See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254465100

Solution of a Class of Riccati Equations

Article · January 2005

CITATIONS READS

5 6,715

2 authors, including:

Krishna Busawon
Northumbria University
259 PUBLICATIONS   2,477 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modelling of solar PV Module View project

Polarization switching properties View project

All content following this page was uploaded by Krishna Busawon on 25 September 2014.

The user has requested enhancement of the downloaded file.

Proceedings of the 8th WSEAS International Conference on APPLIED MATHEMATICS, Tenerife, Spain, December 16-18, 2005 (pp334-338)

Solution of a Class of Riccati Equations

K. BUSAWON, P. JOHNSON
School of Computing, Engineering and Information Sciences
Northumbria University
Ellison Building, Newcastle-upon-Tyne, NE1 8ST
UNITED KINGDOM
krishna.busawon@unn.ac.uk

Abstract: In this paper, we derive the analytical closed form solution of a class of Riccati equations. Since it is
well known that to every Riccati equation there corresponds a linear homogeneous ordinary differential equation
(ODE), the result obtained is subsequently employed to derive the analytical solution of the class of second order
linear homogeneous ODEs corresponding to the class of Riccati equations considered. In addition, we present
some examples in order to demonstrate the validity of the results obtained.

1 Introduction the transformation (4) does not make it any easier to

find the solution of (2). However, as Euler was able
A large class of dynamical systems appearing through- to demonstrate, when one particular solution of (3)
out the field of applied mathematics are described by is obtained, the general solution can be readily found
ODEs of order 2 of the following form: [11]. The difficulty in using (3) to obtain the solu-
tion of (2) therefore resides in the problem of finding a
ẍ(t) + p(t)ẋ(t) + q(t)x(t) = r(t) (1)
particular solution of (3). It ought to be stressed that
dx d2 x when p(t) and q(t) assume some special form, then
where ẋ = , ẍ = 2 , and p(t), q(t) and r(t) are real
dt dt the general solution of (3) can be found. For exam-
valued, scalar functions of the independent variable t. ple, when q(t) ≡ 0, the above Riccati equation reduces
It is not obvious to perform a complete analysis of to the Bernoulli differential equation and thus can be
the equation (1) since there is no universal method reduced to a first order linear ODE by means of the
for solving the linear inhomogeneous ODE of order 1
transformation y(t) = which yields
2. There do however exist well established methods v(t)
for solving (1) in a number of special cases, one of
v̇(t) = p(t)v(t) − 1
which being when p(t) and q(t) are constant [1]. A
survey of the various existing methods (analytical and and whose solution is given by
otherwise) for solving the ODE (1) is given in many t
p(τ )dτ
t
p(τ )dτ
t λ
−p(τ )dτ
reference textbooks (see e.g. [1]-[10]). v(t) = v(t0 )e t0
−e t0
e t0
dλ
t0
On the other hand, it is well known that solving
the ODE (1) in the homogeneous case when r(t) = 0, where t0 is an initial value of t.
amounts to solving a Riccati equation (see e.g. [11]). In other words, when q(t) ≡ 0 the solution of (3)
More precisely, the following second order homoge- can be explicitly written as
t
neous ODE − p(τ )dτ
y(t0 )e t0
y(t) = λ . (5)
ẍ(t) + p(t)ẋ(t) + q(t)x(t) = 0 (2) t −p(τ )dτ
1 − y(t0 ) t0
e t0
dλ
can be transformed into the following Riccati equation In this paper, we treat a special case of the Riccati
2
ẏ(t) + p(t)y(t) − y (t) = q(t) (3) equation (3) where the function q(t) is not identically
zero for all t. More precisely, we consider the case
by making the following change of variable: where the function q(t) is positive for all t ≥ t0 and is
given by
ẋ(t)
y(t) = − . (4) −2
t
p(τ )dτ
x(t) q0 e t0
q (t) = λ 2
In general, it is not at all a simple matter to solve (3) √ t − p(τ )dτ
for arbitrary functions p(t) and q(t). This implies that 1 + K q0 t0
e t0
dλ
Proceedings of the 8th WSEAS International Conference on APPLIED MATHEMATICS, Tenerife, Spain, December 16-18, 2005 (pp334-338)

where q0 = q (t0 ) > 0 and K is a constant. We shall and the functions θn (t) ; n = 1, 2, 3 are given by
give the explicit solution of the Riccati equation (3)
t
when q(t) is as above. Since q (t) depends on p(t), we θn (t) = cn + q (τ )dτ (9)
shall assume, throughout the paper, that the integral t0
t
t0
p(τ )dτ exists. We shall also use the result obtained
to provide the solution of the linear second order ODE where
corresponding to the Riccati equation considered. As √
1 2y0 + q0 (K − α)
far as the authors are aware, the explicit solution of c1 = ln √ (10)
α 2y0 + q0 (K + α)
the class of ODEs considered here does not exist in the √
2 2y0 + K q0
literature. This lays the foundation for the analysis for c2 = tan−1 √ (11)
that particular class of ODEs in terms of stability, ex- β β q0
√
istence of singularities, existence of periodic solutions 2 q0
c3 = − √ . (12)
and so on. 2y0 + K q0
An outline of the paper is as follows: In the next
section, the solution of a class of Riccati equations and
is given along with an example. This result is then
α = K2 − 4
used to derive the solution of the class of second order
linear homogeneous ODEs corresponding to (3) and β = 4 − K 2.
a further two examples are provided to illustrate the
method. Finally, some conclusions are drawn.
Before giving the proof of the above theorem, we
shall make a few remarks on the functions q(t) and
f (t).
2 Solution of a Class of Riccati
Remarks
Equations
1. First of all, note that if q0 = 0 then q(t) = 0 for
In this section, we give the general solution of a class all t ≥ t0 , and, as mentioned in the introduction,
of Riccati equations. The result is summarised in the in this case the solution is given by (5).
following theorem:
2. Next, it can be shown that q (t), which is given
by
Theorem 1 Consider the Riccati equation
t
ẏ(t) + p(t)y(t) − y 2 (t) = q(t) (6) √ − p(τ )dτ
q0 e t0
q (t) = λ ,
with the initial condition y(t0 ) = y0 for some ini- √ t −p(τ )dτ
1 + K q0 t0
e t0
dλ
tial value t0 . Assume that q(t0 ) = q0 > 0 and that
t
the integral t0 p(τ )dτ exists. Assume further that is the solution of the following Bernoulli equation
the function q(t) satisfies the relation in q(t) :
t
−2 p(τ )dτ .
q0 e t0
q(t) = −p(t) q(t) − Kq (t)
q (t) = λ 2
√ t −p(τ )dτ 2
1 + K q0 t0
e t0
dλ = −p(t) q(t) − K q (t) (13)

for some constant K. Then, the general solution with initial condition q (t0 ) = q0 .
of (6) is given by
3. By differentiating the function (8), it can be
y(t) = f (t) q(t) (7) shown after some lengthy calculations that
where the function f (t) is given by
f˙ (t) = f 2 (t) + Kf (t) + 1 q(t). (14)
 αθ1 (t)

 K α 1+e

 − + if K 2 > 4

 2 2 1 − eαθ1 (t) In other words, (8) is the solution of the above dif-

K β βθ2 (t) ferential equation with some specific initial conditions.
f (t) = − + tan if K 2 < 4

 2 2 2 Proof - Theorem 1:



 K 1 Let us first demonstrate that
 − − if K 2 = 4
2 θ3 (t)
(8) yp (t) = f (t) q (t) (15)
Proceedings of the 8th WSEAS International Conference on APPLIED MATHEMATICS, Tenerife, Spain, December 16-18, 2005 (pp334-338)

is, at least, a particular solution of (6). By differenti- of Theorem 1 we can say that the general solution of
ating (15) we have (17) is
.  √ √ a 2 2 
ẏp (t) = f˙ (t) q(t) + f (t) q (t) . (16) 3 5 1 + e 5( 2 (t −t0 )+c1 )
y (t) = − + √ a  at
2 5( 2 (t2 −t20 )+c1 )
2 1−e
Next, by considering the above remarks and by
substituting (14) and (13) into (16) we obtain √
2y0 +at0 (3− 5)
where c1 = √1 ln √ .
5 2y0 +at0 (3+ 5)
ẏp (t) = f 2 (t) q(t) + q(t) − p(t)f (t) q(t)
= yp2 (t) + q (t) − p (t) yp (t)

or alternatively
3 Solution of a Class of 2nd Or-
der Linear Homogeneous Sys-
ẏp (t) + p (t) yp (t) − yp2 (t) = q (t) .
tems
Hence, it is clear that (15) is indeed a particular solu-
As mentioned in the introduction, to every Riccati
tion of (6).
equation there corresponds a linear 2nd order homoge-
Now, the general solution of (6) is given by
neous equation. In particular, to the Riccati equation
y(t) = yp (t) + u(t) (3) there corresponds the homogeneous linear ODE
(2). We can therefore use the solution of the previ-
1 ous class of Riccati equations to derive the solution of
where u(t) = and v(t) satisfies
v(t) its corresponding 2nd order ODE. This is given in the
following corollary:
v̇(t) + (2yp (t) − p(t)) v(t) = −1.
Corollary 1. Consider the initial value problem
More precisely,
t
ẍ(t) + p(t)ẋ(t) + q(t)x(t) = 0 (18)
(2yp (τ )−p(τ ))dτ
u0 e t0
with given initial conditions x(t0 ) = x0 and ẋ(t0 ) =
u(t) = λ

1−
t
u0 t0 e t0
(2yp (τ )−p(τ ))dτ
dλ ẋ0 . Assume that q(t0 ) = q0 > 0 and that the integral
t
t0
p(τ )dτ exists. Assume further that the function
q(t) satisfies the relation
where u0 = u(t0 ). If c1 , c2 and c3 are all chosen as in
t
(10-12), it can be shown that yp (t0 ) = y0 . As a result, −2 p(τ )dτ
q0 e t0
u0 = 0. It is clear from the above that if u0 = 0, then q (t) = λ 2
u(t) ≡ 0. Consequently, if c1 , c2 and c3 in (8) are all √ t −p(τ )dτ
1 + K q0 t0
e t0
dλ
chosen as in (10-12), then yp (t) becomes the general
solution of (6) i.e. y (t) = yp (t). for some constant K. Then the general solution to (18)
Q.E.D. is given by:

Remark  1 − eαθ1 (t)
e( 2 − 2 )(θ1 (t)−c1 ) if K 2 > 4
K α

 x0
Note that for K < 0, the function q(t) and hence 
 1 − e αc1
 βθ2 (t)
the solution x(t) might possess some singularities. cos
x(t) = x0 2 K
e 2 (θ2 (t)−c2 ) if K 2 < 4

 cos
βc2
This will be illustrated by an example in the next sec- 
 2

 x θ3 (t) e K2 (θ3 (t)−c3 )
tion. 0 c3 if K 2 = 4

Example 1. Consider the initial value problem where c1 , c2 and c3 are given by (10-12) in which y0 =
ẋ0
− .
1 x0
ẏ − 3at + y − y 2 = a2 t2 (17)
t
Proof - Corollary 1:
where y (t0 ) = y0 , t0 = 0 and where a is some constant. Employing the transform (4) in (18) yields (6).
Now, the Riccati equation (17) is of the form (6) with Now, by Theorem 1 we know the solution of (6) is
1 given by (7) since we are assuming that (13) holds.
q (t) = a2 t2 and p (t) = − 3at + . It is clear that Solving (4) for x it is clear that
t
q (t0 ) > 0. In addition, it can be shown that q (t) t t √
− y(τ )dτ − f (τ ) q(τ )dτ
satisfies the relation (13) with K = 3. From the result x (t) = x0 e t0 = x0 e t0 (19)
Proceedings of the 8th WSEAS International Conference on APPLIED MATHEMATICS, Tenerife, Spain, December 16-18, 2005 (pp334-338)

and hence the solution of (18) is (19) where f (t) is of Corollary 1 we can say that the general solution of
ẋ0 (20) is
given by (8) and where y0 = − .
x0 √
To determine the explicit form of the solution of 3 1
a ln t
t0 +c2
(18) it is necessary to determine the following integral cos 2
1
t0 2a
x (t) = x0 √
t cos 3c2 t
2
I (t) = f (τ ) q(τ )dτ .
t0
where c2 = √23 tan−1 − 2at√0 ẋ3x0 +x0 . Note that the
t 0
To do this we set u = q(τ )dτ . Then, func-
t0 solution possess a singularity at t = 0. The above
tions θn (t) ; n = 1, 2, 3 described in (9) can be written solution holds as long as t > 0 because of the ln tt0
term appearing in the solution.
θn (t) = cn + u

The function f (t) can therefore be written as a Example 3. Consider the initial value problem
function of u as follows:
ẍ − b + 2aebt ẋ + a2 e2bt x = 0 (21)


 K α 1 + eα(c1 +u)

 − + if K 2 > 4 where x (t0 ) = x0 , ẋ (t0 ) = ẋ0 for some initial value

 2 2 1 − eα(c1 +u)
 t0 and a and b are constants. The form of (21) is
f (u) = K β β (c2 + u)
 − + tan if K 2 < 4 equivalent to that of (18) with q (t) = a2 e2bt and

 2 2 2

 K 1 p (t) = − b + 2aebt . Again, it is a simple matter to

 − − if K 2 = 4
2 c3 + u show that q (t) satisfies the relation (13) with K = 2.
Corollary 1 shows that the solution of (21) is
As a result,
a
ebt − ebt0 + c3 bt
−ebt0 )
e b (e
a
b
t x (t) = x0
I(t) = f (u) du c3
t0
2x0 aebt0
and from this simplified form it is possible to show, where c3 = 2ẋ0 −Kx0 aebt0
.
after some lengthy calculations, that

 α−K 1−eαc1
 (θ1 (t) − c1 ) + ln if K 2 > 4 4 Conclusion

 2 1−eαθ1 (t)
βc
cos 22
I (t) = −K (θ2 (t) − c2 ) + ln 2
if K < 4 In this paper, we have derived the analytical solution

 2 cos
βθ2 (t)


2

−K
2 (θ 3 (t) − c3 ) + ln
c3
θ3 (t) if K 2 = 4 of a class of linear homogeneous ODEs of order 2. We
have first given the solution of a particular class of
This implies that Riccati equations which is then used to characterise
 the solution of a fairly large class of homogeneous sec-
 1 − eαθ1 (t) ond order linear ODEs. Examples were given in each
e( 2 − 2 )(θ1 (t)−c1 )
K α

 x0 2
if K > 4

 1 − e αc1 case to demonstrate the methods at hand. The results
 βθ2 (t)
x(t) = cos K obtained will provide a better insight on the class of
 x0 2
e 2 (θ2 (t)−c2 ) if K 2 < 4

 cos
βc2
dynamical systems considered in terms of stability and


2
 x θ3 (t) e K2 (θ3 (t)−c3 ) 2
if K = 4 phase plane analysis.
0 c3

Q.E.D.
References
Example 2. Consider the initial value problem
[1] R. Bronson, Schaum’s Outline of Theory and
1 1 1
ẍ + 1 + ẋ + 2 2 x = 0 (20) Problems of Differential Equations 2nd Edition,
a t a t
McGraw-Hill Inc., 1994.
where x (t0 ) = x0 , ẋ (t0 ) = ẋ0 with t0 > 0 and [2] D. Zwillinger, Handbook of Differential Equations,
a is some constant. By inspection it can be seen Academic Press Inc., 1989.
that, (20) is of the form (18) with q (t) = a21t2 and
p (t) = 1 + a1 1t . It is trivial to show that q (t) sat- [3] D.L. Kreider et al, An Introduction to Linear
isfies the relation (13) with K = −1. From the result Analysis, Addison-Wesley Inc., 1966.
Proceedings of the 8th WSEAS International Conference on APPLIED MATHEMATICS, Tenerife, Spain, December 16-18, 2005 (pp334-338)

[4] A. D. Polyanin and V. F. Zaitsev, Handbook of

Exact Solutions for Ordinary Differential Equa-
tions 2nd Edition, Boca Raton, Chapman &
Hall/CRC Press, 2003.

[5] A.C. King at al, Differential Equations: Linear,

Nonlinear, Ordinary, Partial, Cambridge Univer-
sity Press, 2003.

[6] J.C. Butcher, The Numerical Methods for Ordi-

nary Differential Equations, John Wiley and Sons
Ltd., 2003.

[7] R.M. Mattheij and J. Molenaar, Ordinary Differ-

ential Equations in Theory and Practice, SIAM,
2002.

[8] J. Hubbard and B. West, Differential Equations:

A Dynamical Systems Approach: Ordinary Dif-
ferential Equations, New York, Springer-Verlag,
2002.

[9] P. Hartman, Ordinary Differential Equations,

SIAM, 2002.

[10] E. Hairer et al, Solving Ordinary Differential

Equations I, New York, Springer-Verlag, 1987.

[11] H.T. Davis, Introduction to Nonlinear Differential

and Integral Equations, New York, Dover Publi-
cations Inc., 1962.

View publication stats