#### A singular initial value problem for some functional differential equations

Hindawi Publishing Corporation
Journal of Applied Mathematics and Stochastic Analysis
RAVI P. AGARWAL 0
DONAL O'REGAN 0
OLEKSANDR E. ZERNOV 0
0 Donal O'Regan: Department of Mathematics, National University of Ireland , Galway , Ireland E-mail address:
For the initial value problem tr x (t) = at + b1x(t) + b2x(q1t) + b3tr x (q2t) + ?(t, x(t), x(q1t), x (t), x (q2t)), x(0) = 0, where r > 1, 0 < qi 1, i ? {1, 2}, we find a nonempty set of continuously differentiable solutions x : (0, ?] ? R, each of which possesses nice asymptotic properties when t ? +0. In this paper, we will consider singular initial value problems for a class of functional differential equations (FDEs). This problem has received very little attention in the literature to date; we refer the reader to [2, 3, 5, 7] where the question of solvability (in various spaces) and the number of solutions have been discussed. We remark that, even now, the asymptotic properties of solutions of FDEs are only partially understood. Our approach to this problem is essentially different from others known in the literature. In particular, we use qualitative methods [1, 7, 8] together with standard fixed point theorems [6]. Our technique leads to the existence of continuously differentiable solutions with nice asymptotic properties. where t ? (0, ?) is a real variable, x : (0, ?) ? R is a real unknown function, r, a, b1, b2, b3, q1, q2 are constants, r > 1, b1 = 0, 0 < qi 1, i ? {1, 2}, and ? : ? R is a continuous
+ ? t; x(t); x q1t; x (t); x q2t
1. Introduction
2. Main result
Consider the initial value problem
tr x (t) = at + b1x(t) + b2x q1t + b3tr x q2t
function; here
? t, y1, y2, y3, y4 ? ? s, y1, y2, y3, y4
? t, y1, y2, y3, y4 ? ? t, z1, z2, z3, z4
=
t, y1, y2, y3, y4 : t ? (0, ?), y1 < ?1t,
y2 < ?1q1t, y3 < ?2t1?r , y4 < ?2 q2t 1?r .
In what follows, we will assume that the following conditions are satisfied:
|a| < ?1 b1 + b2q1 ,
Further, we will assume that
where c = ?a(b1 + b2q1)?1, ? : (0, ?) ? (0, +?) is a continuously differentiable function
and, moreover,
l1(?)|t ? s|,
t, y1, y2, y3, y4 ?
, s, y1, y2, y3, y4 ?
, 0 < ?
t, 0 < ?
s,
l2(t) y1 ? z1 + y2 ? z2 + tr y3 ? z3 + tr y4 ? z4 ,
t, y1, y2, y3, y4 ?
, t, z1, z2, z3, z4 ?
,
where l1 : (0, ?) ? (0, +?) is a continuous nonincreasing function, l2 : (0, ?) ? (0, +?) is
a continuous nondecreasing function, limt?+0 l2(t) = 0, and
(2.3)
(2.4)
(2.5)
(2.6)
(2.7)
(2.8)
lim ?(t) = 0,
t?+0
0
tl?im+0 tr?1 ?(t) ?1 = L1,
tl?im+0 t? (t) ?(t) ?1 = L2,
Li < +?, i ? {1, 2}.
Definition 2.1. For any ? ? (0, ?), a continuously differentiable function x : (0, ?] ? R is
said to be ?-solution of problem (2.1), (2.2) if
(1) (t, x(t), x(q1t), x (t), x (q2t)) ? for all t ? (0, ?];
(2) x identically satisfies the differential equation (2.1) for all t ? (0, ?].
We denote by (?, M) the set of all continuously differentiable functions u : (0, ?] ? R
such that |u(t) ? ct| Mt?(t), t ? (0, ?]; here ?, M are (positive) constants, ? < ?.
The main result of this paper is the following theorem.
Theorem 2.2. There exist constants ?, M such that
(a) if b1 > 0, then problem (2.1), (2.2) possesses an infinite set of ?-solutions x : (0, ?] ?
R, each of which belongs to (?, M). Moreover, for any constant ? which satisfies the
condition |? ? c?| < M??(?), there exists at least one ?-solution x? ? (?, M) such
that x?(?) = ?;
0 < ? < 1miin9 ?i,
where all the ?i are defined below (we choose ? small enough, M and Q large enough so
that the choice of Q, M, and ? ensures the validity of all the reasoning given below).
Let be the space of continuously differentiable functions x : [0, ?] ? R with the
norm
Let
?
be such that for every element u ?
, u : [0, ?] ? R satisfies the inequalities
x
= max
t?[0,?]
x(t) + x (t) .
u(t) ? ctr+1
Mtr+1?(t),
u (t) ? (r + 1)ctr
QMt?(t), t ? (0, ?],
(b) if b1 < 0, then problem (2.1), (2.2) possesses at least one ?-solution x : (0, ?] ? R
which belongs to (?, M).
Proof. We can choose constants Q and M (cf. (2.6)) so that
2 b1 < Q < q2r?1 b3 ?1
Next, we choose ? such that
and u(0) = 0, u (0) = 0; moreover,
?? > 0, ?u ?
, ?ti ? [0, ?], i ? {1, 2} : t1 ? t2
?(?) =?
u t1 ?u t2
where ?(?) = ?(8B(t?))?1; here B(t?) = l1(t?) + t??r , t? is a constant such that t? ? (0, ?),
and if t ? (0, t?], then the following inequalities are satisfied:
QMt?(t)
? ,
33
(1 + r)|c|tr
?
33 .
It is easy to see that is a closed, bounded, and convex set. Moreover,
(use the Arzela? theorem).
There exists a (small enough) ?1 > 0 such that if ? < ?1, then
is a compact set
t, u(t)t?r , u q1t q1t ?r , u (t)t?r ? ru(t)t?r?1, u q2t q2t ?r ? ru q2t q2t ?r?1 ?
(2.15)
for all u ?
, t ? (0, ?].
(2.9)
(2.10)
(2.11)
(2.12)
?,
(2.13)
(2.14)
Let x = yt?r , where y : (0, ?) ? R is a new unknown function. Then we obtain the
following initial value problem:
y (t) = at + b1 y(t)t?r + r y(t)t?1 + b2 y q1t q1t ?r
+ b3q2?r y q2t ? rb3q2?r?1t?1 y q2t
+ ? t, y(t)t?r , y q1t q1t ?r , y (t)t?r ? r y(t)t?r?1,
(2.16)
y q2t q2t ?r ? r y q2t q2t ?r?1 ,
Now we will consider the initial value problem
y (t) = at + b1 y(t)t?r + r y(t)t?1 + b2u q1t q1t ?r
+ b3q2?r u q2t ? rb3q2?r?1t?1u q2t
+ ? t, u(t)t?r , u q1t q1t ?r , u (t)t?r
? ru(t)t?r?1, u q2t q2t ?r ? ru q2t q2t ?r?1 ,
y(0) = 0,
where u ?
is an arbitrary but fixed function. Let
0 = (t, y) : t ? (0, ?], y ? R .
In 0, for (2.17), conditions for the existence and uniqueness theorem [4] and conditions
for the continuous dependence of the initial data theorem [4] are fulfilled.
Now we denote
Let the function A1 : 0 ? [0, +?) be defined by the equality
?1 = (t, y) : t ? (0, ?], y ? ctr+1
= Mtr+1?(t) ,
1 = (t, y) : t ? (0, ?], y ? ctr+1 < Mtr+1?(t) ,
H = (t, y) : t = ?, y ? c?r+1 < M?r+1?(?) .
A1(t, y) = y ? ctr+1 2 tr+1?(t) ?2
and let a1 : 0 ? R be the derivative of the function A1. Using (2.17), we have
a1(t, y) = 2 tr+1?(t) ?2t?r
b1 ? tr?1 1 + t? (t) ?(t) ?1
y ? ctr+1 2
+ y ? ctr+1 ?1(t) ,
(2.17)
(2.18)
(2.19)
(2.20)
(2.21)
(2.22)
b3
Mtr+1?(t) q1 b2 + Q qr?1 +
2
b3 ? 1 c L1 + 1
M
+ ?1(t) ,
tl?im+0 ?1(t) = 0.
There exists a (small enough) ?2 > 0 (see our choice of M at the beginning of the proof)
such that if ? < ?2, then
?1(t) < Mtr+1?(t) b1 , t ? (0, ?].
Since Mtr+1?(t) = |y ? ctr+1| when (t, y) ? ?1, it is easy to see that
sign a1(t, y) = sign b1
when (t, y) ? ?1.
(1) Let b1 > 0 and therefore a1(t, y) > 0 when (t, y) ? ?1. We prove that the integral
curve of (2.17) which intersects ?1 at an arbitrary point (t0, y0) for sufficiently small
|t ? t0| (t ?) lies in 1 if t < t0 and lies outside of 1 if t > t0. In fact, let P(t0, x0) be an
arbitrary point belonging to ?1 and let JP : (t, yP(t)) be the integral curve of (2.17) which
passes through the point P. Then
A1 t0, yP t0
= M2,
a1 t0, yP t0
> 0.
Therefore, if t0 ? (0, ?), then there exists ? > 0 such that
where
But this means that (t, yP(t)) lies in 1 if t ? (t0 ? ?, t0) and (t, yP(t)) lies outside of 1 if
t ? (t0, t0 + ?). If t0 = ?, then there exists ? > 0 such that
A1 t, yP(t) < A1 ?, yP(?) , t ? (? ? ?, ?),
yP(t) ? ctr+1 tr+1?(t) ?1 < M, t ? (? ? ?, ?),
and this means that (t, yP(t)) lies in 1, t ? (? ? ?, ?). It follows from this that each of the
integral curves of (2.17) which intersect H remains in 1 for all t ? (0, ?]. We consider
an arbitrary point G(?, yG) ? H and let Ju : (t, yu(t)) be the integral curve of (2.17) such
that yu(?) = yG. It is easy to see that
yu(t) ? ctr+1
Mtr+1?(t), t ? (0, ?],
yu(t) ? (r + 1)ctr
2 b1 + ?2(t) Mt?(t),
tl?im+0 ?2(t) = 0.
There exists a (small enough) ?3 > 0 such that if ? < ?3 (see our choice of Q at the
beginning of the proof), then
Now we set
(b) If ti ? [t?, ?], i ? {1, 2}, and |t1 ? t2|
(small enough) ?4 > 0 such that if ? < ?4, then
Moreover, there exists a (small enough) ?5 > 0 such that if ? < ?5, then
Therefore,
?. Moreover, since q2ti ? [0, ?], i ? {1, 2}, and
we have, at the same time, |u (q2t1) ? u (q2t2)|
?. Thus,
q2t1 ? q2t2 = q2 t1 ? t2
q2?(?)
?(?),
yu t1 ? yu t2
8? + 58? + B t? ?(?) = 78? < ?.
(2.33)
(2.34)
(2.35)
(2.36)
(2.37)
(2.38)
(2.39)
(2.40)
(2.41)
(c) If t1 ? [0, t?], t2 ? [t?, ?], or vice versa, and |t1 ? t2|
longing to [0, t?], t?, t2 belonging to [t?, ?], and |t? ? t2|
Therefore, in view of (2.35) and (2.41),
.
(2) Let b1 < 0 and therefore a1(t, y) < 0 when (t, y) ? ?1. Similarly, as in the case b1 > 0,
it can be proved that if an integral curve of (2.17) intersects ?1 at an arbitrary point
(t0, y0), then for sufficiently small |t ? t0| (t ?), the curve indicated lies in 1 if t > t0
and it lies outside of 1 if t < t0. It follows that at least one of the integral curves of (2.17)
which intersect H is defined for all t ? (0, ?] and lies in 1 if t ? (0, ?]. In fact, if an
integral curve of (2.17) intersects ?1, then it has to intersect H. Let the mapping ? : ?1 ?
H be defined in the following way: the point ?(P) ? H is assigned to the point P ? ?1
if both of these points belong to a common integral curve of (2.17). We denote ?(?1) =
{?(P) : P ? ?1}. The set H\?(?1) is a nonempty one (H is a closed set, but ?(?1) is a
nonclosed set since ?(?1) is the image of the nonclosed set ?1). Let Ju : (t, yu(t)) be an
integral curve of (2.17) such that (?, yu(?)) ? H\?(?1). Then Ju : (t, yu(t)) cannot have
common points with ?1. Therefore, Ju : (t, yu(t)) is defined for all t ? (0, ?] and it lies in
1 if t ? (0, ?]. As in the case b1 > 0, we can prove that the inequalities (2.31), (2.33) are
fulfilled, and we can introduce notation (2.34) and prove that condition (2.13) is fulfilled.
Thus, yu ? .
Now we will prove in the case b1 < 0 that if t ? +0, then all integral curves of (2.17)
leave the set 1\{(0, 0)}, with one exception, Ju : (t, yu(t)). Let b1 < 0; we denote
?2(?) = (t, y) : t ? (0, ?], y ? yu(t) = ?tr+1?(t)(? ln t) ,
2(?) = (t, y) : t ? (0, ?], y ? yu(t) < ?tr+1?(t)(? ln t) ,
where ? is a parameter, ? ? (0, 1]. Let the function A2 : 0 ? [0, +?) be defined by the
equality
A2(t, y) = y ? yu(t) 2 tr+1?(t)(? ln t) ?2
and let a2 : 0 ? R be the derivative of the function A2. Using (2.17), we have
a2(t, y) = 2 tr+1?(t)(? ln t) ?2t?r y ? yu(t) 2 b1 ? tr?1 1 + t? (t) ?(t) ?1 + (ln t)?1
(2.43)
(2.44)
There exists a (small enough) ?6 > 0 such that if ? < ?6, then a2(t, y) < 0 when (t, y) ?
0, y = yu(t). In particular, a2(t, y) < 0 when (t, y) ? ?2(?) for any ? ? (0, 1]. Therefore,
for any ? ? (0, 1], an integral curve of (2.17) which intersects ?2(?) in an arbitrary point
(t0, y0), for sufficiently small |t ? t0| (t ?), lies in 2(?) if t > t0 and lies outside of 2(?)
if t < t0 (the proof is similar to the one for ?1). Let P?(t?, y?) ? 1\{(0, 0)}, y? = yu(t?).
Then there exists ?? ? (0, 1] such that P? ? ?2(??). As above, the integral curve of (2.17)
J? : (t, y?(t)), which passes through P?, lies outside of 2(??) if t ? (t?, t?), where (t?, t?)
is the left maximal existence interval for the solution y?. On the other hand, if (t, y) ?
1\{(0, 0)}, then
y ? yu(t)
y ? ctr+1 + yu(t) ? ctr+1
2Mtr+1?(t) < ??tr+1?(t)(? ln t)
(2.46)
if t ? (0, t??), where t?? ? (0, ?) and t?? is small enough. Thus, if (t, y) ? 1\{(0, 0)}
and t ? (0, t??), then (t, y) ? 2(??). Let t? = min{t?, t??}. As above, the integral curve
J? : (t, y?(t)) lies outside of 1\{(0, 0)} if t ? (t?, t?) and we have proved our statement.
Now we can introduce an operator T : ? by (Tu)(t) = yu(t). We prove that T :
? is a continuous operator. Let ui ? , i ? {1, 2}, be arbitrary functions and let
Tui = yi, i ? {1, 2}. If u1 = u2, then y1 = y2. Let u1 ? u2 = d, d > 0. We will investigate
the behavior of the integral curves of the differential equation
y (t) = at + b1 y(t)t?r + r y(t)t?1 + b2u1 q1t q1t ?r
+ b3q2?r u1 q2t ? rb3q2?r?1t?1u1 q2t
+ ? t, u1(t)t?r , u1 q1t q1t ?r , u1(t)t?r
? ru1(t)t?r?1, u1 q2t q2t ?r ? ru1 q2t q2t ?r?1 .
(It is obvious that y1 : (0, ?] ? R is the solution of (2.47).) We introduce the following
notations:
?3 = (t, y) : t ? (0, ?], y ? y2(t) = ?d? tr+1?(t) 1?? ,
3 = (t, y) : t ? (0, ?], y ? y2(t) < ?d? tr+1?(t) 1?? ,
where ? = (r + 2)?1, ? = 3|b1|?1(2M)1??(|b2| + 1) + 1.
Let the function A3 : 0 ? [0, +?) be defined by the equality
A3(t, y) = y ? y2(t) 2 tr+1?(t) ?2(1??)
and let a3 : 0 ? R be the derivative of the function A3. Using (2.47), we have
a3(t, y) = 2 t1+r ?(t) ?2(1??)t?r
b1 +
? 1 + ?(1 + r)
? (1 ? ?)t? (t) ?(t) ?1 tr?1
y ? y2(t) 2
+ y ? y2(t) ?3(t) ,
and there exists a (small enough) ?7 > 0 such that if ? < ?7, then
?3(t)
(2M)1??
b2 + 1 d? t1+r ?(t) 1??
since
u1(t) ? u2(t)
u1 qit ? u2 qit
u1(t) ? u2(t)
(1) Let b1 > 0 and therefore a3(t, y) > 0 when (t, y) ? ?3. Then an integral curve of
(2.47) which intersects ?3 in an arbitrary point (t0, y0), for sufficiently small |t ? t0|
(t ?), lies in 3 if t < t0 and lies outside of 3 if t > t0. (The proof is similar to the
one for ?1 in the case b1 > 0.) In addition, y1(?) = y2(?) = yG. Thus, if t decreases from
t = ? to t = 0, then the integral curve of (2.47), J : (t, y1(t)), lies in 3 for all t ? (0, ?]
because this integral curve cannot intersect ?3. We have
y1(t) ? y2(t)
y1(t) ? y2(t)
?d? tr+1?(t) 1??, t ? (0, ?],
b1 + rtr?1 ? + (2M)1??
b2 + 1 t(1+r)(1??)?r ?(t) 1??d?.
Here 1 ? ? > 0, (1 + r)(1 ? ?) ? r = (r + 2)?1 > 0. Thus, there exists a (small enough)
?9 > 0 such that if ? < ?9, then
y1(t) ? y2(t) + y1(t) ? y2(t)
d?, t ? (0, ?].
(2) Let b1 < 0 and therefore a3(t, y) < 0 when (t, y) ? ?3. Then an integral curve of
(2.47) which intersects ?3 in an arbitrary point (t0, y0), for sufficiently small |t ? t0|
(t ?), lies in 3 if t > t0 and lies outside of 3 if t < t0. (The proof is similar to the
one for ?1 in the case b1 < 0.) In addition,
y1(t) ? y2(t)
y1(t) ? ctr+1 + y2(t) ? ctr+1
2Mtr+1?(t) < ?d? tr+1?(t) 1??
(2.56)
when t ? (0, t(d)], where t(d) ? (0, ?) is small enough. Therefore, if t ? (0, t(d)], then the
integral curve of (2.47), J : (t, y1(t)), lies in 3. As above, if t increases from t = t(d) to
t = ?, then the integral curve of (2.47), J : (t, y1(t)), cannot intersect ?3 and therefore this
curve lies in 3 for all t ? (0, ?]. Further, we can get (2.54) and (2.55) as in the case b1 > 0.
Thus, if u1 ? u2 = d < ?(?) = (?/2)1/?, then
max
t?[0,?]
y1(t) ? y2(t) + y1(t) ? y2(t)
The reasoning given above does not depend on the choice of ? > 0 and ui ? , i ?
{1, 2}. Therefore T : ? is the continuous operator. To complete the proof of the
theorem, it suffices to apply the Schauder fixed point theorem to the operator T : .
?
Ravi P. Agarwal: Department of Mathematical Sciences, Florida Institute of Technology,
Melbourne, FL 32901, USA
E-mail address:
Oleksandr E. Zernov: South Ukrainian State Pedagogical University, Odessa, Ukraine
Advances in
ns Research
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
Advances in
Journal of
Algebra
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
bability
and
Statistics
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
The Scientiifc
World Journal
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
International Journal of
Combinatorics
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
Submit your manuscr ipts
Differential Equatio
Journal of
Mathematics
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
En
gineering
Hindawi Publishing Corporation
ht p:/ www.hindawi.com
Discrete
Nature
and Society
International
Journal of
Mathematics
and
Mathematical
Sciences
Journal of
Discrete Mathematics
ht p:/ w w.hindawi.com
Journal of
B. P. Demidovic?, Lectures on the Mathematical Theory of Stability , Izdat. ?Nauka?, Moscow, 1967 .
L. J. Grimm , Analytic solutions of a neutral differential equation near a singular point , Proc. Amer. Math. Soc . 36 ( 1972 ), no. 1 , 187 - 190 .
L. J. Grimm and L. M. Hall , Holomorphic solutions of singular functional differential equations , J. Math. Anal. Appl . 50 ( 1975 ), no. 3 , 627 - 638 .
P. Hartman , Ordinary Differential Equations, John Wiley & Sons, New York, 1964 .
A. I. Shindyapin , A boundary value problem for a singular equation , Differ. Uravn . 20 ( 1984 ), no. 3 , 450 - 455 (Russian).
D. R. Smart , Fixed Point Theorems, Cambridge University Press, London, 1974 .
A. E. Zernov , On the solvability and asymptotics of the solutions of a functional-differential equation with a singularity, Ukrain . Mat. Zh . 53 ( 2001 ), no. 4 , 455 - 465 (Russian). , Qualitative analysis of the implicit singular Cauchy problem , Ukrain. Mat. Zh . 53 ( 2001 ), no. 3 , 302 - 310 (Russian).
Volume 2014 Journal of Hindawi Publishing Corporation ht p:/ www .hindawi.com