equations
Citation for published version (APA):
Brands, J. J. A. M. (1977). Oscillation theorems for second-order functional differential equations. (Eindhoven University of Technology : Dept of Mathematics : memorandum; Vol. 7712). Technische Hogeschool Eindhoven.
Document status and date: Published: 01/01/1977
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
,
TECHNOLOGICAL UNIVERSITY EINDHOVEN
Department of Mathematics
Memorandum 1977-12 august 1977
Oscillation theorems for second-order functional differential eqhations
by J.J.A.M. Brands Technological University Department of Mathematics PO BOX 513, Eindhoven The Netherlands
J.J.A.M. Brands
Technological University, Eindhoven, The Netherlands
ABSTRACT
This paper presents some comparison theorems on the oscillatory behaviour of solutions of second-order functional differential equations.
Here we state one of the main results in a simplified form: Let q, t1,t2
be nonnegative continuous functions on (O,C(» such that t 1 - t
z
is abounded function on [l,~) and t - tl (t) ~ 00 if t ~~. Then
yet)
+ q(t)y(t tl(t»=
0 is oscillatory if and only ifyet)
+ q(t)y(t t2(t» = 0 is oscillatory.
1. INTRODUCTION
Throughout C(A) is the set C+ and C
t are
to 0
this paper we use the following notations and abbreviations:
of all continuous mappings of A into :R, C =C«to'oo»,
to
the sets of all nonnegative and all nonpositive functions in C
t ' respectively. Constant functions are simply denoted by their value.
By Y Se denote the set consisting of all functions y E Co with the property
that jyj is positive and concave (and hence monotonically non-decreasing)
on (0,"") and
I
y( t)I
= o( t), (t ~ ""). We recall that a func tio n y ECo
is called concave if for all a E [O,lJ and all positive numberst
l, t2
we have y(at} + (I - a)t
Z) ~ ay(tl) + (I - a)y(tZ)' By a deZay we mean
a function T €
C~
with the property that t - t(t)~ ~
if t~ ~.
REMARK. With obvious modifications our considerations will remain valid if
OSCILLATION OF DELAY-DIFFERENTIAL EQUATIONS
- 2
-The goal of this paper is to present comparison theorems about the oscillatory behaviour of solutions of second-order functional differential equations. In order to give a clear idea of our method we postpone the general case to section 7 and first treat equations
yet) + q(t)y(t - T(t» • 0 ,
+
where q € Co and T is a delay. We shall refer to this equation by the symbol D
q,T
By a
soLution
of D we mean a continuous function y on some interval q,T(to'oo) where to ~ 0, with the property that there is a number t l, tl ~ to' such that t - T(t) > to if t > tl and such that y is twice continuously differentiable for t > t
l, satisfying
D
q,T for t equation D is called o8cilZato~y if every solution of Dq,T q,T
> t
l • An has
ar-bitrarily large (positive) zeros. A solution y of D is called nont~iviaZ
q,T
if for every r > 0 there is a number t, t > r, such that yet) ~ o. A solution y of D is called a
positive (negative)
solution if there isq,T
a number to' such that yet) > 0
«
0) for t > to.The oscillatory behaviour of special cases of second-order functional differential equations has been already the subject of several papers
([lJ, [3], [5J, [6J, [8J, [9J). The main results of the present paper are of a different nature and seem to be new.
2. MAIN RESULTS
In section 4 we give a proof of the following +
THEOREM I.
Let
ql,q2be functions in
CO'and Let
T1,T2
be delays.
Suppose that for
eve~ypositive
y € Ythere are nonnegative constants
M1,M2,
such that
4for aZl
tsufficientZy Zarge
4the following two
yet) - M2 > 0 ,
Then we have: if
Dis oscillatory then
Dis oscillatory.
ql,t1 q2,T2
Several interesting consequences of theorem I are treated in
section 5. In section 6 we shall show that a condition of the form
DO
f
eut2q(t)dt < "", with u < 2,never guanantees that all nontrivial solutions of
y(
t) + q( t)y( t - 1) .. 0(with q € C~), are nonoscillatory. This result shows that theorem 6
of [6J is incorrect.
Finally, in section 7, we generalize the results of sections 3,4,5
to larger classes of second-order functional differential equations.
3. PRELIMINARIES
In the sequel we need the following lemmas
+
LEMMA 1.
Let
q € Co andlet
Tbe a delay. Suppose that there is a positi7)e
funotion
y €C~o'
whioh is differentiabZe lor
t > t I(where
to~
0 and t 1is
suoh that
t - T(t) > tofor
t >tl),satisfying
co
yet)
~
f
q(8)y(S - T(s»ds,t
Then
Dhas a positive
soZution.v~whiah moreover satisfies
q,T
vet) !> yet), (t > to) and vet) ~ 0, (t ~ 00) •
We have 4 -Vn+1(t) = yet), =
J
t """"
q(s)v (s - t(s»ds nv2(t)
=
f
q(s)y(s - T(s»ds~
yet) .. vI(t) ,t
nOlI 1,2, ••• ) .
It follows that v
2(t) ~ vIet) for t > tl, whence
It ""
f
q(s)v2(s - T(s»ds < "",t
easily follows by induction that:
vn+,(t) ~ v (t) J ,( t > t J ' n == n 0 :s vn+1(t) :s vn(t)
,
(t > t 1 ' 0 ~ vn+1(t) ~v
(t),
(t > t l , n 1,2, ••• ) n =: 1,2, ••• ) n .. 2,3,. •• )We conclude that vn tends to a positive limit function v and that vn
tends to a limit function w. Clearly vet) ~ y(t) , (t > to) and
vet) ~ y(tl) , (t ~ t
I). It is now a routine matter to prove that
v -
w,
and (by Lebesque's theorem on dominated convergence) that""
vet)
=
J
q(s)v(s - T(s»ds ,t
Hence v'is a positive solution of D Obviously vet) ~ 0 (t ~ (0) •
REMARK. Clearly lemma 1 holds for a larger class of L's,namely
LEMMA 2.
Let
to
be a nonnegative number. Let
ube a positive
andconcave
function in
Ct
suah that
u(t) - oCt), (t + =).Then there is a number
tl, tl > to'
2nd a positive funation
y E Ysuoh that
yet)=
u(t), (t 2 t1),-I -1
and moreover
3if
t3 2 t2 2 tl,
then
t2 u(t 2) 2 t3 u(t3).The proof is elementary. The lemma is easily visualized by drawing a diagram.
4. PROOF OF THEOREM 1
We assume that D is oscillatory and D is not. Then there
ql,L1 q2,LZ
exists a solution u of D which has no large positive zeros. Without qZ,LZ
loss of generality we may assume that u is a positive solution. We select numbers t
l, t2, such that u(t) > 0, (t > t1) and t - L2(t) > tl, (t > t2). Clearly u(t) is nonnegative and monotonically non-increasing for t > t
2. Integrating D from t to x where t
z
~ t < x, we getqz
,r
Zx
u( t) - u(x) =
I
q2 (s)u(s - L 2(S) )ds •t
Letting x +~, it follows that
and
J
Q2(s)u(s - L2(s»ds <~
, t co u(t) 2J
Q2(s)u(s - t 2(s»ds, tBy lemma 1 it follows that we may suppose, without loss of generality,
that u(t) +
a
(t + co), Hence, from lemma 2 and the conditions of theorem 1,we infer that there are nonnegative numbers M
6
-with t3 ~ t2 + M
I, such that, for all t > t 3, we have:
Put yet)
=
u(t - MI) - M2• Then
ql(t)y(t-'I(t» =ql(t)(u(t-'l(t) -M I) -M2) sq2(t)u(t-'2(t», (t > t3)·
Hence
t t
By application of lemma 1 it follows that D has a positive solution; q I" I
this contradicts the assumption.
5. APPLICATIONS OF THEOREM I
An immediate consequence of theorem 1 is
THEOREM 2. Let q E
C~.
Let '1' '2 be deZays such that '1 - '2 is boundedon some intel'vaZ [to'co). Then D is osciUatory if and on~y if D is
q"1 Q"2
osci~~atol'Y·
PROOF. Take MI
=
maxI,
1 - L21 ,
M2=
0 and apply theorem 1. t~to+
EXAMPLE 1. Let q E Co and let L be a delay. If D is oscillatory then
'y'(t) ( - 1 q,L
+ 1 + aCt - L(t» )q(t)y(t - L(t»
=
0 is oscillatory for all real a.PROOF. Let y L Y be a pos1t1ve function. If a ~ 0 then trivially
large. If a < 0 then, for all t sufficiently large,
q(t)(y(t - T(t» -
bO»::;;
(I + aCt - t(t»-l)q(t)y(t -ret»~
• +EXAMPLE 2. Let q € CO' Let t be a delay which is bounded on some interval
[ta,oo). Then we have: D
q,T is oscillatory if and only if D q,O is oscillatory. PROOF. Apply theorem 2.
REMARK 1. If in example 2 the condition "t is bounded" is replaced by the condition " t
-I
r (t) -+0,
(t -+ co)" then we get a false statement as is shown by the following example:Let
and
-2 t(t) • t(1 + 2 log t)(1 + log t) •
Then D a is oscillatory (see [4, p 20J), but D q, q,r has the solution yet)
= t!.
+
EXAMPLE 3. Let q E
Co
and let T be a delay. Then we have: Ifyet)
+ (I - t-1T(t»q(t)y(t)=
a is oscillatory thenyet)
+ q(t)y(t - T(t»=
0 is oscillatory.PROOF. Apply lemma 2 and theorem I.
REMARK 2. Observing the proof of theorem 1 we see that the condition in theorem 1 can be weakened as follows: For every positive function y € Y
with the property that 00
I
Q2(t)y(t - T2(t»dt < 00 ,there are nonnegative constants MI , M2 such that, for all t sufficiently large, the following two inequalities hold:
yet) - M2 >
a ,
co 00
J
ql(s)(y(s - rl(s) - M1) - M2)ds ::;;
J
q2(s)y(s - L2(s»ds •8
-We shall not pay attention to this generalization of theorem 1, except for one example.
+
EXAHPLE 4. Let q E Co and let a be a positive number. Then we have:
yet) + q(t)y(t) • 0 is oscillatory if and only if
.. -1
y(t) + (a ft+o q(s)ds)y(t) • 0 is oscillatory.
t
PROOF. Let y E Y be a positive function. Then, changing the order of integration, we easily derive that
.., co s+a
J
q(s)y(s - a)ds ~J
N-l(J
~t+a t s
for all sufficiently large t.
6. A COUNTEREXAMPLE
q(cr)dcr)y(s)ds
co
~
f
q(s)y(s)ds •t
It is well-known (for instance see [4, p 4]) that all nontrivial solutions of
yet) + q(t)y(t)
=
0 ,where q E C;, are nonoscillatory if (1) is not oscillatory. We shall
provide an example showing that such a statement is not true for all equations
yet) + q(t)y(t - 1) • 0 •
Besides, we remark that such a situation is also possible for non-linear equations without delay; for example yet) +
!
t-3y3(t) = 0(1 )
(2)
has both nontrivial oscillatory and nonoscillatory solutions ([7, p 33, 34J). Moreover, in the case of equation (1) one can give conditions, which
tq(t)dt < QOII is such a condition. Our example will show also that
there are no conditions of the form
f
at 2e q(t)dt < QO with a < 2,
which guarantee that all nontrivial solutions of (2) are nonoscillatory. First we give some heuristics. Since (2), and hence (1), (example of section 5) has to be not oscillatory, we try a function q(t) which tends rather fast to zero i f t -+ 00, This implies that the "characteristic
2 -z
equation" of (2) z + qe ... 0 has solutions z with 1m z ~ 'IT. Moreover,
it can be proved that large s'uccessive zeros of a nontrivial oscillatory solution of (2) are less than 1 apart, in the case that (2) is not oscillatory. Therefore it is reasonable to try a solution of the form yet) ...
r(t)ei~(t)
with~(t)
> 'IT.We are now able to start a precise calculation. All asymptotic statements will be with respect to t -+ QO. Substituting in (2)
yet) ... exp(- ~(t) + i'IT(t - Set»~), ~ =
n
and using the abbreviation~f = f(t - I) - f(t), we get
.
n
= -
('IT 2 (I - S) • 2 + 'lTacot 'IT~e) - 2'IT(1 8)cot • 'IT~e • n + n 2 sin 'lT68e-~~q
... 2'IT(1 -e
)n + 'ITS.Several choices for
e
are possible. We shall calculate the case-) 2
Set) ... t • Setting p ...
n -
2t + 2t in (3), we find-2. 2
t p
=
aCt) + 8(t)p + y(t)p ,h a -)
were a, ~, y have asymptotic series in terms of powers of t ,the first
~ew terms being given by
aCt) ~ - 4 + 4t- 1 + •••
,
S(t) ~ 2 - 2t- 1 + ••.,
y( t)=
t -2 (3) (4)- 10
-Hence (5) is a so-called unstable Riccati equation, which has exactly one bounded solution p. This solution has an asymptotic series
which formally satisfies equation (5). For a detailed treatment of this result on Riccati equations see [2, p 184 - 186J. By means of
sub-. 2 -2
stitution we find rO .. 2, r 1 .. O. It follows that wet) '" 2t - 2t + 2 + OCt ), hence
Applying this to (4) we find
4 -2t2 + 4t - 11/3 -1
q(t) .. 4t e (I + OCt
» ,
which corresponds with nonoscillatory equations (I) and (2). Neverthe-less we have found a nontrivial oscillating solution of (2), and, moreover,
< 00 if a < 2.
-n
REMARK. We can get even sharper results. If we choose S (t) .. t , where n is a 2 -1 n+2
positive integer, then we find similarly wet) ~ 2(n + 2n) t ,
( -I n+ I n . ( -t
q t)
=
exp [- 2n t + OCt )J; and ~f we choosee
t)=
e then we get 1JJ(t)~
2et and q(t) .. exp [- (1 - e-l)et(1 + 0(1»].7.
GENERALIZATIONSFirst we introduce same more notation. We shall use Church's lambda notation: A Af(x) means a function with domain A and with function
x€
values given by the expression f (x). For example; if f .. A R x2 then
2 x€
f(x) .. x for all real x, The letter 6 always denotes a variable which runs through all reals of [O,lJ, Therefore we simply write
Ae
insteadWe shall also use some abbreviations. C is the Banach space
+
-C([O,I]) with the sup-norm. C and C are the sets of all non-negative and all nonpositive functions in C, respectively.
G
to is the set of all functions g € C«to'~) x C) satisfying, for all t > to' the conditions:
( i) g(t,O) == 0 (H) if q> I ,(jl2 E:
c ,
+ such that f+l J + g(t,Ql2)'-
Ql 2 € C then g(t,Qll) 2:-(iii) if ipl,Ql2 € C
,
such that Ql1 - Ql2 € C then g(t,'Pj ) s g(t,({lZ)'We write g E G if we do not want to specify to'
Let t1,t
Z
be nonnegative numbers. If g EG
t and if L is a delay 2
such that t - L(t) > tl for t > t
2, then f(t,y) • g(t,A e yet - T(t)e» defines a function a domain including on (t
2,oo) x C . tl The set of all functions
£
with(t2'~) x C , such that the restriction of f to
t}
the form (*) is denoted by F t ' If we do not want t}, 2
to specify t
1,t2 then we write f € F.
We remark that for a given f E F the representation in the form At,yg(t,Ae yet - T(t)e» is not unique.
+
EXAMPLE. Let q € CO' Let f(t,y) • q(t)y(t - I),
Then f(t,y) • g(t,Ae yet - T(t)e» with g(t,q»
=
q(t)q>(l), T(t)=
but also with g(t,Ql)=
q(t)ip(!), .(t)=
2.12
-The class of equations we want to consider are of the form yet) + f(t,y) - 0 ,
where f L F. We shall refer to this equation by the symbol D
f or D g,! . By a solution of D
f we mean a continuous function y on some interval
(to'~)' with to ~ 0 , having the property that there is a number tI,
tl ~ to' such that f € Ft t and such that y is twice continuously
0' 1
differentiable on (tl'~)' satisfying D
f for t > tl• The definitions of "D
f is oscillatory", "nontrivial solution", "positive and negative solutions" are the same as in the introduction. The results are analogous to previous results.
+
LEMMA 1'.
Let
to ~ O.Let
f € F tand "let
Y E Ct
be differentiable
0' t} 0for
t > t l ,satisfying
~yet)
~
f
f(s,y)ds , tThen
Df
has a positive solution
v,which moreover satisfies
v ( t) !:> Y ( t) , (t > to) andv (
t) + 0 (t + co) •The proof is analogous to the proof of lemma 1.
THEOREM I'.
Let
f 1 ,f 2 € Fsimultaneous.ly. Suppose that for every
y € Ythere are non:negative constan,ts
M} ,M2,
such that
lfor aU
tSUfficientLy
Zarge
lthe foLLowing two inequaUties hold:
ly(t)1 - M2 > 0 ,
Ifj(t,Aty(t - M1) - M2 sgn y)1 s; If2(t,y)1 •
(sgn y
equaLs
+1or
-1accopding to whether
yis a positive
or
a negative
function). Then we have: If
Df
is osciZlatory then
Dfis oscillatory.
1 2
The proof is analogous to the proof of theorem 1, but one remark may be useful. The phrase "without loss of generality we may assume
that u is a positive solution" must be changed in "We assume u to be a positive solution. If u is a negative solution then the
same with f
l,f2 instead of f1,f2• (f is defined by f(t,y) Further we assume that £1
=
At,ygl(t,A
e
yet - 't(t)e» £2=
At ,y8 Z(t,y(t - "2(t)6» where gl and g2 in G and "1
delays. I I ,
proof is the
III - f ( t, -y».
and
and "2 are
THEOREM 2'.
Let
g € Gand let
11,12
be
de~ays,suoh that
"1 - 12is
bounded on some interval
[to'~)'Then
0is osoiZlatory if and only
. g, "1
if
0is osoiZlatory.
g,12
Ex&~LE I. Let f E F. Let f E F be defined for all real ~ by
-I (l. f~(t,y) ... f(t,A
t(! + ~t )y(t». Then we have: If Of is oscillatory then Of is oscillatory for all real ~.
~
EXAMPLE 2. Let (l. be a positive number. Let q € c~. Let 1 be a delay.
Then we have: If yet) + q(t)(1 +
~(I
+ Iy(t - t(t)I)-I)y(t - t(t) ... 0 is oscillatory then yet) + q(t)y(t - 1(t» ... 0 is oscillatory.PROOF. Let y € Y. Without loss of generality we may assume that y is
a positive function. We define a constant M as follows: If yet) + L, (t + ~)
-I
then M
=
L(l + a) ; if yet) +=,
(t + 00) then M=
~. Then we have, for all t sufficiently large, that q(t)(1+~(l
+y(t-t<'t» -M)-I)(y(t-t(t) -M)~ q{t)y(t - 1(t» and yet) - M > O. Then we apply theorem I'.
EXAMPLE 3. Let ql,q2 be functions
inc~.
Let 11,12 be delays. Suppose that 12 is bounded on some interval [t
o
'=)' Then we haveis oscillatory if and only if yet) + ql(t)y(t-'t(t» +q2(t)y(t) ... 0 is oscillatory. The latter equation is oscillatory if
.. -I
14
-PROOF. Apply theorem 1 I and lemma 2.
Finally we present a sufficient condition for D to be oscillatory
gtT in the case that T is unbounded.
THEOREM 4.
Let
g E G.Let
Tbe a
de~ay ~hichis differentiabZe on some
interval
(tl'~)and
f(t) < 1for
t > t1•
Then
D g,Tis osciZZatory if
D*
is oscillatory. Here
g*is defined by
g ,0
*
-)
g (t,<p) '" (l-Hh(t») g(h(t),rp), (t > tl ' rp E C) ,where
h E C tis defined by
h(t) - T(h(t» = t. IPROOF, We suppose that D
*
°
is oscillatory and D is not. As in theg , g,T
proof of theorem 1 I we may assume that there exists a positive solution
u of Dg,T' Then there is a number t
z
with t2 > t) such that u(t-,(t» >0if t > t2 and
00
u(t)
~
f
g(s,Aeu(s - T(s)8»ds ,t
Let t3 be such that t - T(t) > t2 if t > t3' Then, since u(t) ~ 0
if t > t
2, we have
00
u(t)
~
I
g(s,AeU(. - T(s»)ds ,t Put s - h(o). Then
s
u(t) ~
f
g(h(o), Aeu(o»(1 - f(h(o») do -1t-T(t) co
~
f
g(h(o),Aeu(o»(l t -I - Hh(o») do,Application of lemma I' completes the proof.
EXAMPLE 4. Let q E
C~.
Then we have: If yet) +2q(2t)y3(t) 0 isoscillatory then yet) +
g(t)y3(~t)
... 0 is oscillatory.REMARK. Applying lemma 2 and theorem I ' we get the result:
- 1 3 . 3
If yet) +
8'
q(t)y (t) ... 0 lS oscillatory then yet) +q(t)y Ot) '" 0is oscillatory. This statement is weaker than the statement of example 3, which can be seen as follows: If we put t ... 2s, y(2s) ... w(s), then the equation yet) +
~
q(t)y3(t) ... 0 is transformed into the equationwI! (s ) + ~ q ( 28 )",,3 ( s) 0: O.
ACKNOWLEDGEMENTS
I would like to thank N.G. de Bruijn and M.L.J. Hautus for their help and valuable suggestions.
REFERENCES
1. Bradley, J.S., Oscillation theorems for a second-order delay equation. J. Di££. Eq. 8 (1970), 397-403.
2. Bruijn, N.G. de, Asymptotic Methods in Analysis. North-Holland Publishing Co. Amsterdam. First Ed. 1958, Third Ed. 1970.
3. Burton, T. and Grimmer, R., Oscillation, continuation and uniqueness of solutions of retarded differential equations. Trans. Amer. Math. Soc. 179 (1973),193-209.
4. Coppel, W.A., Disconjugacy. Lecture Notes in Mathematics 220. Springer-Verlag, Berlin etc. 1971.
5. Gollwitzer, H.E., On nonlinear oscillation for a second-order delay equation. J. Math. Anal. Appl. 26 (1969), 385-389.
6. Kung, G.C.T., Oscillation and nonoscillation of differential equations with a time lag. SIAM. J. Appl. Math. 21 (1971), 207-213.
16
-7. Moore, R.A. and Nehari, Z., Nonoscillation theorems for a class of nonlinear differential equations. Trans. Amer. Math. Soc. 93 (1959),
30-52.
S.
Staikos, V.A. and Petsoulas, A.G., Some oscillation criteria forsecond-order nonlinear delay-differential equations. J. Math. Anal.
Appl. 30 (1970), 695-701.
9. Waltman, P., A note on an oscillation criterion for an equation with
17 -GREEK LETTERS a alpha
a
beta y gamma A capital delta ql phi If psi A lambda•
n eta p rho t taue
theta 0' sigma MATHEMATICAL SYMBOLS(the usual notation for "is element of")