J Eng Math (2014) 89:193–205

DOI 10.1007/s10665-014-9696-3

A class of exact solutions of the Liénard-type ordinary

nonlinear differential equation
Tiberiu Harko · Francisco S. N. Lobo ·
M. K. Mak

Received: 1 August 2013 / Accepted: 11 March 2014 / Published online: 30 August 2014
© Springer Science+Business Media Dordrecht 2014

Abstract A class of exact solutions is obtained for the Liénard-type ordinary nonlinear differential equation. As
a first step in our study, the second-order Liénard-type equation is transformed into a first Abel kind, first order
differential equation. With the use of an exact integrability condition for the Abel equation (Chiellini lemma), the
exact general solution of the Abel equation can be obtained, leading to a class of exact solutions of the Liénard
equation, expressed in parametric form. We also extend the Chiellini integrability condition to the case of the general
Abel equation. As an application of the integrability condition the exact solutions of some particular Liénard-type
equations, including a generalized van der Pol–type equation, are explicitly obtained.

Keywords Abel equation · Exact solutions · Integrability condition · Liénard equation

1 Introduction

The Liénard-type second-order nonlinear differential equation of the form [1,2]

ẍ(t) + f (x)ẋ(t) + g(x) = 0, (1)
where f (x) and g(x) are arbitrary real functions of x, with f (x), g(x) ∈ C ∞ (I ), defined on a real interval I ⊆ R,
and its generalization, the Levinson–Smith-type equation [3]
ẍ(t) + f (x, ẋ)ẋ(t) + g(x) = 0, (2)
where a dot represents the derivative with respect to the time t, and f is a function of x and ẋ, play an important role
in many areas of applied sciences [where Eq. (1) appears as the Rayleigh or van der Pol equation], cardiology (mod-
eling the electric activity of the heart), neurology (propagation of electric pulses in neurons), biology, mechanics,

T. Harko (B)
Department of Mathematics, University College London, Gower Street, London WC1E 6BT, UK
e-mail: t.harko@ucl.ac.uk
F. S. N. Lobo
Centro de Astronomia e Astrofísica da Universidade de Lisboa, Campo Grande, Ed. C8, 1749-016 Lisbon, Portugal
e-mail: flobo@cii.fc.ul.pt
M. K. Mak
Department of Computing and Information Management, Hong Kong Institute of Vocational Education, Chai Wan,
Hong Kong, People’s Republic of China
e-mail: mkmak@vtc.edu.hk

123
194 T. Harko et al.

seismology, chemistry, physics, and cosmology [4–15]. Equation (2) can model, for example, oscillating circuits,
seismic waves, concentration of chemical reactants, and scalar fields in a cosmological background.
A particular type of the general Liénard equation, the van der Pol equation [4–6],
ẍ(t) − μ[1 − x 2 (t)]ẋ(t) + x(t) = 0, (3)
where μ is a positive parameter describing a nonconservative oscillator with nonlinear damping, is extensively
applied in both the physical and biological sciences. In the 1920s the Dutch physicist van der Pol, working as an
engineer for Philips Company, studied the differential equation (3), which describes the circuit of a vacuum tube.
A few years later, the electric activity of the heart rate was modeled using a Liénard-type model [7]. In the 1960s
Fitzhugh [8] and Nagumo et al. [9] extended the van der Pol equation in a planar field as a model for action potentials
of neurons. The van der Pol equation, originally introduced to describe relaxation oscillators in electronic circuits,
has been frequently used in theoretical models of the heart function [10–12]. The van der Pol equation is a useful
phenomenological model for the heartbeat since it displays many features that supposedly occur in a biological
setting, such as complex periodicity, entrainment, and chaotic behavior [10–12].
Most models based on chemical kinetics can be formulated as Liénard-type coupled nonlinear first-order rate
equations in several variables [13]. The first-order approximation for a Liénard system works well near a bifurcation
point, with higher-order terms being required the further the system is from the bifurcation point. The dynamics
of a scalar inflaton field (the cosmological scalar field driving the inflationary phase in the evolution of the early
Universe) with a symmetric double-well potential can also be formulated mathematically as a Liénard system
[14,15]. For this case one can prove rigorously the existence of a limit cycle in its phase space, and, using analytical
and numerical arguments one can show that the limit cycle is stable, and its period can be obtained by an analytical
formula.
Liénard-type equations can also be used to model fluid mechanical phenomena. Linearly forced isotropic turbu-
lence can be described in terms of a cubic Liénard equation with linear damping of the form [16]
ẍ(t) + [ax(t) + b]ẋ(t) + cx(t) − x 3 (t) + d = 0, (4)
where a, b, c, and d are constants, and x(t) is proportional to a uniquely specified similarity length scale [17].
Liénard-type equations also naturally appear in the mathematical description of some important astrophysical
phenomena. For example, the time dependence φ(t) of the perturbations of the stationary solutions of spherically
symmetric accretion processes can be described by a generalized Liénard-type equation of the form [18]
 
φ̈ +  f φ, φ̇ φ̇ + V  (φ) = 0, (5)
where  is a small parameter, and V (φ) is the potential of the system, with the prime indicating the derivative
with respect to φ. Equation (5) is obtained by subjecting stationary solutions of spherically symmetric accretion
processes to time-dependent radial perturbations and including nonlinearities of any arbitrary order.
From a physical point of view, the Liénard equation represents a generalization of the equation of damped
oscillations,
ẍ + γ ẋ + ω2 x = 0, (6)
where γ and ω2 are constant parameters [19]. For γ = 0 we obtain the equation of the linear harmonic oscillator,
which represents one of the fundamental equations of both classical and quantum physics. Generally, a linear
oscillation can be described by the equation ẍ + f (t)ẋ + g(t)x = 0.
The mathematical properties of Liénard-type equations have been intensively investigated from both mathemati-
cal and physical points of view, and their study remains an active field of research in mathematical physics [20–30].
Several methods of integrability, such as the Lie symmetries method [31,32] and the Weierstrass integrability, intro-
duced in [33], were used to study the Liénard equation and the relations between the Riccati and Liénard equations,
respectively. Liénard systems that have a generalized Weierstrass first integral or a generalized Weierstrass inverse
integrating factor were studied in [34].
The present paper aims to introduce some exactly integrable classes of the Liénard equation, Eq. (1), whose
solutions can be obtained in an exact analytical form, and to formulate the integrability condition for this class of

123
Exact solutions of the Liénard-type differential equation 195

differential equations. To obtain the functional form of integrable Liénard-type equations, we reduce them first to
an Abel-type equation of the form [35,36]
y  = p(x)y 3 + q(x)y 2 . (7)
Then we apply to the latter Abel equation an integrability condition, equivalent to the initial Liénard equation, that
was obtained by Chiellini [36,37]. In fact, the Chiellini condition has been recently used in [38–44] for the study
of Abel differential equations and second-order differential equations reducible to an Abel-type equation.
Bandic [38] studied the nonlinear differential equation
y  + ψ(y)y 2 + φ(y)y  + f (y) = 0 (8)
and showed that it could be solved using quadratures if
  
f (y) = F(y) exp −2 ψ(y) dy (9)

and   
φ(y) = G(y) exp − ψ(y) dy , (10)

where F(y) and G(y) are the coefficients of the integrable Abel equation w  = F(y)w 3 +G(y)w 2 . The integrability
of this Liénard-type equation was obtained using the Chiellini condition.
In [42] the differential Chiellini integrability condition was reformulated in an integral form, and the general
form of the solution of the Abel equation was obtained in a simpler form. In [44] the Chiellini integrability condition
of the first-order Abel equation of the first kind dy/dx = f (x)y 2 + g(x)y 3 was extended to the case of the general
Abel equation of the form
dy
= a(x) + b(x)y + f (x)y α−1 + g(x)y α , (11)
dx
where α ∈ R and α > 1.
Several other methods can be used for the integration of the general Abel-type equation
y  = p(x)y 3 + q(x)y 2 + r (x)y + s(x). (12)
If y = y1 (x) is a particular solution of the general Abel equation, then by means of the transformations
E(x)
u(x) = , (13)
y(x) − y1 (x)
where  
E(x) = exp [3 p(x)y12 + 2q(x)y1 + r (x)] dx , (14)

the Abel equation can be transformed into

du 1
+ + 2 = 0, (15)
dx u
where 1 (x) = p(x)E 2 (x), and 2 (x) = [3 p(x)y1 (x) + q(x)]E(x) [36]. Therefore, if y1 = −q(x)/(3 p(x)),
then 2 = 0, and the general solution of the Abel equation can be obtained from the integration of a differential
equation with separable variables.
Thus, it turns out that if the coefficients f (x) and g(x) of the Liénard equation satisfy two specific conditions,
then the general solution of the Lié nard equation can be obtained in an exact analytical form. Some examples of
exactly integrable Liénard equations, of physical interest, are also considered. The generalization of the method to
the case of the Levinson–Smith-type equations of the form (2) is briefly discussed.
The present paper is organized as follows. In Sect. 2, we introduce the Abel equation representation for the
Liénard equation, and we formulate the integrability condition of the first-order Abel equation. In Sect. 3, we obtain
the general solution of Liénard-type equations satisfying the integrability condition of the Abel equation. The exact
solutions of some nonlinear Liénard-type differential equations are obtained in Sect. 4. We discuss and conclude
our results in Sect. 5.

123
196 T. Harko et al.

2 Reduction of Liénard equation to an integrable Abel-type equation

As a first step in our study of the Liénard equation (1) we reduce it to an Abel-type first-order nonlinear differential
equation. Then, using an integrability condition for this equation, which involves a differential relation between
the coefficients f (x) and g(x) of the equations, we obtain the general solution of the Liénard equation in an exact
parametric form.
If we use the notation ẋ = u, the Liénard equation (1) can be written as
du
u + f (x)u + g(x) = 0. (16)
dx
By introducing a new dependent variable v = 1/u, Eq. (16) takes the form of the standard Abel differential equation
of the first kind,
dv
= f (x)v 2 + g(x)v 3 . (17)
dx

2.1 Chiellini integrability condition for reduced Abel equation

In this context, an exact integrability condition for the Abel equation (17) was obtained by Chiellini [37] (see also
[36]) and can be formulated as the Chiellini lemma as follows.
Chiellini Lemma If the coefficients f (x) and g(x) of a first-kind Abel-type differential equation of the form
dv
= f (x)v 2 + g(x)v 3 (18)
dx
satisfy the condition
 
d g(x)
= k f (x), (19)
dx f (x)
where k = constant = 0, then the Abel equation (18) can be exactly integrated.
To prove the Chiellini lemma, we introduce a new dependent variable w defined as [36,37]
f (x)
v= w. (20)
g(x)
Then Eq. (18) can be written as
 
1 d f (x) f (x) dg(x) f (x) dw f 3 (x) 3
− 2 w+ = 2 (w + w 2 ). (21)
g(x) dx g (x) dx g(x) dx g (x)
On the other hand, the condition given by Eq. (19) can be written in an equivalent form as
f (x) dg(x) 1 d f (x) f 3 (x)
− = k . (22)
g 2 (x) dx g(x) dx g 2 (x)
Therefore, Eq. (21) becomes
dw f 2 (x)
= w w2 + w + k , (23)
dx g(x)
which is a first-order separable differential equation, with the general solution given by
 2 
f (x) dw 1
dx = ≡ F(w, k). (24)
g(x) w(w 2 + w + k) k
With the use of condition (19), the left-hand side of Eq. (24) can be written as [42]
 2 
f (x) 1 d g(x) 1 g(x)
dx = ln dx = ln + C0 , (25)
g(x) k dx f (x) k f (x)

123
Exact solutions of the Liénard-type differential equation 197

where C0 is an arbitrary constant of integration. Therefore, the general solution of Eq. (23) is obtained as
g(x)
= C −1 e F(w,k) , (26)
f (x)
where C −1 = exp(−kC0 ) is an arbitrary constant of integration, and
⎧
⎪
⎪ √ w exp − √4k−1
1
arctan √1+2w k > 41 ,
⎪
⎪ w 2 +w+k 4k−1
⎨
= exp 1+2w − 2arctanh(1 + 4w) k = 41 ,
F(w,k) 1
e (27)
⎪
⎪ √ √
⎪
⎪ −1/2 1−4k 1/2 1−4k
⎩√ w 1 − √1−4k
1+2w
1 + √1−4k
1+2w
k < 41 .
2 w +w+k
Equation (26) determines w as a function of x.
The integrability condition given by Eq. (19) can be written as
dg(x) 1 d f (x)
= g(x) + k f 2 (x), (28)
dx f (x) dx
representing a first-order linear differential equation in g. As a function of g, the function f satisfies the differential
equation
1 d f (x) 1 1 dg(x)
= −k f 2 (x) + . (29)
f (x) dx g(x) g(x) dx
To solve Eq. (29), we introduce a new dependent variable f (x) = 1/σ (x), and using the notation σ 2 (x) = ξ(x),
we obtain a first-order differential equation for ξ :
dξ(x) 1 dg(x) 2k
= −2 ξ(x) + . (30)
dx g(x) dx g(x)
Therefore, the Chiellini lemma can be reformulated as follows.

Lemma 1 If the coefficients f (x) and g(x) of the Abel equation (18) satisfy the conditions
  
g(x) = f (x) C1 + k f (x) dx (31)

or
g(x)
f (x) = ±   , (32)
C2 + 2k g(x) dx

where C1 , C2 , and k are arbitrary constants, then the Abel equation is exactly integrable, and its solution is given
by
v(x) = Ce−F(w(x),k) w(x), (33)
where the functions F(w, k) are given by Eqs. (27).

A similar result was obtained in [42].

2.2 Chiellini integrability condition for general Abel equation

The Chiellini lemma can be extended to a general Abel equation of the form
dv
= a(x) + b(x)v + f (x)v 2 + g(x)v 3 , (34)
dx

123
198 T. Harko et al.

where a(x), b(x), f (x), g(x) ∈ C ∞ (I ) are defined on a real interval I ⊆ R, and a(x), b(x) = 0, ∀x ∈ I , as
follows. If we introduce a new function p(x), defined as

v(x) = e b(x)dx
p(x), (35)
Eq. (34) becomes
dp   
= a(x)e− b(x)dx + f (x)e b(x)dx p 2 + g(x)e2 b(x)dx p 3 . (36)
dx
We assume now that the functions b(x), f (x), and g(x) satisfy the condition
  
d g(x)e b(x)dx 
= k1 f (x)e b(x)dx , (37)
dx f (x)
where k1 is an arbitrary constant. Then, if we introduce the transformation
f (x)
p(x) =  s(x), (38)
g(x)e b(x)dx
Eq. (36) becomes
ds g(x) f 2 (x)
= a(x) + s s 2 + s + k1 . (39)
dx f (x) g(x)
Hence, we have obtained the following generalization of the Chiellini lemma.
Lemma 2 If the coefficients of the general Abel equation (34) satisfy the conditions (37) and
f 3 (x)
a(x) = k2 , (40)
g 2 (x)
where k2 is an arbitrary constant, then the Abel equation can be exactly integrated, and its general solution is given
by
f (x)
v(x) = s(x), (41)
g(x)
with s(x) a solution of the equation
 2 
f (x) 1 g(x)e b(x)dx
G 0 (s, k1 , k2 ) = dx = ln + K0, (42)
g(x) k1 f (x)
where K 0 is an arbitrary constant of integration and

ds
G 0 (s, k1 , k2 ) = . (43)
s 3 + s 2 + k1 s + k2
Using Eq. (37), Eq. (42) becomes

g(x)e b(x)dx
= K 1 eG(s,k1 ,k2 ) , (44)
f (x)
where K 1 = exp(−k1 K 0 ) is an arbitrary constant of integration, and G(s, k1 , k2 ) = k1 G 0 (s, k1 , k2 ).

3 A class of exact solutions of the Liénard equation

As we have already seen, the second-order nonlinear Liénard equation (1) can be reduced to an Abel-type equation
of the form given by Eq. (18), with the general solution given by v(x) = C exp(−F(w(x), k))w(x), where w(x)
is determined, as a function of x, by Eq. (26). Alternatively, Eq. (26) fixes x as a function of w,
x = x(w). (45)

123
Exact solutions of the Liénard-type differential equation 199

To find the time dependence of x, we start from

dx dx dw 1 g(x) 1
= =u= = , (46)
dt dw dt v f (x) w
which gives
dw dw g(x) 1
= . (47)
dt dx f (x) w
With the use of Eq. (23), satisfied by the function w(x), we obtain for dw/dt the equivalent expression
dw
= f (x)(w 2 + w + k). (48)
dt
Therefore, we have obtained the following theorem.

Theorem If the coefficients of the Liénard equation (1) satisfy the condition
  
g(x) = f (x) C1 + k f (x) dx (49)

or
g(x)
f (x) = ±   , (50)
C2 + 2k g(x) dx

where C1 , C2 , and k are arbitrary constants, then the general solution of the Liénard equation (1) can be obtained
in an exact parametric form, with w taken as a parameter, as

dw
t − t0 = , x = x(w), (51)
f (x(w))(w 2 + w + k)
with x(w) obtained as a solution of the equation
g(x)
= C −1 e F(w,k) (52)
f (x)
and

dw
F(w, k) = k . (53)
w(w 2 + w + k)
Similar results were previously obtained in [42], where the integral form of the Chiellini integrability condition
were explicitly formulated. A particular integrable case of the Liénard equation can be obtained for the case k = 0.
In this case, the Chiellini condition given by Eq. (19) immediately provides
g(x) = A f (x), (54)
with A an arbitrary constant. Thus, the Liénard equation takes the particular form
ẍ + f (x)ẋ + A f (x) = 0, (55)
with the associated Abel equation given by
dv
= f (x)v 2 (1 + Av), (56)
dx
where ẋ = 1/v. The general solution of Eq. (56) is given by

1 1
f (x)dx = A ln + A − + K 1 , (57)
v v

123
200 T. Harko et al.

where K 1 is an arbitrary constant of integration. Therefore, the general solution of Eq. (55) can be written in a
parametric form, with v taken as a parameter, in the following form:

dv
t − t0 = , x = x(v), (58)
f (x(v))v(1 + Av)
where x = x(v) is the solution of Eq. (57).
In the general solution for the time, given by Eqs. (51) and (58), one can take the arbitrary integration constant t0
as zero, without any loss of generality. The arbitrary integration constant C and the initial value w0 of the parameter
w can be determined from the initial conditions at t = 0 for the position x and the velocity ẋ, given by
x(0) = x0 , ẋ(0) = ẋ0 , (59)
where x0 and ẋ0 are the initial values of x and ẋ at t = 0. Evaluating Eq. (52) for x = x0 we obtain
g(x0 )
= C −1 e F(w0 ,k) , (60)
f (x0 )
while evaluating Eq. (46) at t = 0 gives the equation ẋ0 = [g(x0 )/ f (x0 )]w0−1 , which determines the initial value
of the parameter w0 as
1 g(x0 )
w0 = . (61)
ẋ0 f (x0 )
Once the initial value of the parameter w0 is known, the value of the integration constant C −1 is obtained as
g(x0 ) −F(w0 ,k)
C −1 = e . (62)
f (x0 )

3.1 Integrability condition for Levinson–Smith equation

The procedure for the exact integration of the Liénard-type equations based on the Chiellini lemma can be easily
extended to the generalized Li énard equations of the Levinson–Smith form, given by Eq. (2), if they can be trans-
formed to an Abel-type equation. As a particular case of the integrable Levinson–Smith-type nonlinear differential
equations we consider the equation
ẍ + [γ (x)ẋ 2 + δ(x)ẋ + f (x)]ẋ + g(x) = 0, (63)
where γ (x) and δ(x) are some arbitrary functions of the variable x. Using the notation ẋ = 1/v, Eq. (63) takes the
form of the general Abel equation
dv
= γ (x) + δ(x)v + f (x)v 2 + g(x)v 3 = 0. (64)
dx

If γ (x) = 0, by means of the transformation v(x) = e δ(x)dx h(x), Eq. (64) can be written in the standard form
of the Abel equation
dh
= A(x)h 2 + B(x)h 3 , (65)
dx
 
where A(x) = f (x)e δ(x)dx , and B(x) = g(x)e2 δ(x)dx . If the coefficients A(x) and B(x) of the equation satisfy
the conditions of Lemma 1, then the general solution of Eq. (65) can be obtained through quadratures. If γ (x) = 0,
then from Lemma 2 it follows that if the functions γ (x), δ(x), f (x), and g(x) satisfy the conditions
  
d g(x)e δ(x)dx  f 3 (x)
= k1 f (x)e δ(x)dx , γ (x) = k2 2 , (66)
dx f (x) g (x)
where k1 and k2 are two arbitrary constants, then the generalized Liénard-type equation (63) can be integrated exactly.
Therefore, all the integrability results obtained for the Liénard equation can be applied to the Levinson–Smith-type
equations of the form (63).

123
Exact solutions of the Liénard-type differential equation 201

4 Examples of exactly integrable Liénard-type equations

In this section, we consider some exactly integrable Liénard-type equations that represent the generalizations of
Eqs. (3) and (4). As a first case we assume that the functional form of the function f (x) is known. Then the Chiellini
integrability condition fixes the form of the function g(x) and allows us to find the general solution of the Liénard
equation in an exact parametric form. The case in which the function g(x) is fixed is also considered. Furthermore,
an integrable generalization of the van der Pol equation is also explored.

4.1 First case: f (x) = ax + b

As a first case we assume that the function f (x) is given by

f (x) = ax + b, (67)
where a and b are arbitrary constants. Then from the first integrability condition, given by Eq. (49), we obtain the
function g(x) as
1 2 3 3
g(x) = a kx + abkx 2 + (aC1 + b2 k)x + bC1 , (68)
2 2
where C1 and k are arbitrary integration constants. Therefore, the exactly integrable Liénard equation is given by
1 3
ẍ + (ax + b)ẋ + a 2 kx 3 + abkx 2 + (aC1 + b2 k)x + bC1 = 0. (69)
2 2
As a function of the parameter w, x is determined by Eq. (52), which gives for x the quadratic algebraic equation
ak 2
x + bkx + C1 = C −1 e F(w,k) , (70)
2
which determines x as a function of w as

−bk ± b2 k 2 − 2ak[C1 − C −1 e F(w,k) ]
x(w) = . (71)
ak
The time dependence of x is determined as a function of w as

kdw
t − t0 = ±  . (72)
b2 k 2 − 2ak[C1 − C −1 e F(w,k) ](w 2 + w + k)
Equations (71) and (72) give the exact solution of the Liénard equation (69). Depending on the value of the constant
k, k < 1/4, k = 1/4, and k > 1/4, there are three distinct classes of solutions of this equation.

4.2 Second case: g(x) = cx + d

Secondly, we consider the case in which the function g(x) is fixed. By analogy with the van der Pol equation (3),
we assume that
g(x) = cx + d, (73)
where c and d are arbitrary constants. Then, after determining the function f (x) from the integrability condition
equation (50), we obtain the Liénard equation
cx + d
ẍ ±  ẋ + cx + d = 0. (74)
ckx 2 + 2dkx + C2
Equation (52) gives the equation
ckx 2 + 2dkx + C2 = C −2 e2F(w,k) , (75)

123
202 T. Harko et al.

with the solution


−dk ± d 2 k 2 − ck[C2 − C −2 e2F(w,k) ]
x(w) = . (76)
ck
The parametric time dependence of the solution is obtained as

k e F(w,k) dw
t − t0 = ±  . (77)
C d 2 k 2 − ck[C2 − C −2 e2F(w,k) ](w 2 + w + k)
Equations (76) and (77) give the exact analytic solution, in parametric form, of the Liénard equation (74).

4.3 Third case: generalization of van der Pol equation

Finally, we consider the integrable generalization of the van der Pol equation (3), in which we fix the function f (x)
as f (x) = −μ(1 − x 2 ) and obtain the function g(x) from the integrability condition equation (49). Therefore, the
integrable generalization of the van der Pol equation is given by
1 4
ẍ − μ(1 − x 2 )ẋ + kμ2 x 5 − kμ2 x 3 + C1 μx 2 + kμ2 x − C1 μ = 0. (78)
3 3
The parametric dependence of x is determined from the algebraic equation
1
C1 − kμx + kμx 3 = C −1 e F(w,k) . (79)
3
Equation (79) can be rewritten in the form
x 3 − 3x + H (F) = 0, (80)
where we have used the notation
3[CC1 − e F(w,k) ]
H (F(w, k)) = . (81)
Ckμ
The solution of the algebraic equation (80) is given by
 1/3
21/3 H 2 (F(w, k)) − 4 − H (F(w, k))
x(w) =  1/3 + . (82)
21/3
H 2 (F(w, k)) − 4 − H (F(w, k))
To have a real solution of the cubic equation (80), the conditions
H 2 (F(w, k)) − 4 > 0 (83)
and

H 2 (F(w, k)) − 4 − H (F(w, k)) > 0 (84)
must be satisfied for all w and k.
The parametric time dependence of the solution of the generalized van der Pol equation is obtained as

1 φ 2 (w, k)dw
t − t0 = , (85)
μ [φ (w, k) + φ (w, k) + 1](w 2 + w + k)
4 2

where we have used the notation

21/3
φ(w, k) =  . (86)
( H 2 (F(w, k)) − 4 − H (F(w, k)))1/3
Depending on the numerical values of the parameters k, μ, C1 , C, a large class of dynamical evolutions of the
solutions of the generalized van der Pol equation can be obtained.

123
Exact solutions of the Liénard-type differential equation 203

5 Discussions and final remarks

In this paper we have introduced a class of exactly integrable Liénard and generalized Liénard-type equations. If the
coefficients of the second-order nonlinear equations satisfy some specific conditions that follow from the Chiellini
emma, then the general solution of the Liénard differential equation can be obtained in an exact parametric form.
As an application of the integrability procedure obtained, we have considered some specific examples of exactly
integrable nonlinear differential equations that could be of physical interest. One of these equations, Eq. (69), is
similar in form to Eq. (4) and, in fact, represents the exactly solvable generalization of the equation describing the
linearly forced isotropic turbulence [20]. We have also considered an exactly solvable generalization of the classical
van der Pol oscillator equation, in which higher-order force terms are also present. In all these cases of physical
interest the general solution of the corresponding Liénard equation can be obtained in an exact parametric form.
The existence of an analytical solution may allow a deeper understanding of the highly nonlinear physical processes
that govern most natural phenomena.
The exact solutions also allow us to obtain some approximate solutions of the considered differential equa-
tions, corresponding to small and large values of the parameter w. In the limit of small w, i.e., w k, giving
exp(F(w, k)) ≈ w, Eq. (52) takes the simple form
g(x)
≈ C −1 w, (87)
f (x)
while the parametric time evolution can be obtained as

1 dw
t − t0 ≈ . (88)
k f (x(w))
In the limit of large w, so that w k and w 2 w,
    
dw k
exp(F(w, k)) ≈ exp k = exp − 2 , (89)
w3 2w
and the approximate asymptotic solution of the exactly integrable Liénard equation is given by
g(x)
≈ C −1 e−k/2w
2
(90)
f (x)
and

dw
t − t0 ≈ . (91)
f (x(w))w 2
As an application of the previous asymptotic equations we consider the case f (x) = ax +b, with a, b = constant,
giving
g(x) kax 2
= C1 + + kbx. (92)
f (x) 2
In the limit of small x, by neglecting the x 2 term, we obtain
C −1 w − C1
x(w) ≈ (93)
bk
and
bC aw
t − t0 ≈ ln − aC1 + b2 k , (94)
a C
giving
1 a(t−t0 )/bC b
x(t) ≈ e − . (95)
abk a

123
204 T. Harko et al.

In the limit of large x, so that kbx C1 and ax/2 b, we obtain

g(x) kax 2
≈ , (96)
f (x) 2

2C −1 −k/4w2
x(w) ≈ e (97)
ka
and

dw
t − t0 ≈ √ . (98)
w2 2a/kC e−k/4w + b
2

In the range of the values of w for which


2a −k/4w2
e b, (99)
kC
we obtain
 √ 
πC k
t − t0 ≈ − erfi , (100)
2a 2w
where erfi(z) gives the imaginary error function erfi(z) = erf(iz)/i. In the large time limit the solution of the Lié
nard equation (69) can be obtained only in parametric form.

Acknowledgments We would like to thank the four anonymous referees and the editor for comments and suggestions that helped us
to improve our manuscript.

References

1. Liénard A (1928) Étude des oscillations entreténues. Revue générale de l’électricité 23:901–912
2. Liénard A (1928) Étude des oscillations entreténues. Revue générale de l’électricité 23:946–954
3. Levinson N, Smith O (1942) A general equation for relaxation oscillations. Duke Math J 9:382–403
4. van der Pol B (1927) On relaxation-oscillations. Lond Edinb Dublin Philos Mag J Sci 2:978–992
5. Andronov AA, Leontovich EA, Gordon II, Maier AG (1973) Qualitative theory of second order dynamic systems. Wiley, New York
6. Strogatz SH (1994) Nonlinear dynamics and chaos. Addison-Wesley, Reading
7. van der Pol B, van der Mark J (1928) The heartbeat considered as a relaxation oscillation, and an electrical model of the heart.
Lond Edinb Dublin Philos Mag J Sci 6:763–775
8. Fitzhugh F (1961) Impulses and physiological states in theoretical models of nerve membranes. Biophys J 1:445–466
9. Nagumo J, Arimoto S, Yoshizawa S (1962) An active pulse transmission line simulating nerve axon. Proc Inst Radio Eng 50:2061–
2070
10. Glass L (1990) Theory of heart. Springer, New York
11. Nayfeh A, Balachandran B (1995) Applied nonlinear dynamics. Wiley, New York
12. Edelstein-Keshet L (1988) Mathematical models in biology. Random House, New York
13. Poland D (1994) Loci of limit cycles. Phys Rev E 49:157–165
14. Salasnich L (1995) Instabilities, point attractors and limit cycles in an inflationary universe. Mod Phys Lett A 10:3119–3127
15. Salasnich L (1997) On the limit cycle of an inflationary universe. Nuovo Cimento B 112:873–880
16. Ran Z (2009) One exactly soluble model in isotropic turbulence. Adv Appl Fluid Mech 5:41–47
17. de Karman T, Howarth L (1938) On the statistical theory of the isotropic turbulence. Proc R Soc Lond A 164:192–215
18. Sen S, Ray AK (2012) Implications of nonlinearity for spherically symmetric accretion. arXiv:1207.1070
19. DiBenedetto E (2011) Classical mechanics: theory and mathematical modeling. Birkhäuser/Springer, New York
20. Dumortier F, Rousseau C (1990) Cubic Liénard equations with linear damping. Nonlinearity 3:1015–1039
21. Dumortier F, Kooij RE, Li CZ (2000) Cubic Liénard equations with quadratic damping having two antisaddles. Qual Theory Dyn
Syst 3(1):163–209
22. Cheb-Terrab ES, Roche AD (2000) Abel ODEs: equivalence and integrable classes. Comput Phys Commun 130:204–231
23. Chandrasekar VK, Senthilvelan M, Kundu A, Lakshmanan M (2006) A nonlocal connection between certain linear and nonlinear
ordinary differential equations/oscillators. J Phys A 39:9743–9754
24. Liu XG, Tang ML, Martin RR (2008) Periodic solutions for a kind of Liénard equation. J Comput Appl Math 219:263–275

123
Exact solutions of the Liénard-type differential equation 205

25. Zou L, Chen XW, Zhang WN (2008) Local bifurcations of critical periods for cubic Liénard equations with cubic damping. J
Comput Appl Math 222:404–410
26. Pradeep RG, Chandrasekar VK, Senthilvelan M, Lakshmanan M (2009) Nonstandard conserved Hamiltonian structures in dissi-
pative/damped systems: nonlinear generalizations of damped harmonic oscillator. J Math Phys 50:052901
27. Pandey SN, Bindu PS, Senthilvelan M, Lakshmanan M (2009) A group theoretical identification of integrable cases of the Linard-
type equation ẍ + f (x)ẋ + g(x) = 0. I. Equations having nonmaximal number of Lie point symmetries. J Math Phys 50:082702
28. Pandey SN, Bindu PS, Senthilvelan M, Lakshmanan M (2009) A group theoretical identification of integrable equations in the
Liénard-type equation ẍ + f (x)ẋ + g(x) = 0. II: Equations having maximal Lie point symmetries. J Math Phys 50:102701
29. Banerjee D, Bhattacharjee JK (2010) Renormalization group and Liénard systems of differential equations. J Phys A 43:062001
30. Messias M, Gouveia MRA (2011) Time-periodic perturbation of a Liénard equation with an unbounded homoclinic loop. Physica
D 240:1402–1409
31. Garcia IA, Giné J, Llibre J (2008) Liénard and Riccati differential equations related via Lie algebras. Discr Contin Dyn Syst B
10:485–494
32. Carinena JF, de Lucas J (2011) Lie systems: theory, generalisations, and applications. Dissertationes Mathematicae (Rozprawy
Matematyczne) 479:1–162
33. Giné J, Grau M (2010) Weierstrass integrability of differential equations. Appl Math Lett 23:523–526
34. Giné J, Llibre J (2011) Weierstrass integrability in Liénard differential systems. J Math Anal Appl 377:362–369
35. Polyanin AD, Zaitsev VF (2003) Handbook of exact solutions for ordinary differential equations. Chapman & Hall/CRC, Boca
Raton/Washington
36. Kamke E (1959) Differentialgleichungen: Lösungsmethoden und Lösungen. Chelsea, New York
37. Chiellini A (1931) Sull’integrazione dell’equazione differenziale y  + P y 2 + Qy 3 = 0. Bollettino dell Unione Matematica Italiana
10:301–307
38. Bandic I (1961) Sur le critère d’intégrabilité de l’équation différentielle généralisée de Liénard. Bollettino dell Unione Matematica
Italiana 16:59–67
39. Mak MK, Chan HW, Harko T (2001) Solutions generating technique for Abel-type nonlinear ordinary differential equations.
Comput Math Appl 41:1395–1401
40. Mak MK, Harko T (2002) New method for generating general solution of Abel differential equation. Comput Math Appl 43:91–94
41. Harko T, Mak MK (2003) Relativistic dissipative cosmological models and Abel differential equation. Comput Math Appl 46:849–
853
42. Mancas SC, Rosu HC (2013) Integrable dissipative nonlinear second order differential equations via factorizations and Abel
equations. Phys Lett A 377:1234–1238
43. Mancas SC, Rosu HC (2013) Integrable Ermakov–Pinney equations with nonlinear Chiellini ‘damping’. arXiv:1301.3567
44. Harko T, Lobo FSN, Mak MK (2013) A Chiellini type integrability condition for the generalized first kind Abel differential equation.
Univers J Appl Math 1:101–104

123