R E S E A R C H
Open Access
Oscillatory properties of half-linear difference
equations: two-term perturbations
Simona Fišnarová
**Correspondence:
fi[email protected] Department of Mathematics, Mendel University in Brno, Zem ˇed ˇelská 1, Brno, CZ-613 00, Czech Republic
Abstract
We consider the nonoscillatory half-linear difference equation
(
rk(
xk)
)
+ck(xk+1) = 0,
(x) :=|x|p–1sgnx,p> 1,
and we study the influence of the perturbations˜r,c˜on the oscillatory properties of the equation
(rk+˜rk)
(
xk)+ (ck+ck˜ )
(xk+1) = 0.
The presented oscillation and nonoscillation criteria are obtained using the variational principle and the so-called modified Riccati technique.
1 Introduction
In this article, we study oscillatory properties of the second-order half-linear difference equation of the form
L[xk] :=
rk(xk)
+ck(xk+) = , (x) :=|x|p–sgnx, p> , ()
wherer,care real-valued sequences,rk= . Ifp= , then () reduces to the linear Sturm-Liouville difference equation
(rkxk) +ckxk+= . ()
The basic qualitative theory of () has been established in the article [] and is summarized in the books [, ]. Many oscillatory properties of () are very similar to that of (), however the absence of the linearity requires sometimes to use different methods in half-linear case. In this article, we deal with the so-called perturbation principle. We suppose that equa-tion () is nonoscillatory and thathis a solution of () and we give conditions under which the perturbed equation
˜
L[xk] :=
(rk+r˜k)(xk)
+ (ck+c˜k)(xk+) = , ()
whererk+r˜k= , is oscillatory or nonoscillatory. Similar problem has been studied in [, ], where the caser˜k= has been considered. We extend some results of those papers to
the general caser˜k= and we also show that the assumptionlimk→∞rkhk(hk) =∞ considered in [] can be replaced by alternative conditions. We are motivated also by the results of [, ], where the two-term perturbations of the half-linear differential equation
r(t)x+c(t)(x) = , r(t) >
are studied.
The article is organized as follows. In the next section, we recall the basic methods of oscillation theory for (), in particular the variational principle and the Riccati technique. Section is devoted to the so-called modified Riccati technique. In Section , we present the main results of this article, the oscillation and nonoscillation criteria for the perturbed equation () and in the last section, we show how the results can be applied to the per-turbed equation of the Euler type.
2 Preliminaries
Oscillatory properties of () are defined using the concept of the generalized zero. We say that a solutionxof () has ageneralized zeroin an interval (m,m+ ] ifxm= and rmxmxm+≤. Equation () is said to bedisconjugateon an interval [m,n] if any solution
of () has at most one generalized zero on (m,n+ ] and the solution for whichxm= has no generalized zero on (m,n+ ]. Consequently, equation () is said to benonoscillatoryif there existsm∈Nsuch that this equation is disconjugate on [m,n] for everyn>m. In the opposite case, () is said to beoscillatory.
One of the basic methods used to investigate (non)oscillation of () is the variational technique which relates nonoscillation of () to a positivity of a certainp-degree func-tional.
Lemma [] Equation () is nonoscillatory if and only if there exists m∈Nsuch that
F(y,m,∞) :=
∞
k=m
rk|yk|p–ck|yk+|p
>
for every nontrivial sequence y∈U(m), where
U(m) :=y={yk}∞k=;yk= ,k≤m,∃n>m:yk= ,k≥n
.
The second basic method used is the Riccati technique which is based on the relation-ship between nonoscillation of () and the solvability (in a neighborhood of infinity) of the Riccati-type equation
R[wk] :=wk++ck–
rkwk
(–(r
k) +–(wk))
= , ()
where–is the inverse function of,i.e.–(s) =|s|q–sgns,q:= p
p–. Indeed, ifxis a
solution of () such thatxk= on some discrete interval [m,∞), thenwk=rk(xk/xk) is a solution of () on [m,∞). More precisely, we have the following equivalent statements.
(i) Equation () is nonoscillatory.
(ii) There exists a solutionwof () such thatrk+wk> for largek.
(iii) There exists a solutionwof the Riccati inequalityR[wk]≤such thatrk+wk> for
largek.
If () is nonoscillatory, then there exists a solutionwkof () such thatrk+wk> for large k. Among all solutionswwith this property, there is the so-calledminimalsolutionw˜ for whichw˜k<wkon some interval [m,∞), whererk+wk> andrk+w˜k> . The minimal solutionw˜ can be constructed as follows. Let () be disconjugate on [n,∞) and letN>n. Denote byxN the solution of () which satisfiesxN
N= ,xNN+= and letwN=r(xN/xN)
be the solution of () associated withxN. Then
˜ wk= lim
N→∞w N
k for everyk∈[n+ ,∞). ()
For details of this construction see []. The solutionx˜of () which is associated with the minimal solutionw˜of () by the substitutionw˜=r(x˜/x˜), is called therecessivesolution of (). The recessive solution is defined uniquely up to the multiplication by a real constant. Next, we formulate a comparison statement for minimal solutions of two Riccati equa-tions. The Riccati equation associated with () is
˜
R[wk] :=wk++ck+˜ck–
(rk+r˜k)wk
(–(r
k+r˜k) +–(wk))
= . ()
Lemma []
(i) Let () be nonoscillatory and letr˜k≤,c˜k≥for largek. Further, letw˜ andw¯ be the
minimal solutions of the corresponding Riccati equations () and (), respectively. Then there existsm∈Zsuch thatw˜k≤ ¯wkfork∈[m,∞).
(ii) Ifck≥and ∞rk–q=∞, then the minimal solution of () satisfiesw˜k≥for
largek.
Note that the statement (ii) of the previous lemma is a special case of the statement (i). Condition ∞r–k q=∞implies that the recessive solution of the equation(rk(xk)) = is a constant sequence, hence the minimal solution of the associated Riccati equation is
w= and this is compared with the minimal solution of ().
3 Modified Riccati equation and related results
In this section, we suppose that equation () is nonoscillatory, byhwe denote a positive so-lution of this equation and suppose that both the coefficientsrk,rk+r˜kare positive for large k. This sign restriction is needed when proving inequalities (), () and estimate () be-low, for details see []. Since these estimates play the crucial role in the (non)oscillation criteria based on the modified Riccati technique presented in this section, we suppose that
rk> ,rk+r˜k> for largekthroughout the whole Section . Denote
define the function
H(k,v) :=v+ (rk+r˜k)hk+(hk) –
(rk+˜rk)(v+Gk)hpk+
(hqk–(r
k+r˜k) +–(v+Gk))
, ()
and consider the so-called modified Riccati equation
Rm[vk] :=vk+
˜rk(hk)
+c˜k(hk+)
hk++H(k,vk) = . ()
We show that there is a relation between operators given in () and () and estimate the functionH(k,v) by a term involvingv, which, in turn, enables us to compare Riccati equation associated with () with the Riccati equation related to a certain linear equation.
Lemma Let w be a sequence such that rk+˜rk+wk= and suppose that vk=hpkwk–Gk. Then
Rm[vk] =hpk+R˜[wk].
Proof By a direct computation
vk=hpk+wk+–hpkwk–
(rk+r˜k)(hk)
hk+– (rk+˜rk)|hk|p
and
H(k,vk) =hpkwk+ (rk+˜rk)|hk|p–
(rk+˜rk)hpkh p k+wk
(hqk–(r
k+˜rk) +–(hpkwk))
=hpkwk+ (rk+˜rk)|hk|p–
(rk+˜rk)hpk+wk
(–(r
k+˜rk) +–(wk)) .
Hence,
vk+
(rk+r˜k)(hk)
hk++H(k,vk) =hpk+wk+–
(rk+r˜k)hpk+wk
(–(r
k+r˜k) +–(wk)) .
Adding the termhpk+(ck+c˜k) to both sides of the last equality, we obtain
vk+hk+L˜[hk] +H(k,vk) =hpk+R˜[wk],
whereL˜ is the operator defined in (). Sincehis a solution of (), we have the required
identity.
Lemma Let H(k,v)be defined in ().
(i) It holdsH(k,v)≥forv> –(rk+˜rk)hk((hk) +(hk))with the equality if and
only ifv= .
(ii) Suppose thathkhk> fork∈[m,∞)and denoteRk=q(rk+r˜k)hkhk+|hk|p–.
Then we have the following inequalities forv≥andk∈[m,∞):
(Rk+v)H(k,v)≥v, p∈(, ], ()
(iii) Suppose thatlim infk→∞Gk> , then for largek:
and since the first condition in () holds, we have
–
Consequently, again using () and conditions (),
ask→ ∞. Hence,H(k,v) can be written in the form
) follows from the second condition in ().
This means, by the first condition in (), that ∞H(k,v) =∞.
4 Oscillation and nonoscillation criteria
We start with a statement based on the variational principle. This statement generalizes a result of [] dealing with the case˜rk= . Here, we do not need the sign restriction onrk,
Proof The proof is based on Lemma . LetN∈Nbe arbitrary and let K,L, M,N be
positive integers satisfyingN<K<L<M<N, whereK is such that () is disconjugate
on [K,∞) and(r˜k/rk)≤ holds on this interval, andLis such thathhas no generalized gN = . The fact that such a solution really exists follows from the disconjugacy of () on [K,∞) and from the homogeneity of the solution space of (). Indeed, ifxis a solution of
the corresponding solutions of Riccati equation (). Set
F(y;K,L– ) =
Next, using summation by parts, and sincehis a solution of (), we have
F(y;L,M– ) =
Concerning interval [M,N], sincegN= andhhas no generalized zero in this interval, it follows from [, Lemma ] thatw[kg]<wk[h]fork∈[M,N– ]. This means that
rk(gk)gk<rk
(hk)
(hk) |
At the same time, < rkgkgk+
Next, again summing by parts and using the boundary conditions and the fact thatgis a solution of (),
= n–
k=M
r˜k(hk)
+˜ck(hk+)
hk+.
Combining the above computations, we have
F(y;M,N– )≤–
+r˜M
rM
w[Mg]|hM|p– n–
k=M
˜rk(hk)
+c˜k(hk+)
hk+. ()
Consequently, using (), () and (), we obtain
F(y;N,∞) =
∞
k=N
(rk+r˜k)|yk|p– (ck+˜ck)|yk+|p
=F(y;K,L– ) +F(y;L,M– ) +F(y;M,N– )
≤α+α+
+˜rM
rM
|hM|p
w[Mh]–w[Mg]
– n–
k=L
˜rk(hk)
+c˜k(hk+)
hk+.
Letα> be arbitrary. It follows from condition () thatMcan be taken so large that
n–
k=L
r˜k(hk)
+c˜k(hk+)
hk+≥α+α+α.
Sincehis the recessive solution of (),i.e. w[h] is the minimal solution of (), from its
construction () we have
w[Mh]= lim
N→∞w
[g]
M.
Hence,Ncan be taken so large that
+r˜M
rM
|hM|p
w[Mh]–w[Mg]<α.
Consequently, ifM,Nare taken as above, then
F(y;N,∞)≤,
and this means that equation () is oscillatory by Lemma . The proof is complete.
The following results are based on the modified Riccati technique. Here we suppose that
his a positive solution of the nonoscillatory equation () andrk> ,rk+r˜k> for large k. Denote
Ck:=
˜rk(hk)
+c˜k(hk+)
hk+, Rk:=
q(rk+˜rk)hkhk+|hk|
First we present a statement, where, using the global inequalities (), (), equation () is compared with the linear equation
(Rkyk) +Ckyk+= . ()
This statement generalizes [, Theorems . and .].
Theorem Let h be a solution of () such that hk> , hkhk> for large k.
(i) Letp≥, and suppose thatCk≥for largekand ∞Rk =∞. If the linear
equation () is nonoscillatory, then equation () is also nonoscillatory.
(ii) Letp≤, and suppose thathis the recessive solution of (),˜rk≤,c˜k≥for largek.
If the linear equation () is oscillatory, then equation () is also oscillatory.
Proof (i) Nonoscillation of equation () means that there exists a solutionvof the Riccati equation
vk+Ck+ vk Rk+vk
=
such thatRk+vk> for largek. Assumptions of the theorem imply, in view of Lemma (ii), thatvk≥ for largek. Hence, from Lemma (ii) it follows
(Rk+vk)H(k,vk)≤vk for largek,
i.e. vksolves the inequality
vk+Ck+H(k,vk)≤
for large k. Consequently, by Lemma , we have that the sequence w=h–p(v+G) is a solution of Riccati inequalityR˜[wk]≤. Moreover,
rk+r˜k+wk=rk+˜rk+h–kpvk+ (rk+r˜k)(hk/hk) =h–kpvk+
+˜rk
rk
rk+w[kh]
,
wherew[kh] is a solution of () related to a nonoscillatory solutionhof () and hencerk+ w[kh]> for largek. Sincevk≥ for largek, we haverk+r˜k+w˜k> for largek. Hence, equation () is nonoscillatory according to Lemma .
(ii) Suppose, by contradiction, that equation () is nonoscillatory and letwbe a solution of the associated Riccati equation (). Then, by Lemma ,v=hpw–Gsolves the modified Riccati equation (). Sincer˜k≤,c˜k≥, it follows from Lemma thatwk≥w[kh]for large k, wherew[kh] =rk(hk/hk) is the minimal solution of (). This means thatvk=hpkwk –Gk=hpk(wk– ( +rk˜rk)w[kh])≥ for largek. Applying inequality () we have
(Rk+vk)H(k,vk)≥vk
for largekand hencevsolves the inequality
vk+Ck+ vk Rk+vk≤
which, together with the fact thatRk+vk> for largek, means that () is nonoscillatory.
The next two statements are based on the estimate () which holds for allp> . Hence we do not have to distinguish between the casesp> ,p< . The idea of the proofs is the same as in Theorem and we skip the proofs since they are similar to that in [, Theo-rems . and .]. The only difference is that we replacerkbyrk+˜rk.
Theorem Suppose that h is a solution of () such that hk> , hkhk> for large k and let
∞
Rk =∞,
∞
Ck<∞, Ck≥for large k ()
and
lim inf
k→∞ Gk> . ()
If there existsε> such that the linear equation
(Rkyk) + ( +ε)Ckyk+= ()
is nonoscillatory, then equation () is also nonoscillatory.
Theorem Suppose that h is the recessive solution of () such that hk> , hkhk> , ˜
ck≥,˜rk≤for large k, conditions () hold and
lim
k→∞Gk=∞. ()
If there existsε> such that the linear equation
(Rkyk) + ( –ε)Ckyk+= ()
is oscillatory, then equation () is also oscillatory.
The next statement is a version of Theorem . In the proof of the statement we use Lemma (iv). This enables to replace condition () by alternative conditions. This state-ment is new also in caser˜= .
Theorem Suppose that h is the recessive solution of () such that hk > , hkhk> , ˜
ck≥,˜rk≤for large k and conditions (), () and () hold. If there existsε> such that the linear equation () is oscillatory, then equation () is also oscillatory.
Proof Suppose, by contradiction, that equation () is nonoscillatory and letwbe a solu-tion of the associated Riccati equasolu-tion () andv=hpw–Gthe associated solution of the modified Riccati equation (). Similarly as in the proof of Theorem (ii), we conclude that
vk≥ for largek. SinceCk≥ for largekandH(k,vk) is nonnegative, we have for largek
This means that there exists a finite limitlimk→∞vk. Summing the modified Riccati equa-tion fromN tok,Nbeing sufficiently large, and using the fact thatvis nonnegative, we obtain
vN≥ k
j=N Cj+
k
j=N H(j,vj).
Lettingk→ ∞and since ∞Ck<∞, we have ∞H(k,vk) <∞. From Lemma (iv) it follows thatlimk→∞vk= . Now we can use the estimate (). There existsNsuch that
H(k,vk) >
–ε
v
k Rk+vk
fork>N.
We have
Rk=
q(rk+˜rk)hkhk+|hk| p–=
qGk
+ hk
hk
>
qGk.
Hence, by (),limk→∞ vk
Rk = and similarly as in [, Theorem .], we obtain for
suffi-ciently largekthe estimate Rk
–ε+vk < –ε
Rk+vk. Hence,
H(k,vk) >
–ε
v
k Rk+vk
> v
k Rk
–ε+vk
for sufficiently largek. This means thatvksolves the inequality
vk+Ck+ vk Rk
–ε+vk
< ,
which is the Riccati inequality associated with equation () and since –Rkε +vk> , we
have nonoscillation of (). This is a contradiction.
It is known (see []), that if ck≥,rk> , ∞r–k =∞, ∞
ck <∞, then the linear equation () is nonoscillatory provided
lim sup
k→∞ k–
r–j ∞
j=k cj<
, ()
and it is oscillatory provided
lim inf
k→∞ k–
rj– ∞
j=k cj>
. ()
Applying this criterion to equations (), (), we obtain the following result.
(i) If
lim sup
k→∞ k–
(rj+r˜j)hjhj+|hj|p–
∞
j=k Cj<
q, ()
then equation () is nonoscillatory.
(ii) Let moreoverhbe the recessive solution of (),˜ck≥,˜rk≤for largekand let either
condition () or conditions () and () be satisfied. If
lim inf
k→∞ k–
(rj+r˜j)hjhj+|hj|p–
∞
j=k Cj>
q,
then equation () is oscillatory.
Proof Consider,e.g., the case (i), the case (ii) is analogous. Condition () can be written in the form
lim sup
k→∞ k–
R–j ∞
j=k Cj<
.
Consequently, there existsε> such that
lim sup
k→∞ k–
R–j ∞
j=k
( +ε)Cj< ,
hence equation () is nonoscillatory and this implies, by Theorem , nonoscillation of
().
Remark If˜rk= ,c˜k≥, then the statements of Theorem reduce to [, Theorems . and .]. More precisely, (non)oscillation of
rk(xk)
+ (ck+˜ck)(xk+) = ()
is compared with that of the linear equation
Rkyk
+Ckyk+= , Rk:=
qrkhkhk+|hk| p–,C
k:=c˜khpk+. ()
In part (ii) of Theorem we suppose thatp≤,r˜k≤,c˜k≥. Under these conditions, if equation () is oscillatory, then () is oscillatory by [, Theorem .] and oscillation of () follows then by the Sturm comparison theorem, see,e.g., []. However, Theorem , part (ii) extends [, Theorem .] in case when the perturbation˜ckis ‘not too much positive’ so that equation () and hence also () is nonoscillatory.
Remark The conditions˜rk≤,˜ck≥ considered in Theorem (ii), Theorem and Theorem are used to show thatwk– ( +˜rk/rk)wk[h]≥, wherewandw[h]are the solu-tions of the Riccati type equasolu-tions associated with equasolu-tions () and (), respectively. We conjecture thatwkand (+˜rk/rk)w[kh]can be compared by another argument than Lemma , so the sign restriction on the perturbation termsr˜k,c˜kcan be relaxed, similarly as in the continuous case [].
5 Application
In this section, we apply the previous results to the perturbed Euler-type difference equa-tion
( +r˜k)(xk)
+
γp (k+ )p +c¯k
(xk+) = , γp:=
p–
p
p–
. ()
This equation is considered as a perturbation of the nonoscillatory equation
(xk)
+ck(xk+) = , ()
where
ck= –
(hk)
(hk+)
, hk=k p–
p . ()
It is easy to see thathk=k p–
p is a solution of () and it was shown in [] that ifp≥,
then it is the recessive solution. By a direct computation, as shown in [], the coefficient
ckis of the form
ck=
γp (k+ )p
+O(k+ )–, ask→ ∞
and also
hk= p–
p k
–p +Ok–,
hk(hk) =
p–
p
p–
+Ok–,
hkhk+(hk)p–=
p–
p
p–
k +Ok–,
(hk) = –
p–
p
p
(k+ )–+p +O(k+ )–,
ask→ ∞. Consequently,
hk hk
=p–
p k
– +Ok–, ask→ ∞,
hence conditions () are satisfied and we have
Gk=
p–
p
p–
( +˜rk)
+Ok–, Rk=
p–
p
p–
( +r˜k)k
and
ask→ ∞. Using these computations, Corollary applied to () reads as follows.
Corollary Let Ckbe given in () and suppose that
∞
then equation () is nonoscillatory.
(ii) Suppose moreover thatp≥,limk→∞c¯k(k+ )p+=∞,˜rk≤for largek. If
then equation () is oscillatory.
Competing interests
The author declares that she has no competing interests.
Acknowledgements
Research supported by the Grant P201/10/1032 of the Czech Science Foundation.
Received: 23 January 2012 Accepted: 11 June 2012 Published: 5 July 2012
References
1. ˇRehák, P: Oscillatory properties of second order half-linear difference equations. Czechoslov. Math. J.51, 303-321 (2001)
2. Agarwal, RP, Bohner, M, Grace, SR, O’Regan, D: Discrete Oscillation Theory. Hindawi Publishing Corporation, New York (2005)
3. Došlý, O, ˇRehák, P: Half-Linear Differential Equations. North-Holland Mathematics Studies, vol. 202. Elsevier, Amsterdam (2005)
4. Došlý, O, Fišnarová, S: Linearized Riccati technique and (non)oscillation criteria for half-linear difference equations. Adv. Differ. Equ.2008, Article ID 438130 (2008)
5. Došlý, O, ˇRehák, P: Recessive solution of half-linear second order difference equations. J. Differ. Equ. Appl.9, 49-61 (2003)
6. Došlý, O, Fišnarová, S: Half-linear oscillation criteria: Perturbation in term involving derivative. Nonlinear Anal., Theory Methods Appl.73, 3756-3766 (2010)
7. Došlý, O, Fišnarová, S: Variational technique and principal solution in half-linear oscillation criteria. Appl. Math. Comput.217, 5385-5391 (2011)
8. Došlý, O: Oscillation criteria for higher order Sturm-Liouville difference equations. J. Differ. Equ. Appl.4, 425-450 (1998)
9. Erbe, LH, Zhang, BG: Oscillation of second order linear difference equations. Chin. J. Math.16, 239-252 (1988) 10. Došlý, O, Fišnarová, S: Summation characterization of the recessive solution for half-linear second order difference
doi:10.1186/1687-1847-2012-101
Cite this article as:Fišnarová:Oscillatory properties of half-linear difference equations: two-term perturbations.