equations by Lies method:
Equations admitting three-dimensional Lie algebras, Lie Groups and Their Appl. 1, 49 (1994).
V. Tofrisi and M.C. Nucci, Apptication of Lie group analysis to a mathematical model which describes HIV
transm~ion in male horn~exu~/~~u~
cohorts, Preprint RT 199510.
E.O. l&k and L.W. Schwartz, A numerical and asymptotic study of some third-order ordinary differential
equations relevant to draining and coating flows, SIAM Rev. 32, 453 (1990).
E.N. Lorenz, Deterministic nonperiodic flow, J. Atmos. Sci. 20, 130 (1963).
R.M. Anderson, The role of mathematical models in the study of HIV transmission and the epidemiology
of AIDS, J. of AIDS 1, 240 (1988).
G.W. Bluman and J.D. Cole, The general similarity solution of the heat equation, J. Math. Mech. 18, 1025
(1969).
M.C. Nucci, Nonclassical symmetries and Backlund transformations, J. Math. Anal. Appl. 178, 294 (1993).
M.C. Nucci, Iterating the nonclassical symmetries method, Phyeica D 78, 124 (1994).
F. Allassia and M.C. Nucci, Symmetries and heir equations for the laminar boundary layer model, J. Math.
Anal. Appl. 201,911
(1996).

Solving Diff~ential Equations

193

21. M. C. Nucci, Interactive REDUCE programs for calculating Lie point, non-classical, Lie-l%ickiund, and
approximate symmetries of differential equations: Manual and floppy disk, In CRC Handbook of Lie Croup
Analysis of Rltferential Equations, Vol. III: New tinds, (Ediled by N.H. Ibragimov), CRC Press, Boca
R&on, FL, (1995).
22. W.F. Ford, Solution to Problem 91-3, SIAM Reu. 84, 121 (1992).
23. T. Sen and M. Tabor, Lie symmetries of the Lorens model, Physica Lz44, 313 (1990).
24. G.M. Murphy, Ordinary L%#emntiaE.&pations and Their Solutions, Van Nostrand, Princeton, NJ, (1960).
25. N.H. Ibragimov, Group analysis of ordinary differential equations and the invariance principle in msthematicai physics, Russian Math. Surneys 42, 89 (1992).
26. R.M. Anderson and R.M. May, Infectious D&eases of Humans, ~~~arn~~sand CO&T&~Oxford University
Press, Oxford, (1991).
27. J.D. Murray, ~athernot~~~ &oiogy, Springer-Verlsg, Berlin, (1989).
28. R.M. Grant et of., Infectivity of the human immun~deficiency virus: Estimates from a prospective study
of homosexual men, J. 1r&& L&s.156, 189 (1987).
29. R.M. May and R.M. Anderson, Heterogeneities, cofactors and other aspects of the transmission clynsmics
of HIV/AIDS, In Cwrent Topics in AIDS, Vol. 2, John Wiley & Sons, New York, (19893.
30. D.R. Hoover et al., Estimating the 1978-1990 and future spread of human immunodeficiency virus Type I
in subgroups of homosexual men, Am, J. Epidemioi. 134, 1190 (1991).
31, W. Winkelsteiu et al., The San Francisco mens study: III. Reduction in human immunodeficierIcy virus
transmission among homo~~xu~l/b~exual men, 1982-1986, A.J.P.N. 76, 685 (1987).
32. Centers for Disease Control, Update: Acquired immunodeficiency syndrome in the San Francisco cohort
study, 1978-1985, M.M. W.R. 84, 573 (1985).
33. D. Levi and P. Winternitr Non-classical symmetry reductions-Example of the Boussinesq equation, 3.
Pays. A: Math. Gen. 22, 2915 (1989).
34. P.A. Clarkson and P. Winter&z, Nonclassical symmetry reductions for the Kadomtsev-Petvisshviii equation,
Ph$&&?aI) 49, 257 (1991).
35. M.C. Nucci and P.A. CIarkson, The nonclsssical method is more general than the direct method for symmetry
reductions: An example of the Fit~hugb-Nagumo equation, P&s. .&et&A 164, 49 (1992).
36. M.C. Nucci and W.F. Ames, CIsssicdt and nonctsssical symmetries of the Helmholtz equation, J. Math.
Anal. Appl. 178, 584 (1993).
37. P.A. Clarkson and E.L. Mansfield, On a shallow water wave equation, Nonlinearity 7, 975 (1994).
38. C. Rogers and W.F. Shsdwick, Bidcklund Pransfownations and Their Applications, Academic Press, New
York, (1982).
39. V.A. Galaktionov, On the new exsct blow-up solutions for nonlinear heat conduction equations with source
and applications, Differential Integral Equations 8, 863 (1990).
40. J.R. King, Exact polynomial solutions to some nonlinear diffusion equations, Physica D 64, 35 (1!)93).
41. E.M. Vorobev, Partial symmetries and multidimensionai integrable differential equations, Di&renbial Equutiaras25, 322 (1989).
42. EM. Vorobev, Symmetries of ~mp~tibifity conditions for systems of differential equations, Acta Appb
butt. 26, 61 (1992).
43. G. Guthrie, Con~rueting Miura tr~sformations using symmetry groups, -arch
Report No. 85 (2993).
44, P.3. Uiver, Direct reduction and di&rential constraints, PTCX.R. Sot. hd.
A 444, 599 (1994).
45. L. Prandtl, &er Fl~~i~keits~egung bei sehr kleiner R&bung, In ~~ff~d~~~gen
des IH. I~ternizt~u~~~e~
Mathematiker-iCongresses,
Heidelberg, 190& Teubner, Leipzig, (1905).
46. N.V. ignatovitch and E.M. Vorobev, Group theoretic analysis of the boundary value problem for the lsminsr
layer equation, Math. Mcdelling 8, 116 (1991).

Вам также может понравиться