Citation: Yun Xin, Hao Wang. Positive periodic solution for third-order singular neutral differential equation with time-dependent delay[J]. AIMS Mathematics, 2020, 5(6): 7234-7251. doi: 10.3934/math.2020462
[1] | Ahmed Morsy, Kottakkaran Sooppy Nisar, Chokkalingam Ravichandran, Chandran Anusha . Sequential fractional order Neutral functional Integro differential equations on time scales with Caputo fractional operator over Banach spaces. AIMS Mathematics, 2023, 8(3): 5934-5949. doi: 10.3934/math.2023299 |
[2] | Xiulin Hu, Lei Wang . Positive solutions to integral boundary value problems for singular delay fractional differential equations. AIMS Mathematics, 2023, 8(11): 25550-25563. doi: 10.3934/math.20231304 |
[3] | Zhibo Cheng, Lisha Lv, Jie Liu . Positive periodic solution of first-order neutral differential equation with infinite distributed delay and applications. AIMS Mathematics, 2020, 5(6): 7372-7386. doi: 10.3934/math.2020472 |
[4] | Veliappan Vijayaraj, Chokkalingam Ravichandran, Thongchai Botmart, Kottakkaran Sooppy Nisar, Kasthurisamy Jothimani . Existence and data dependence results for neutral fractional order integro-differential equations. AIMS Mathematics, 2023, 8(1): 1055-1071. doi: 10.3934/math.2023052 |
[5] | Shabir Ahmad, Aman Ullah, Ali Akgül, Manuel De la Sen . A study of fractional order Ambartsumian equation involving exponential decay kernel. AIMS Mathematics, 2021, 6(9): 9981-9997. doi: 10.3934/math.2021580 |
[6] | Axiu Shu, Xiaoliang Li, Bo Du, Tao Wang . Positive periodic stability for a neutral-type host-macroparasite equation. AIMS Mathematics, 2025, 10(3): 7449-7462. doi: 10.3934/math.2025342 |
[7] | Kanagaraj Muthuselvan, Baskar Sundaravadivoo, Kottakkaran Sooppy Nisar, Suliman Alsaeed . Discussion on iterative process of nonlocal controllability exploration for Hilfer neutral impulsive fractional integro-differential equation. AIMS Mathematics, 2023, 8(7): 16846-16863. doi: 10.3934/math.2023861 |
[8] | Yan Yan . Multiplicity of positive periodic solutions for a discrete impulsive blood cell production model. AIMS Mathematics, 2023, 8(11): 26515-26531. doi: 10.3934/math.20231354 |
[9] | Zaitao Liang, Xuemeng Shan, Hui Wei . Multiplicity of positive periodic solutions of Rayleigh equations with singularities. AIMS Mathematics, 2021, 6(6): 6422-6438. doi: 10.3934/math.2021377 |
[10] | Velusamy Kavitha, Dumitru Baleanu, Jeyakumar Grayna . Measure pseudo almost automorphic solution to second order fractional impulsive neutral differential equation. AIMS Mathematics, 2021, 6(8): 8352-8366. doi: 10.3934/math.2021484 |
The main purpose of this paper is to consider the existence of a positive periodic solution for third-order neutral differential equation with a singularity
(u(t)−cu(t−τ(t)))‴+a(t)u(t)=f(t,u(t))+e(t), | (1.1) |
where τ∈C(R,R) is a ω-periodic function, a∈C(R,R+) is ω-periodic functions, c is a constant and c∈(−mM+m,mM+m), here M:=maxt∈[0,ω]|a(t)| and m:=mint∈[0,ω]|a(t)|, e∈L1(R) is an ω-periodic function, the nonlinear term f∈Car(R×R+,R) is a L2-Carathˊeodory function and is an ω-periodic function on t, f has a singularity of repulsive type at the origin, i.e.,
limu→0+f(t,u)=+∞,uniformly in t. |
Note that when c=0 the neutral operator u↦(u(t)−cu(t−τ(t))‴ reduces to the linear operator u↦u‴ and then equation (1.1) is of the differential equation form
u‴+a(t)u=f(t,u)+e(t). | (1.2) |
During the last two decades, there are some good amount of works on periodic solutions for neutral differential equations (see [2,3,6,7,8,11,14,16,17,19,22,23] and the references cited therein). Some classical tools have been used to study neutral differential equation in the literature, including the fixed point index theorem [2,16], Krasnoselskii's fixed point theorem [3,6,7,8], fixed point theorem of Leray-Schauder type [14], Mawhin's continuous theorem [11,23], Continuation theorem of coincidence degree theory [19], the fixed point theorem in cones [17,22]. For example, Wu and Wang [22] in 2007 discussed a kind of second-order neutral differential equation
(u(t)−cu(t−τ))″+a(t)u(t)=λb(t)f(u(t−δ(t))), | (1.3) |
where c∈(−mM+m,0), b,δ∈C(R,R) are ω-periodic functions, λ is a constant and 0<λ<1. By a fixed point theorem in cones and the property of neutral operator (A1u)(t):=u(t)−cu(t−τ), they obtained sufficient conditions for the existence of positive periodic solutions to (1.3). Afterwards, Ren et al. [17] in 2011 considered second-order neutral differential equation with variable delay as follows
(u(t)−cu(t−τ(t)))‴+a(t)u(t)=f(t,u(t−τ(t)), | (1.4) |
where |c|<1. The authors presented the existence result for a positive periodic solution for (1.4) by applications of Krasnoselskii's fixed point theorem.
Besides, recently there have been published some results on second-order or third-order singular equations (see [4,5,9,10,12,13,15,18,20,21,24,25]). Torres [18] in 2007 investigated the existence of periodic solutions for the following second-order equation with a singularity of repulsive type
u″+a(t)u=f(t,u)+e(t). | (1.5) |
His proof was based on Schauder's fixed point theorem. After that, Ma et al. [15] in 2014 improved Torres's result and given an assumption which is relatively weaker than condition in [18].
We are mainly motivated by the recent work [15,16,18,22] and focus on Eq (1.1). By employing two available operators and applying Krasnoselskii's fixed point theorem, we obtain the existence of a positive periodic solution for (1.1). We would like to emphasize that the inclusion of the neutral operator in the singularity implies a new technical difficulty concerning the right choice of the operator.
Firstly, we recall Krasnoselskii's fixed point theorem, which can be found in [1].
Lemma 2.1. Let Y be a Banach space. Assume K is a bounded closed convex subset of Y. If Q, S:K→Y satisfy
(ⅰ) Qu+Sy∈K,∀u,y∈K;
(ⅱ) S is a contractive operator;
(ⅲ) Q is a completely continuous operator in K.
Then Q+S has a fixed point in K.
Consider the following third-order linear nonhomogeneous differential equation
{u‴+a(t)u=h(t),u(0)=u(ω),u′(0)=u′(ω),u″(0)=u″(ω), | (2.1) |
where h∈C+ω:={h∈C(R,(0+∞)):h(t+ω)≡h(t),∀t∈R}. Obviously, the calculation of the Green's function of (2.1) is very complicated. In order to get around the calculation of the Green's function of (2.1), we discuss the Green's function of differential equation as follows
{u‴+Mu=h(t),u(0)=u(ω),u′(0)=u′(ω),u″(0)=u″(ω), | (2.2) |
where M=maxt∈[0,ω]|a(t)| is defined in Section 1. The Eq (2.2) has a unique ω-periodic solution
u(t)=∫ω0G(t,s)h(s)ds. | (2.3) |
We introduce the positiveness of the Green's function G(t,s), which can be found in [8,Lemma 2.2].
Lemma 2.2. (see [8,Lemma 2.2]) Assume that M<64π381√3ω3 holds, then the Green's function G(t,s) satisfies
0<l:=13M23(exp(M13ω)−1)≤G(t,s)≤3+2exp(−M13ω2)3M23(1−exp(−M13ω2))2:=L, |
for all (t,s)∈[0,ω]×[0,ω]. Furthermore, ∫ω0G(t,s)ds=1M.
On the other hand, we give the property of neutral operator (Au)(t):=u(t)−cu(t−τ(t)).
Lemma 2.3. (see [16,Lemma 2.1]) If |c|<1, then the operator Au has a continuous inverse A−1u on
X:={u∈C(R,R):u(t+ω)≡u(t),∀t∈R}, |
satisfying
(1) (A−1f)(t)=f(t)+∞∑j=1cjf(s−j−1∑i=1τ(Di)),∀f∈X;
(2) |(A−1f)(t)|≤‖f‖1−|c|,∀f∈X,
where t−τ(t)=s and Dj=s−j−1∑i=1τ(Di), ‖f‖:=maxt∈[0,ω]|f(t)|.
Let v(t)=(Au)(t), then from Lemma 2.3, we obtain that u(t)=(A−1v)(t). Hence (1.1) can be transformed into
v‴(t)+a(t)(A−1v)(t)=f(t,A−1v(t))+e(t), | (2.4) |
which can be further rewritten as
v‴(t)+a(t)v(t)−a(t)H(v(t))=f(t,A−1v(t))+e(t), | (2.5) |
where H(v(t))=v(t)−(A−1v)(t)=−c(A−1v)(t−τ(t)).
Furthermore, we consider
v‴(t)+a(t)v(t)−a(t)H(v(t))=h(t),forh∈C+ω. | (2.6) |
Formula (2.6) is rewritten in the following form
v‴+Mv=(M−a(t))v+a(t)H(v(t))+h(t). | (2.7) |
Define operators T,B:X→X by
(Th)(t)=∫ω0G(t,s)h(s)ds,(Bv)(t)=(M−a(t))v+a(t)H(v(t)). |
Obviously, (Th)(t)>0 if M<64π381√3ω3, for all t∈[0,ω] and h∈C+ω. By Lemma 2.3, ‖B‖≤M−m+M|c|1−|c| if |c|<1. By (2.3), the solution of (2.7) can be written as the following form
v(t)=(Th)(t)+(TBv)(t). |
In view of c∈(−mM+m,mM+m), we arrive at
‖TB‖≤‖T‖‖B‖≤M−m+m|c|M(1−|c|)<1, | (2.8) |
where we used the fact ∫ω0G(t,s)ds=1M. Hence
v(t)=(I−TB)−1(Th)(t). | (2.9) |
Remark 2.1. If |c|>1, by Lemma [16,Lemma 2.1] and (2.8)−(2.9), we get ‖TB‖≤1−mM+|c||c|−1. Since 1−mM+|c||c|−1>1, we can not get (I−TB)−1. Therefore, the above method does not apply to the case of |c|>1.
Define an operator P:X→X by
(Ph)(t)=(I−TB)−1(Th)(t). |
If M<64π381√3ω3, then (2.6) has the unique periodic solution v(t)=(Ph)(t). Moveover, we get the following conclusion.
Lemma 2.4. Assume that M<64π381√3ω3, c<0 and |c|<min{mM+m,σ} hold. Then P satisfies
(Th)(t)≤(Ph)(t)≤M(1−|c|)m−(M+m)|c|‖Th‖,forh∈C+ω, | (2.10) |
where σ:=lL and 0<σ≤1.
Proof. By the Neumann expansion of P, we have
P=(I−TB)−1T=(I+TB+(TB)2+⋯+(TB)n+⋯)T=T+TBT+(TB)2T+⋯+(TB)nT+⋯. | (2.11) |
From (2.11) and recalling that ‖TH‖≤M−m+m‖c‖M(1−‖c‖)<1, we get
(Ph)(t)=(I−TB)−1(Th)(t)≤‖Th‖I−‖TB‖≤M(1−|c|)m−(M+m)|c|‖Th‖. |
On the other hand, applying Lemma 2.2 and M<64π381√3ω3, we obtain
(Th)(t)=∫ω0G(t,s)h(s)ds≥l∫ω0h(s)ds=lLL∫ω0h(s)ds≥σmaxt∈[0,ω]∫ω0G(t,s)h(s)ds=σ‖Th‖>0. |
Since c∈(−mM+m,0) and |c|≤σ, using Lemma 2.3, we arrive at
(A−1Th)(t)=(Th)(t)+∞∑j=1cj(Th)(s−j−1∑i=1τ(Di))=(Th)(t)+∑j≥1evencj(Th)(s−j−1∑i=1τ(Di))−∑j≥1odd|c|j(Th)(s−j−1∑i=1τ(Di))≥σ‖Th‖1−c2−|c|‖Th‖1−c2=(σ−|c|)‖Th‖1−c2≥0, |
and from (2.11), we see that
(BTh)(t)=(M−a(t))(Th)(t)+a(t)(−c(A−1Th)(t−τ(t)))≥0,forh∈C+ω. |
Clearly, (TBTh)(t)≥0 if h∈C+ω. Then we have from the above analysis that
(Ph)(t)=(Th)(t)+(TBTh)(t)+((TB)2Th)(t)+((TB)3Th)(t)+⋯≥(Th)(t),forh∈C+ω. |
Lemma 2.5. Assume that M<64π381√3ω3 and c∈(0,mM+m) hold. Then P satisfies
m−(M+m)cM(1−c)(Th)(t)≤(Ph)(t)≤M(1−c)m−(M+m)c‖Th‖,forh∈C+ω. | (2.12) |
Proof. In view of ‖TB‖<1, similarly as the proof of Lemma 2.4, we get that (Ph)(t)≤M(1−c)m−(M+m)c‖Th‖.
Since c∈(0,mM+m), we can not get (TBh)(t)≥0 for all h∈C+ω. From (2.11), it is clear
P=(I+TB+(TB)2+(TB)3+⋯)T=(I+(TB)2+(TB)4+⋯)T+(TB+(TB)3+(TB)5+⋯)T=(I+(TB)2+(TB)4+⋯)T+(I+(TB)2+(TB)4+⋯)TBT=(I+(TB)2+(TB)4+⋯)(I+TB)T. |
Then, we get
(Ph)(t)≥(I+TB)(Th)(t)≥(I−‖TB‖)(Th)(t)≥m−(m+M)cM(1−c)(Th)(t)>0,forh∈C+ω. |
Define operators Q,S:X→X by
(Qu)(t)=P(f(t,u)+e(t)),(Su)(t)=cu(t−τ(t)). | (2.13) |
From (2.6) and (2.13), the existence of periodic solutions to (1.1) is equivalent to the existence of solutions to operator equation as follows
Qu+Su=u,inX. | (2.14) |
In this section, we establish the existence of a positive periodic solution for (1.1) in the case that c∈(0,mM+m] by using Krasnoselskii's fixed point theorem. Define the function γ:R→R
γ(t):=∫ω0G(t,s)e(s)ds, |
and
γ∗:=maxt∈Rγ(t),γ∗:=mint∈Rγ(t). |
By analysis of γ(t), we consider the following three cases.
Case (Ⅰ) γ∗>0.
Theorem 3.1. Suppose M<64π381√3ω3 and c∈(0,mm+M] hold. Furthermore, assume that the following conditions hold:
(H1) There exist continuous, non-negative functions g(u),h(u) and ζ(t) such that
0≤f(t,u)≤ζ(t)(g(u)+h(u))for all(t,u)∈[0,ω]×(0,∞), |
and g(u)>0 is non-increasing and h(u) is non-decreasing in u∈(0,∞).
(H2) There exists a positive constant R>0 such that
Mm−(M+m)c(g(m−(M+m)cM(1−c)γ∗)(1+h(R)g(R))Λ∗+γ∗)≤R, |
where Λ(t)=∫ω0G(t,s)ζ(s)ds and Λ∗:=maxt∈RΛ(t).
If γ∗>0, then (1.1) has at least one positive periodic solution.
Proof. An ω-periodic solution of (1.1) is just a fixed point of the following operator equation
(Qu)(t)+(Su)(t)=u(t). | (3.1) |
Let R be the positive constant satisfying (H2) and
r:=m−(M+m)cM(1−c)γ∗. |
Then we have R>r>0 since R>γ∗>m−(M+m)cM(1−c)γ∗. Now we define the set
K={u∈X:r≤u(t)≤Rfor allt}. | (3.2) |
Obviously, K is a bounded closed convex set in X. Moreover, for any u∈K, it is easy to verify that Q, S are continuous and (Qu)(t+ω)=(Qu)(t),(Su)(t+ω)=(Su)(t), that is, Q(K)⊂X,S(K)⊂X.
Next we claim that Qu+Sy∈K, for any u,y∈K. By Lemma 2.5 and non-negative sign of G(t,s) and f(t,u), we have
(Qu)(t)+(Sy)(t)=P(f(t,u(t))+e(t)))+cy(t−τ(t))≥m−(M+m)cM(1−c)T(f(t,u(t))+e(t)))+cy(t−τ(t))=m−(M+m)cM(1−c)(∫ω0G(t,s)(f(s,u(s))ds+γ(t))+cy(t−τ(t))≥m−(M+m)cM(1−c)γ∗:=r>0, | (3.3) |
where we used the fact ∫ω0G(t,s)ds=1M.
On the other hand, using Lemma 2.5, we have
(Qu)(t)+(Sy)(t)=P(f(t,u(t))+e(t)))+cy(t−τ(t))≤M(1−c)m−(M+m)cmaxt∈[0,ω]|T(f(t,u(t))+e(t)))|+cy(t−τ(t))=M(1−c)m−(M+m)cmaxt∈[0,ω]|∫ω0G(t,s)f(s,u(s))ds+γ(t)|+cy(t−τ(t))≤M(1−c)m−(M+m)c{maxt∈[0,ω]|∫ω0G(t,s)f(s,u(s))ds|+‖γ‖}+cy(t−τ(t)), |
since γ∗>0, then γ(t)>0 and ‖γ‖:=maxt∈[0,ω]|γ(t)|=γ∗. By conditions (H1) and (H2), we have
(Qu)(t)+(Sy)(t)≤M(1−c)m−(M+m)c{maxt∈[0,ω]|∫ω0G(t,s)ζ(s)(g(u(s))+h(u(s)))ds|+γ∗}+cy(t−τ(t))≤M(1−c)m−(M+m)c{g(r)(1+h(R)g(R))Λ∗+γ∗}+cR≤R. | (3.4) |
Combining (3.3) and (3.4), we get Qu+Sy∈K for all u,y∈K.
Furthermore, for any u1,u2∈K, we have
‖(Su1)(t)−(Su2)(t)‖=|cu1(t−τ(t))−cu2(t−τ(t))|≤|c|‖u1−u2‖, |
which implies that
‖(Su1)(t)−(Su2)(t)‖≤|c|‖u1−u2‖. | (3.5) |
In view of |c|<1, S is a contractive operator.
By [16,Theorem 3.1], we get that Q is a completely continuous. Therefore, by Krasnoselskii's fixed point theorem, Q+S has a fixed point u∈K, that is to say, (1.1) has a positive ω-periodic solution u(t) with u∈[r,R].
Remark 3.1. If |c|>1, from (3.11), we can not get S is a contractive operator. Therefore, the above method of Theorem 3.1 does not apply to the case of |c|>1.
Corollary 3.1. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. Assume the following condition holds:
(F1) There exist a continuous function d(t)≻0 and a constant ρ>0 such that satisfy
0≤f(t,u)≤d(t)uρ,for all(t,u)∈[0,ω]×(0,∞). |
If γ∗>0, then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 3.1. We take
ζ(t)=d(t),g(u)=1uρ,h(u)=0. |
Then condition (H1) is satisfied. Next, we consider the condition (H2) is also satisfied. In fact, we take R>0 with
Mm−(M+m)c(Mρ(1−c)ρΨ∗(m−(M+m)c)ργρ∗+γ∗)≤R, |
since c∈(0,mM+m) and Ψ(t):=∫ω0G(t,s)d(s)dt.
Corollary 3.2. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. Assume the following condition holds:
(F2) There exist a continuous function d(t)≻0 and constants ρ>0,0≤η<1 such that satisfy
0≤f(t,u)≤d(t)uρ+d(t)uη,for all(t,u)∈[0,ω]×(0,∞). |
If γ∗>0, then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 3.1. We take
ζ(t)=d(t),g(u)=1uρ,h(u)=uη. |
Then condition (H1) is satisfied and the existence condition (H2) is also satisfied. The existence condition (H2) becomes
Mm−(M+m)c(Ψ∗(Mρ(1−c)ρ(m−(M+m)c)ργρ∗+(R)η)+γ∗)≤R, | (3.6) |
Since ρ>0,0≤η<1 and c∈(0,mM+m), we can choose R>0 large enough such that (3.6) is satisfied.
In the following, we investigate (1.1) in the case that attractive-repulsive singularities.
Corollary 3.3. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. Assume that the following condition holds:
(F3) There exist positive constants α>β>0 and μ>0 such that
f(t,u)=1uα−μuβ. |
If γ∗>0, then there exists a positive constant μ1 such that (1.1) has at least one positive periodic solution for each 0≤μ≤μ1.
Proof. We apply Theorem 3.1. Take
g(x)=1uα,h(u)≡0,ζ(t)≡1. |
Firstly, we consider that condition (H2) is satisfied. Take R>0 with
R=Mm−(M+m)c(1M(γ∗)α+γ∗), |
where ∫ω0G(t,s)ds=1M. Next, we consider that the condition (H1) is also satisfied. In fact, f(t,u)≥0 if and only if μ≤uβ−α.
In view of β<α, then we have μ<Rβ−α. As a consequence, the result holds for
μ1:=(Mm−(M+m)c(1M(γ∗)α+γ∗))β−α. |
By Theorem 3.1, we consider a special case of c, i.e., c=0.
Theorem 3.2. Suppose M<64π381√3ω3 and c=0 hold, and f(t,u) satisfies condition (H1). Furthermore, assume that the following conditions holds:
(H∗2) There exists a positive constant R>0 such that
Mm(g(γ∗)(1+h(R)g(R))Λ∗+γ∗)≤R. |
If γ∗>0, then (1.2) has at least one positive periodic solution.
Remark 3.2. Theorem 3.2 extends and improves [26,Theorem 3.3].
Case (Ⅱ) γ∗=0.
Theorem 3.3. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. And f(t,u) satisfies (H1). Furthermore, assume that the following conditions hold:
(H3) For each L>0, there exists a continuous function ϕL≻0 such that f(t,u)≥ϕL(t) for all (t,u)∈[0,ω]×(0,L].
(H4) There exists R>0 such that R>(ΦR)′∗:=m−(M+m)cM(1−c)(ΦR)∗ and
Mm−c(m+M)(g((ΦR)′∗)(1+h(R)g(R))Λ∗+γ∗)≤R, |
where ΦR(t)=∫ω0G(t,s)(ϕR)(s)ds.
If γ∗=0, then (1.1) has at least one positive periodic solution.
Proof. We follow the same strategy and notation as in the proof of Theorem 3.1. Let R be the positive constant satisfying (H4) and let r:=m−(M+m)cM(1−c)(ΦR)∗; then R>r>0 since R>(ΦR)′∗. Next we prove that Qu+Sy∈K for all u,y∈K.
For each u,y∈K and for all t∈[0,ω], by Lemma 2.5 and (H3), we get
(Qu)(t)+(Sy)(t)=P(f(t,u(t))+e(t)))+cy(t−τ(t))≥m−(M+m)cM(1−c)T(f(t,u(t))+e(t)))+cy(t−τ(t))=m−(M+m)cM(1−c)(∫ω0G(t,s)f(s,u(s))ds+γ(t))+cy(t−τ(t))≥m−(M+m)cM(1−c)(∫ω0G(t,s)ϕR(s)ds+γ(t))+cy(t−τ(t))≥m−(M+m)cM(1−c)(ΦR)∗:=r>0. | (3.7) |
On the other hand, using Lemma 2.5, we see that
(Qu)(t)+(Sy)(t)=P(f(t,u(t))+e(t)))+cy(t−τ(t))≤M(1−c)m−(M+m)cmaxt∈[0,ω]|T(f(t,u(t))+e(t)))|+cy(t−τ(t))=M(1−c)m−(M+m)cmaxt∈[0,ω]|∫ω0G(t,s)f(s,u(s))ds+γ(t)|+cy(t−τ(t))≤M(1−c)m−(M+m)cmaxt∈[0,ω]|∫ω0G(t,s)f(s,u(s))ds+γ(t)|+cy(t−τ(t))≤M(1−c)m−(M+m)c{maxt∈[0,ω]|∫ω0G(t,s)f(s,u(s))ds|+‖γ‖}+cy(t−τ(t)). | (3.8) |
By conditions (H1) and (H4), we have
(Qu)(t)+(Sy)(t)≤M(1−c)m−(M+m)c{maxt∈[0,ω]|∫ω0G(t,s)ζ(s)(g(u(s))+h(u(s)))ds|+‖γ‖}+cy(t−τ(t))≤M(1−c)m−(M+m)c{g(r)(1+h(R)g(R))Λ∗+‖γ‖}+cR≤R. | (3.9) |
Combining (3.7) and (3.9), we get Qu+Sy∈K for all u,y∈K.
Similarly, we get that Q is a completely continuous and S is a contractive operator in X. Therefore, by Krasnoselskii's fixed point theorem, our result is proven.
Corollary 3.4. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. Assume that the following condition holds:
(F4) There exist continuous functions d(t),ˆd(t)≻0 and 0<ρ<1 such that satisfy
0≤ˆd(t)uρ≤f(t,u)≤d(t)uρ,for all(t,u)∈[0,ω]×(0,∞). |
If γ∗=0, then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 3.3. We take
ϕL(t)=ˆd(t)Lρ,ζ(t)=d(t),g(u)=1uρ,h(u)=0. |
Then conditions (H1) and (H3) are satisfied, and the existence condition (H4) becomes
R>(m−(M+m)c)ˆΨ∗M(1−c)Rρ=r,Mm−(M+m)c((M(1−c)Rρ(m−(M+m)c)ˆΨ∗)ρΨ∗+‖γ‖)≤R, | (3.10) |
where ˆΨ=∫ω0G(t,s)ˆd(t)dt. Note that Ψ∗>0, since 0<ρ<1, we can choose appropriate R>0 so that (3.10) is satisfied and the proof is complete.
Case (Ⅲ) γ∗<0.
Theorem 3.4. Suppose M<64π381√3ω3 and c∈(0,mM+m] hold. And f(t,u) satisfies (H1) and (H3). Furthermore, assume that the following condition holds:
(H5) There exists R>0 such that R>m−(M+m)cM(1−c)((ΦR)∗+γ∗)>0 and
Mm−(M+m)cg(m−(M+m)cM(1−c)((ΦR)∗+γ∗))(1+h(R)g(R))Λ∗≤R. |
If γ∗<0, then (1.1) has at least one positive periodic solution.
Proof. We follow the same strategy and notation as in the proof of Theorem 3.1. Let R be the positive constant satisfying (H5) and let r:=m−(M+m)cM(1−c)((ΦR)∗+γ∗); then R>r>0 since R>m−(M+m)cM(1−c)((ΦR)∗+γ∗). Next we prove that Qu+Sy∈K for all u,y∈K.
For each u,y∈K and for all t∈[0,ω], by Lemma 2.5 and (H3), we deduce
(Qu)(t)+(Sy)(t)=P(f(t,u(t))+e(t)))+cy(t−τ(t))≥m−(M+m)cM(1−c)T(f(t,u(t))+e(t)))+cy(t−τ(t))=m−(M+m)cM(1−c)(∫ω0G(t,s)f(s,u(s))ds+γ(t))+cy(t−τ(t))≥m−(M+m)cM(1−c)(∫ω0G(t,s)ϕR(s)ds+γ(t))+cy(t−τ(t))≥m−(M+m)cM(1−c)((ΦR)∗+γ∗):=r>0. | (3.11) |
On the other hand, applying Lemma 2.5 and (3.8), we arrive at
\begin{align*} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ \leq&\frac{M(1-c)}{m-(M+m)c}\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds+\gamma(t)\right| +cy(t-\tau(t))\\ \leq&\frac{M(1-c)}{m-(M+m)c}\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds\right|+cy(t-\tau(t)), \end{align*} |
since \gamma^*\leq 0, G(t, s) and f(t, u(t) are non-negative, (\Phi_R)_*+\gamma_* > 0 , then we know \left|\int^\omega_0G(t, s)f(s, u(s))ds+\gamma(t)\right|\leq\left|\int^\omega_0G(t, s)f(s, u(s))ds\right| . By conditions (H_1) and (H_5) , we get
\begin{equation} \begin{split} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ \leq&\frac{M(1-c)}{m-(M+m)c}\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)\zeta(s)(g(u(s))+h(u(s)))ds\right|+cy(t-\tau(t))\\ \leq&\frac{M(1-c)}{m-(M+m)c}g(r)\left(1+\frac{h(R)}{g(R)}\right)\Lambda^*+cR\\ \leq&R. \end{split} \end{equation} | (3.12) |
Combining (3.11) and (3.12), we get \mathcal{Q}u+\mathcal{S}y\in \mathcal{K} for all u, \; y\in \mathcal{K} .
Similarly, we get that \mathcal{Q} is a completely continuous and \mathcal{S} is a contractive operator in X . Therefore, by Krasnoselskii's fixed point theorem, our result is proven.
Corollary 3.5. Suppose M < \frac{64\pi^3}{81\sqrt{3}\omega^3} , c\in\left(0, \frac{m}{M+m}\right] and (F_4) hold. If \gamma^* < 0 and \gamma_*\geq\left(\hat{\Psi}_*\frac{(m-(M+m)c)^\rho}{(M\Psi^*)^\rho}\rho^2\right)^{\frac{1}{1-\rho^2}} \left(1-\frac{1}{\rho^2}\right) , then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 3.4. We take
\phi_L(t) = \frac{\hat{d}(t)}{L^\rho},\; \; \; \; \zeta(t) = d(t),\; \; \; \; g(u) = \frac{1}{u^\rho},\; \; \; \; h(u) = 0. |
Then conditions (H_1) and (H_3) hold. Next, we consider that the condition (H_5) is also satisfied. In fact, take R > 0 with
R: = \frac{M}{m-(M+m)c}\left(\frac{\Psi^*}{r^\rho}\right), |
then \frac{m-(M+m)c}{M(1-c)}\left((\Phi_R)_*+\gamma_*\right)\geq r holds if r verifies
\begin{equation*} \frac{m-(M+m)c}{M(1-c)}\left(\hat{\Psi}_*\frac{(m-(M+m)c)^\rho}{(M\Psi^*)^\rho}(r)^{\rho^2}+\gamma_{*}\right)\geq r, \end{equation*} |
or equivalently,
\begin{equation*} \gamma_{*}\geq f(r): = \frac{m-(M+m)c}{M(1-c)}r-\hat{\Psi}_*\frac{(m-(M+m)c)^\rho}{(M\Psi^*)^\rho}(r)^{\rho^2}. \end{equation*} |
The function f(r) possesses a minimum at r_0: = \left(\hat{\Psi}_*\frac{(m-(M+m)c)^\rho}{(M\Psi^*)^\rho}\rho^2\right)^{\frac{1}{1-\rho^2}}. Let r = r_0. Then the (\Phi_R)_*+\gamma_* > 0 holds if \gamma_{*}\geq f(r_0) , which is just the condition \gamma_*\geq\left(\hat{\Psi}_*\frac{(m-(M+m)c)^{\rho+1}}{(M\Psi^*)^\rho M(1-c)}\rho^2\right)^{\frac{1}{1-\rho^2}} \left(1-\frac{1}{\rho^2}\right) . The the condition (H_5) holds directly by the choice of R , and it would remain to prove that R = \frac{M}{m-(M+m)c}\left(\frac{\Psi^*}{(r_0)^\rho}\right) > r_0. This is easily verified through elementary computations.
In the end of this section, we illustrate our results with one example.
Example 3.1. Consider the following a singular equation
\begin{equation} \left(u-\frac{1}{8}u(t-\cos^2t)\right)'''+\frac{1}{8}(\sin(2t)+2)u = \frac{\cos(2t)+4}{u^\rho}+(\cos(2t)+4)u^{\frac{1}{2}}+e^{\sin^2t}, \end{equation} | (3.13) |
where \rho is a real constant and \rho > 0 .
Comparing Eq (3.13) with Eq (1.1), it is easy to see that
\begin{align*} c = \frac{1}{8},\; \tau = \cos^2t,\; a(t) = \frac{1}{8}(\sin(2t)+2),\; \omega = \pi,\; f(t,u) = \frac{\cos(2t)+4}{u^\rho}+(\cos(2t)+4)u^{\frac{1}{2}},\; e(t) = e^{\sin^2t}. \end{align*} |
Furthermore, we have
\begin{align*} m = \frac{1}{8},\; M = \frac{3}{8} \lt 0.456 \lt \frac{64}{81\sqrt{3}},\; c\in\left(0,\frac{1}{4}\right),\; d(t) = \cos(2t)+4,\; \eta = \frac{1}{2}, \end{align*} |
and
\gamma(t) = \int^\omega_0G(t,s)e(t)ds \gt 0. |
Obviously, condition (F_2) holds, and \gamma_* > 0 . Therefore, applying Corollary 3.2, we get that Eq (3.13) has at least one positive \pi -periodic solution.
In this section, we investigate the existence of positive periodic solutions for (1.1) in the case that c\in\left(-\frac{m}{M+m}, 0\right) by using Krasnoselskii's fixed point theorem.
Case (Ⅰ) \gamma_* > 0 .
Theorem 4.1. Suppose M < \frac{64\pi^3}{81\sqrt{3}\omega^3} , c < 0 and |c| < \min\left\{\frac{m}{M+m}, \sigma\right\} hold, and f(t, u) satisfies conditions (H_1) . Furthermore, assume that the following condition holds:
(H_2^{**}) There exists a positive constant R such that \frac{\gamma_*}{1+|c|} < R < \frac{\gamma_*}{|c|} and
\frac{M(1-|c|)}{m-(M+m)|c|}\left(g\left(\gamma_*-|c|R\right)\left(1+\frac{h(R)}{g(R)}\right)\Lambda^*+\gamma^*\right)\leq R. |
If \gamma_* > 0 , then (1.1) has at least one positive periodic solution.
Proof. We follow the same strategy and notation as in the proof of Theorem 3.1. Let R be the positive constant satisfying (H_2^{**}) and let r: = \gamma_*-|c|R ; then R > r > 0 since R > \frac{\gamma_*}{1+|c|} . Next we prove that \mathcal{Q}u+\mathcal{S}y\in \mathcal{K} for all u, \; y\in \mathcal{K} .
For any u, \; y\in \mathcal{K} , by Lemma 2.4 and non-negative sign of G(t, s) and f(t, u) , we see that
\begin{equation} \begin{split} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ = &\mathcal{P}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ \geq& \mathcal{T}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ = &\int^\omega_0G(t,s)f(s,u(s))ds+\gamma(t)-|c|y(t-\tau(t))\\ \geq&\gamma_*-|c|R: = r \gt 0. \end{split} \end{equation} | (4.1) |
On the other hand, applying Lemma 2.4, we obtain
\begin{align*} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ = &\mathcal{P}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\max\limits_{t\in[0,\omega]}|\mathcal{T}(f(t,u(t))+e(t))|+cy(t-\tau(t))\\ = &\frac{M(1-|c|)}{m-(M+m)|c|}\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds+\gamma(t)\right|-|c|y(t-\tau(t))\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left(\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds\right|+\gamma(t)\right), \end{align*} |
since c < 0 and y\in \mathcal{K} . By conditions (H_1) and (H_2^{**}) , we get
\begin{equation} \begin{split} &(Qu)(t)+(Sy)(t)\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left\{\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)\zeta(s)(g(u(s))+h(u(s)))ds\right| +\gamma^*\right\}\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left\{g(r)\left(1+\frac{h(R)}{g(R)}\right)\Lambda^*+\gamma^*\right\}\\ \leq&R. \end{split} \end{equation} | (4.2) |
Combining (4.1) and (4.2), we get \mathcal{Q}u+\mathcal{S}y\in \mathcal{K} for all u, \; y\in \mathcal{K} .
Similarly, we get that \mathcal{Q} is a completely continuous and \mathcal{S} is a contractive operator in \mathcal{X} . Therefore, by Krasnoselskii's fixed point theorem, our results is proven.
Corollary 4.1. Suppose M < \frac{64\pi^3}{81\sqrt{3}\omega^3} , c < 0 , |c| < \min\left\{\frac{m}{M+m}, \sigma\right\} and (F_1) hold. If
\begin{equation} \gamma_* \gt \left(\frac{(m-(M+m)|c|}{\rho |c|M(1-|c|)\Psi^*}\right)^{\frac{1}{\rho+1}}+\frac{1}{\rho}\left(\frac{(m-(M+m)|c|}{\rho |c|M(1-|c|)\Psi^*}\right)^{-\frac{2\rho+1}{\rho+1}}+\frac{|c|M(1-|c|)}{m-(M+m)|c|}\gamma^* \gt 0, \end{equation} | (4.3) |
then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 4.1. We take
\zeta(t) = d(t),\; \; \; \; g(u) = \frac{1}{u^\rho},\; \; \; \; h(u) = 0. |
Then condition (H_1) is satisfied. Next, we consider the condition (H_2^{**}) is also satisfied. In fact, taking R = \frac{M(1-|c|)}{m-(M+m)|c|}\left(\frac{\Psi^*}{r^\rho}+\gamma^*\right), the \gamma_*-|c|R\geq r holds if and only if r verifies
\begin{equation*} \gamma_*-\frac{|c|M(1-|c|)}{m-(M+m)|c|}\left(\frac{\Psi^*}{r^\rho}+\gamma^*\right)\geq r, \end{equation*} |
or equivalently,
\begin{equation*} \gamma_{*}\geq f(r): = r+\frac{|c|M(1-|c|)}{m-(M+m)|c|}\left(\frac{\Psi^*}{r^\rho}+\gamma^*\right). \end{equation*} |
The function f(r) possesses a minimum at r_0: = \left(\frac{(m-(M+m))|c|}{\rho |c|M(1-|c|)\Psi^*}\right)^{\frac{1}{\rho+1}}. Let r = r_0. Then \gamma_*-cR\geq r holds if \gamma_{*}\geq f(r_0) , which is just the condition (4.3). The (H_2^{**}) holds directly by the choice of R , and it would remain to prove that R = \frac{M(1-|c|)}{m-(M+m)|c|}\left(\frac{\Psi^*}{r_0^\rho}+\gamma^*\right) > r_0. This is easily verified through elementary computations.
Case (Ⅱ) \gamma_* = 0 .
Theorem 4.2. Suppose M < \frac{64\pi^3}{81\sqrt{3}\omega^3} , c < 0 and |c| < \min\left\{\frac{m}{M+m}, \sigma\right\} hold. And f(t, u) satisfies (H_1) and (H_3) . Furthermore, assume that the following condition holds:
(H_4^{**}) There exists R > 0 such that \frac{(\Phi_R)_*}{1+|c|} < R < \frac{(\Phi_R)_*}{|c|} and
\frac{M(1-|c|)}{m-(M+m)|c|}\left(g\left((\Phi_R)_*-|c|R\right)\left(1+\frac{h(R)}{g(R)}\right)\Lambda^*+\|\gamma\|\right)\leq R. |
If \gamma_* > 0 , then (1.1) has at least one positive periodic solution.
Proof. We follow the same strategy and notation as in the proof of Theorem 4.1. Let R be the positive constant satisfying (H_4^{**}) and let r: = (\Phi_R)_*-|c|R ; then R > r > 0 since R > \frac{(\Phi_R)_*}{1+|c|} . Next we prove that \mathcal{Q}u+\mathcal{S}y\in \mathcal{K} for all u, \; y\in \mathcal{K} .
For any u, \; y\in \mathcal{K} , by Lemma 2.5 and (H_3) , we have
\begin{equation} \begin{split} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ = &\mathcal{P}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ \geq& \mathcal{T}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ = &\int^\omega_0G(t,s)f(s,u(s))ds+\gamma(t)-|c|y(t-\tau(t))\\ \geq&\int^\omega_0G(t,s)\phi_R(s)ds+\gamma(t)-|c|y(t-\tau(t))\\ \geq&(\Phi_R)_*-|c|R: = r \gt 0. \end{split} \end{equation} | (4.4) |
On the other hand, applying Lemma 2.4, we get
\begin{align*} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ = &\mathcal{P}(f(t,u(t))+e(t))+cy(t-\tau(t))\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\max\limits_{t\in[0,\omega]}|\mathcal{T}(f(t,u(t))+e(t))|+cy(t-\tau(t))\\ = &\frac{M(1-|c|)}{m-(M+m)|c|}\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds+\gamma(t)\right|-|c|y(t-\tau(t))\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left(\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)f(s,u(s))ds\right|+\|\gamma\|\right), \end{align*} |
since c < 0 and y\in K . By conditions (H_1) and (H_4^{**}) , it is clear
\begin{equation} \begin{split} &(\mathcal{Q}u)(t)+(\mathcal{S}y)(t)\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left(\max\limits_{t\in[0,\omega]}\left|\int^\omega_0G(t,s)\zeta(s)(g(u(s))+h(u(s)))ds\right| +\|\gamma\|\right)\\ \leq&\frac{M(1-|c|)}{m-(M+m)|c|}\left(g(r)\left(1+\frac{h(R)}{g(R)}\right)\Lambda^*+\|\gamma\|\right)\\ \leq&R. \end{split} \end{equation} | (4.5) |
Combining (4.4) and (4.5), we get \mathcal{Q}u+\mathcal{S}y\in \mathcal{K} for all u, \; y\in \mathcal{K} .
Similarly, we get that \mathcal{Q} is a completely continuous and \mathcal{S} is a contractive operator in X . Therefore, by Krasnoselskii's fixed point theorem, our results is proven.
Corollary 4.2. Suppose M < \frac{64\pi^3}{81\sqrt{3}\omega^3} , c < 0 , |c| < \min\left\{\frac{m}{M+m}, \sigma\right\} and (F_5) hold. If \gamma_* = 0 , then (1.1) has at least one positive periodic solution.
Proof. We apply Theorem 4.2. We take
\phi_L(t) = \frac{\hat{d}(t)}{L^\rho},\; \; \; \; \zeta(t) = d(t),\; \; \; \; g(u) = \frac{1}{u^\rho},\; \; \; \; h(u) = 0. |
Then conditions (H_1) and (H_2) hold, and the existence condition (H_4^{**}) becomes
\begin{equation} \frac{\hat{\Psi}_*}{|c|R^\rho} \gt R \gt \frac{\hat{\Psi}_*}{(1+|c|)R^\rho} = r, \end{equation} | (4.6) |
and
\begin{equation} \begin{split} \frac{M(1-|c|)}{m-(M+m)|c|}\left(\left(\frac{(1+|c|)R^\rho}{\hat{\Psi}_*}\right)^\rho\Psi^*+\|\gamma\|\right) \leq R. \end{split} \end{equation} | (4.7) |
Note that \Psi_* > 0 , since 0 < \rho < 1 , we can choose appropriate R > 0 so that (4.6) and (4.7) are satisfied and the proof is complete.
The paper is devoted to the existence of positive periodic solutions for (1.1), where the nonlinear function f has a singularity at u = 0 and sub-linearity condition at u = \infty for an appropriately chosen parameter. By employing Green's function and the Krasnoselskii fixed point theorem in cones, we prove the existence of positive periodic solutions to (1.1) with time-dependent delay for the first time. We would like to emphasize that the inclusion of the neutral operator in the singularity implies a new technical difficulty concerning the right choice of the operator.
The authors are grateful to anonymous referees for their constructive comments and suggestions which have greatly improved this paper. This work was supported by Fundamental Research Funds for the Universities of Henan Province (NSFRF170302).
The authors declare that they have no competing interests.
[1] | R. Agarwal, S. Grace, D. O'Regan, Oscillation Theory for Difference and Functional Differential Equations, Kluwer Acasemic, 2000. |
[2] | A. Ardjouni, A. Djoudi, Existence, uniqueness and positivity of solutions for a neutral nonlinear periodic differential equation, Comput. Appl. Math., 34 (2015), 17-27. |
[3] |
T. Candan, Existence of positive periodic solutions of first order neutral differential equations with variable coefficients, Appl. Math. Lett., 52 (2016), 142-148. doi: 10.1016/j.aml.2015.08.014
![]() |
[4] |
Z. Cheng, Q. Yuan, Damped superlinear Duffing equation with strong singularity of repulsive type, J. Fixed Point Theory Appl., 22 (2020), 1-18. doi: 10.1007/s11784-019-0746-3
![]() |
[5] |
Z. Cheng, F. Li, Weak and strong singularities for second-order nonlinear differential equations with a linear difference operator, J. Fixed. Point Theory Appl., 21 (2019), 1-23. doi: 10.1007/s11784-018-0638-y
![]() |
[6] |
Z. Cheng, F. Li, S. Yao, Positive solutions for second-order neutral differential equations with time-dependent deviating arguments, Filomat, 33 (2019), 3627-3638. doi: 10.2298/FIL1912627C
![]() |
[7] |
Z. Cheng, F. Li, Positive periodic solutions for a kind of second-order neutral differential equations with variable coefficient and delay, Mediterr. J. Math., 15 (2018), 1-19. doi: 10.1007/s00009-017-1047-y
![]() |
[8] |
W. Cheung, J. Ren, W. Han, Positive periodic solution of second-order neutral functional differential equations, Nonlinear Anal., 71 (2009), 3948-3955. doi: 10.1016/j.na.2009.02.064
![]() |
[9] |
J. Chu, P. Torres, M. Zhang, Periodic solution of second order non-autonomous singular dynamical systems, J. Differ. Equations, 239 (2007), 196-212. doi: 10.1016/j.jde.2007.05.007
![]() |
[10] |
J. Chu, Z. Zhou, Positive solutions for singular non-linear third-order periodic boundary value problems, Nonlinear Anal. Theor., 64 (2006), 1528-1542. doi: 10.1016/j.na.2005.07.005
![]() |
[11] |
B. Du, Y. Liu, I. Abbas, Existence and asymptotic behavior results of periodic solution for discrete-time neutral-type neural networks, J. Franklin Inst., 353 (2016), 448-461. doi: 10.1016/j.jfranklin.2015.11.013
![]() |
[12] |
A. Fonda, R. Manásevich, F. Zanolin, Subharmonics solutions for some second order differential equations with singularities, SIAM J. Math. Anal., 24 (1993), 1294-1311. doi: 10.1137/0524074
![]() |
[13] |
R. Hakl, P. Torres, On periodic solutions of second-order differential equations with attractive-repulsive singularities, J. Differ. Equations, 248 (2010), 111-126. doi: 10.1016/j.jde.2009.07.008
![]() |
[14] |
L. Lv, Z. Cheng, Positive periodic solution to superlinear neutral differential equation with time-dependent parameter, Appl. Math. Lett., 98 (2019), 271-277. doi: 10.1016/j.aml.2019.06.024
![]() |
[15] | R. Ma, R. Chen, Z. He, Positive periodic solutions of second-order differential equations with weak singularities, Appl. Math. Comput., 232 (2014), 97-103. |
[16] | J. Ren, Z. Cheng, S. Siegmund, Neutral operator and neutral differential equation, Abst. Appl. Anal., 2011 (2011), 1-22. |
[17] | J. Ren, S. Siegmund, Y. Chen, Positive periodic solutions for third-order nonlinear differential equations, Electron. J. Differ. Eq., 66 (2011), 1-19. |
[18] |
P. Torres, Weak singularities may help periodic solutions to exist, J. Differ. Equations, 232 (2007), 277-284. doi: 10.1016/j.jde.2006.08.006
![]() |
[19] | D. Wang, Positive periodic solutions for a nonautonomous neutral delay prey-predator model with impulse and Hassell-Varley type functional response, P. Am. Math. Soc., 142 (2014), 623-638. |
[20] |
H. Wang, Positive periodic solutions of singular systems with a parameter, J. Differ. Equations, 249 (2010), 2986-3002. doi: 10.1016/j.jde.2010.08.027
![]() |
[21] |
Z. Wang, T. Ma, Existence and multiplicity of periodic solutions of semilinear resonant Duffing equations with singularities, Nonlinearity, 25 (2012), 279-307. doi: 10.1088/0951-7715/25/2/279
![]() |
[22] |
J. Wu, Z. Wang, Two periodic solutions of second-order neutral functional differential equations, J. Math. Anal. Appl., 329 (2007), 677-689. doi: 10.1016/j.jmaa.2006.06.084
![]() |
[23] | T. Xiang, R. Yuan, Existence of periodic solutions for p-Laplacian neutral functional equation with multiple deviating arguments, Topol. Method. Nonlinear Anal., 37 (2011), 235-258. |
[24] | Y. Xin, H. Liu, Singularity problems to fourth-order Rayleigh equation with time-dependent deviating argument, Adv. Differ. Equ., 368 (2018), 1-15. |
[25] | Y. Xin, H. Liu, Existence of periodic solution for fourth-order generalized neutral p-Laplacian differential equation with attractive and repulsive singularities, J. Inequal. Appl., 259 (2018), 1-18. |
[26] | S. Yao, J. Liu, Study on variable coefficients singular differential equation via constant coefficients differential equation, Bound. Value Probl., 3 (2019), 1-24. |