Вы находитесь на странице: 1из 7

ON SOME BOUNDED AND STABLE

SOLUTIONS OF A NON-AUTONOMOUS
LIENARD EQUATION
Juan E. Npoles Valdes1;2 ,
Luciano M. Lugo Motta Bittencurt1 and Samuel I. Noya1
September 1, 2014
Abstract
In this note using a bounded region we obtain su cient conditions
under which we can guarantee the boundedness and stability of solutions
of a non-autonomous Linard equation.

UNNE, FaCENA, Av. Libertad 5450, (3400) Corrientes, Argentina;


UTN, FRRE, French 414, (3500) Resistencia, Chaco, Argentina;
jnapoles@exa.unne.edu.ar; jnapoles@frre.utn.edu.ar
2

AMS Subject Classication (2010): 34C11


Key words and phrases: Boundedness, stability, Linard equation.

Preliminars

The Liapunovs Second Method (o Direct) has long been recognized as the
most general method for the study of the stability of equilibrium points of
systems described by dierential equations. The method was rst presented in
his classical memoir, which appeared in Russian in 1892 and was translated into
French in 1907 and English in 1949 [9]. Statements and proofs of mathematical
results underlying the method and numerous examples and references can be
found in the books of Antosiewicz [1], Barbashin and Krasovskii [2], Cesari [3],
Demidovich [4], Hahn [6] and Yoshizawa [17] and bibliography listed therein.
First we consider the system
x0 = f (t; x);

(1)

where f 2 C(IxSr ) and f (t; 0) 0. With I we denote the interval 0 t <


+1, Rm will stand for Euclidean m-space, k:k will be an arbitrary norm in Rm
and Sr = fx 2 Rm : kxk < rg.
This method is a powerful tool because his simplicity for the research of the
stability. For instance, if within a neighborhood of the equilibrium point an
appropriate "energy" function is always decreasing we expect the equilibrium
to be asymptotically stable. The Second Method generalizes this idea, if such
function V (t; x), called Liapunov Function, can be constructed for the system
(1) in a neighborhood of the equilibrium point and if in that neighborhood
V 0 (t; x) 0 for x 6= 0 being V (t; x) positive dened, then the equilibrium point
is asymptotically stable. If one knows only that V 0 (t; x) 0 for x 6= 0, then, in
general, one can conclude only the origin is stable.
The main di culty is to construct a suitable V (t; x), this requires experience
and technique. So, many times it may be di cult to construct a Liapunov
Function which satises all above conditions, in such cases it is often useful to
prove the boundedness and stability properties by using qualitative methods
based in a bounded closed set. We apply this to the well-know non-autonomous
Linard equation
x00 + f (x)x0 + a(t)g(x) = 0;

(2)

under suitable assumptions. The classical Linard equation


x00 + f (x)x0 + g(x) = 0;

(3)

appears as simplied model in many domains in science and engineering [10].


It was intensively studied during the rst half of 20th century as it can be used
to model oscillating circuits or simple pendulums. In the simple pendulum case,
f and g represents the friction and acceleration terms. One of the rst models
where this equation appears was introduced by Balthasar van der Pol [15,16],
considering the equation
x00 + (x2

1)x0 + x = 0;

(4)

for modeling the oscillations of a triode vacuum tube. The Linard equation, which is often taken as the typical example of nonlinear self-excited vibration problem, can be used to model resistor-inductor-capacitor circuits with
nonlinear circuit elements. It can also be used to model certain mechanical systems which contain the nonlinear damping coe cients and the restoring force
or stiness. Moreover, some nonlinear evolution equations (such as the BurgersKorteweg-de Vries equation) which arise from various physical phenomena can
also be transformed to equation (3). Therefore, the study of equation (3) is
of physical signicance. We recommend [5] for other references about more
applications.
The main purpose of this note is to present qualitative methods rather than
to obtain general results of boundedness and stability properties of solutions
of (2), without making uses of common conditions. By this, we need some

clarications. V (t; x) denote an arbitrary Liapunovs Function dened on an


open set S
IxRm with continuous partial derivatives with respect to all
arguments, corresponding to V (t; x); we dene the function
0
V(1)
(t; x) := Lim sup
h!0+

V (t + h; x + hf (t; x))
h

V (t; x)

called the total derivative of V (t; x) for system (1). Under the above conditions,
@V
@V
+
f (t; x):
(5)
@t
@x
We now mention two theorems which will play an important role in the
proofs of our main results.
Theorem A. Suppose that there exists a Liapunovs function V (t; x) dened
on IxSr satisfying the following conditions:
1) V (t; 0) 0,
2) a(x) V (t; x); where a(x) is a positive denite function and continuous
increasing function on R.
0
3) V(1)
(t; x) 0:
Then the trivial solution x(t) 0 of (1) is stable.
For the proof, see [17, Theorem 8.1].
Theorem B (Barbashin and Krasovskii). If there exists a function
V(t,x) which is everywhere positive denite, radially unbounded, decreasing,
and whose total derivative (4) for system (1) is negative denite, then the solution x(t) 0 of (1) is asymptotically stable in the whole.
For the proof, see [4, p. 248].
0
(t; x) =
V(1)

Results

In (2) consider that f : R ! R is a continuous function, g : R ! R is a locally


Zx
lipschitzian continuous function, with G(x) = g(s)ds, and a : [0; 1) ! R is a
0

function of class C 1 and such that 0 < a a(t) A < +1.


For a continuous function M (t) dene M (t)+ = maxfM (t); 0g and M (t) =
maxf M (t); 0g. It is clear that M (t) = M (t)+ M (t) .
Theorem 1. Suppose that
a) g(x)F (x) > 0, for 0 < jxj < k;
b) G(x) < l, implies jxj < k.
Then, every solution x(t) of (2) are bounded and the trivial solution is stable.
Rx
Proof. Letting F (x) = 0 f (s)ds and y = x0 we obtain an equivalent system

to equation (2):

x0 = y

F (x);
(6)

y0 =

a(t)g(x):

Here a suitable Liapunov Function is


V (t; x; y) =

y2
+ G(x):
2a(t)

(7)

Then along the solutions of (5) we have


0
V(3)
(t; x; y)

a0 (t)+ 2
y
2a2 (t)

g(x)F (x):

(8)

Let l and m arbitrary positive numbers and consider the region D dened
by the inequalities
l < W (x; y) < l and (y + F (x))2 < m2
2

y
where W (x; y) = 2a
+ G(x). The conditions a) and b) assure us that the
0
region D is bounded (see also [8] or [11]) and that within D, V(3)
(t; x; y) 0.
We select l and m so large that (x0 ; y0 ) will be in the interior of D. We shall
show than any solution (x(t); y(t)) with initial conditions (x0 ; y0 ) cannot leave
the bounded region D, so we can obtain that all solutions of (2) are bounded,
since (x(t); y(t)) is arbitrary.
Is clear that in order to leave D, the solution (x(t); y(t)) must either cross
the locus of points determined by W (x; y) = l or one of loci determined by
y +F (x) = m. We can select m su ciently large that the part of y +F (x) = m
which is the boundary of D corresponds to x > 0 and the part of y+F (x) = m
0
corresponds to x < 0. Since V(3)
(t; x; y) 0 and V (t; x; y) W (x; y) a solution
starting inside cannot cross W (x; y) = l.
Now

d
2
(y + F (x)) = 2(y + F (x))a(t)g(x):
dt
Along the part of y + F (x) = m or y + F (x) = m which makes up the
boundary of D, we have
d
2
(y + F (x)) = 2am(y + F (x)) jg(x)j < 0:
dt
Hence, the solution (x(t); y(t)) cannot cross outside of D through the part
of the boundary determined by y + F (x) = m. Therefore, every solution of
y2
(2) is bounded for t 0. From this and taking a(x; y) = 2A
+ G(x) we obtain
the desired result under Theorem A.

Remark 1. A di culty in using the Theorem A often is that one can construct
a Liapunov Function satisfying all assumptions but not the existence of function
a(x,y). We illustrate in the next result that it may then be easier to establish the
boundedness of the solutions as a separate problem. So, we continue our work
with a weaker assumption on g(x) and a stronger assumption on the damping,
but taking 0 < a a(t) < +1.

Theorem 2. Under hypotheses


a) xg(x) > 0, for x 6= 0.
b) f (x) > 0, for x 6= 0.
c) jG(x)j ! 1 as jxj ! 1,
then all solutions satises jx(t)j < M , jx0 (t)j < M , where M may depend on
the solution.
Proof. We use the same Liapunov Function as before. From (4) and con0
ditions a), b) and c) we obtain again that V(3)
(t; x; y) 0. So, as above every
solution of (2) is bounded for t 0.
Denition 1. By F(I) we denote the class ofR non decreasing continuous func1 du
= 1.
tions ' on I such '(u) > 0 for all u 2 I and 0 '(u)
Theorem 3. Under conditions
a) 0 < a a(t) A < +1,
b) xg(x) > 0 for x 6= 0;
c) f (x) > 0 for all x,
d) G(x) ! 1, when x ! 1:
Then the trivial solution x 0 of the system (5) is asymptotically stable in
the whole.
Proof. Let V (t; x; y) as in (5) and let
U (t; x) :=

V (t;x;y)

dr
;
(r)

where
2 F(R+ ); then U (t; x) satises the conditions of the BarbashinKrasovski Theorem (Theorem B above), and the proof of the theorem is complete.
Remark 2. The simple case
x00

2x0 + x = 0;

(9)

with unbounded solution x(t) = et , shows that positivity of f is probably necessary


in some sense. This is an open problem.
Remark 3. If we take f and g strictly increasing, our results are consistent with
those of [12], mainly Theorem 2 a).

Remark 4. Finally we give examples of functions f(x) which show that our
results contains those in [13] and [14] refer to the boundedness and stability of
solutions of equation (2).
x; if jxj 1;
Example 1. f (x) =
1
x
; if jxj > 1:
8
1;
if x 1;
<
x; if jxj < 1;
Example 2. f (x) =
:
1; if x
1:
These examples do not satisfy the conditions of Repilado and Ruiz, but they
are covered by results of this paper.
Remark 5. When a(t) 1 our results improve the obtained in [7] and those of
[8] refer to classical Linard Equation (3).

References
[1] Antosiewicz, H. A.,A survey of Lyapunovs second method contributions to
the theory of noninear oscillations IV in Annals of Math. Studies No.44,
Princeton Univ. Press, Princeton, NJ, 1958.
[2] Barbashin, E. A.,Liapunovs Functions, Science Publishers, Moscow, 1970
(Russian).
[3] Cesari, L., Asymptotic behavior and stability problems in ordinary di erential equations, Springer-Verlag, Berlin, 1959.
[4] Demidovich, B. P.,Lectures on the Mathematical Theory of the Stability,
Science Publishers, Moscow, 1967 (Russian).
[5] Guckenheimer, J. and P. Holmes, Nonlinear oscillations, dynamical systems, and bifurcations of vector elds, volume 42 of Applied Mathematical
Sciences. Springer-Verlag, New York, 2002. Revised and corrected reprint
of the 1983 original.
[6] Hahn, W., Theory and Application of Liapunovs Direct Method, PrenticeHall, Englewood Clis, New Jersey, 1963.
[7] Hasan, Y. Q. and L. M. Zhu, The bounded solutions of Linard equation,
J. of Applied Sciences 7(8): 1239-1240, 2007.
[8] LaSalle, J. P., Some extensions of Liapunovs second method, IRE Transaction on Circuit Theory, Dec. 1960, 520-527.
[9] Liapounov, A. M., Problme gnral de la stabilit du mouvement, in Annals
of Math. Studies No.17, Princeton Univ. Press, Princeton, NJ, 1949.
[10] Linard, A., tude des oscillations entretenues, Revue Gnerale de
llectricit 23: 901912, 946954 (1928).

[11] Miller, R. K. and A. N. Michel, Ordinary Di erential Equations, New York:


Lawa State University, 1982.
[12] Npoles Valdes, Juan E., On the boundedness and asymptotic stability in
the whole of Linard equation with restoring term, Revista de la Unin
Matemtica Argentina, 41(4) 2000, 47-59 (Spanish).
[13] Repilado, J. A. and A. I. Ruiz, On the behavior of solutions of equation
x+f(x)x+a(t)g(x)=0 (I), Revista Ciencias Matemticas, University of Havana, VI (1), 1985, 65-71 (Spanish).
[14] Repilado, J. A. and A. I. Ruiz, On the behavior of solutions of equation
x+f(x)x+a(t)g(x)=0 (II), Revista Ciencias Matemticas, University of
Havana, VII (3), 1986, 35-39 (Spanish).
[15] Van der Pol, B., On oscillation hysteresis in a triode generator with two
degrees of freedom, Phil. Mag (6) 43, 700719 (1922).
[16] Van der Pol, B., On relaxation-oscillations, Philosophical Magazine,
2(11):978992, 1926.
[17] Yoshizawa, T., Stability theory by Liapunovs second method, The Mathematical Society of Japan, 1966.

Вам также может понравиться