Mukhiddin I. Muminov and Ali H. M. Murid

# Existence conditions for periodic solutions of second-order neutral delay differential equations with piecewise constant arguments

Open Access
De Gruyter | Published online: March 10, 2020

# Abstract

In this paper, we describe a method to solve the problem of finding periodic solutions for second-order neutral delay-differential equations with piecewise constant arguments of the form x″(t) + px″(t − 1) = qx([t]) + f(t), where [⋅] denotes the greatest integer function, p and q are nonzero real or complex constants, and f(t) is complex valued periodic function. The method reduces the problem to a system of algebraic equations. We give explicit formula for the solutions of the equation. We also give counter examples to some previous findings concerning uniqueness of solution.

MSC 2010: 35-xx; 47-xx; 47A10

## 1 Introduction

In the study of almost periodic differential equations, many useful methods have been developed in the classical references such as Hale and Lunel [1], Fink [2], Yoshizawa [3], and Hino et al. [4].

Differential equations with piecewise constant arguments are usually referred to as a hybrid system, and could model certain harmonic oscillators with almost periodic forcing. For some excellent works in this field we refer the reader to [5, 6] and references therein, and for a survey of work on differential equations with piecewise constant arguments we refer the reader to [7, 8].

A recently published paper [9] has studied the differential equation of the form

x ( t ) + p x ( t 1 ) = q x ( 2 [ t + 1 2 ] ) + f ( t ) , (1)

where [⋅] denotes the greatest integer function, p and q are nonzero constants, and f(t) is a 2n-periodic continuous function. The 2n-periodic solvable problem (1) is reduced to the study a system of n + 1 linear equations. Furthermore, by applying the well-known properties of linear system in algebra, all existence conditions are described for 2n-periodic solutions that yields explicit formula for the solutions of (1).

In this paper we study certain functional differential equation of neutral delay type with piecewise constant arguments of the form

x ( t ) + p x ( t 1 ) = q x ( [ t ] ) + f ( t ) , (2)

where p and q are nonzero real or complex constants, f(t) is a complex valued n-periodic continuous function defined on R.

The papers [5, 6] have investigated the existence of almost periodic solutions of (2), when f is an almost periodic function, while Nth-order differential equation of neutral delay type are studied in [10, 11]. In these works some theorems for the existence and uniqueness of the almost periodic solutions have been obtained. However, there are some incorrectness of uniqueness results given in [5, 10, 11, 12].

In the present paper, by adapting the method in [9], we give all exact conditions for the uniqueness, infiniteness and emptiness of n-periodic solutions of (2), in the case when f is n-periodic. We give explicit formula for the exact solutions of the equation. For some functions f, we also show that equation (2) may have infinitely many 3-periodic solutions.

Throughout this paper, we use the notations R for the set of reals and Z for the set of integers.

## 2 Definition of periodic solution

A solution of (2) is defined in [13] as follows

## Definition 2.1

A function x is called a solution of (2) if the following conditions are satisfied:

1. x and xare continuous on R;

2. the second-order derivative of x(t) exists everywhere, with possible exception at points t = n, nZ, where one-sided second-order derivatives of x(t) exist;

3. x satisfies (2) on each interval (n, n + 1) with integer nZ.

Comments: The importance of the differentiability condition imposed on x is given in Definition 13 of [13], since we deal with second-order differential equations. This condition may admit uniqueness condition of periodic solution of (2), however it is a sufficient condition for the uniqueness condition (see Theorem 3.2 (ii) of this paper). Equation (2) may have infinite number of periodic continuous functions, satisfying (2) on each interval (n, n + 1) and may not have derivative at each point n, nZ (see Examples 1 and 2 below). Moreover, if we omit the continuity condition of x′ on R, then the uniqueness of periodic solution of (2) does not hold, and well-posedness of (2) is not true. In the definition of solution, the continuity condition of x′ on R is omitted in many works (see, for example, [5], [11] and [12]) consequently, the uniqueness of pseudo periodic and hence almost periodic solution does not hold. We illustrate this in the following two examples.

## Example 1

Let f(t) = −3π2 cos π t, p = 2, q = 4. Then the function

x ( t ) = 15 2 α + α t 4 3 Q ( t ) + 3 3 cos π t , t [ 0 , 2 ] (3)

satisfies (2) with the function Q defined on [0, 2] as

Q ( t ) = 9 t 2 2 , t [ 0 , 1 ) , 9 2 9 ( t 1 ) + ( 27 + 3 α ) ( t 1 ) 2 2 , t [ 1 , 2 ] ,

where α is any number. One can easily check that x(t) is continuous on R. But x′(t) is discontinuous at kZ when α ≠ −6. If α = −6, x(t) is the exact solution of (2) (see Figures 1 and 2).

### Figure 1

The graph of x(t) when α = −10.

### Figure 2

The graph of x(t) when α = −6.

## Example 2

Let f(t) = 1 π 2 cos π t and p2 ≠ 1, 8 − 8 p + q ≠ 0. Then, by the definition of solution given in the papers [5], [10] and [11], the 2-periodic functions

x α ( t ) = x ( 0 ) + α t + q 1 p 2 Q ( t ) + 1 cos π t 1 p (4)

are solutions of the equation (2), where α is any number and the function Q defined on [0, 2] as

Q ( t ) = ( x ( 0 ) p x ( 1 ) ) t 2 2 for t [ 0 , 1 ) ,

Q ( t ) = Q ( 1 0 ) + ( x ( 0 ) p x ( 1 ) ) ( t 1 ) + ( x ( 1 ) p x ( 0 ) ) ( t 1 ) 2 2 for t [ 1 , 2 ) , x ( 0 ) = 2 ( 6 q + 18 p q 4 α + 4 p 2 α q α + p q α ) q ( 8 8 p + q ) , x ( 1 ) = 2 ( 18 q 6 p q 4 α + 4 p 2 α + q α p q α ) ( 8 + 8 p q ) q .

Note that x(0) = xα(2), which gives continuity of xα on R.

It has been claimed in the papers [5], [10] and [11] that if ∣λi∣ ≠ 1, i = 1, 2, 3, then equation (2) has a unique solution, where λi, i = 1, 2, 3, are the eigenvalues of the matrix

A = 1 + q 2 p 1 p q 1 0 1 0 0 .

For the case, when p = 1 2 , q = 3, the matrix A has a form

A = 2 1 1 2 3 1 0 1 0 0 .

One can easily check that the eigenvalues λi, i = 1, 2, 3 are reals and ∣λi∣ ≠ 1, i = 1, 2, 3. Moreover, the conditions of the Main results of these papers are satisfied. Example 2 shows incorrectness of the results Theorem 2.11 in [10], Theorem 3.3 in [11] and Theorem 1 in [5], which claim uniqueness of the almost periodic solutions of (2).

## 3 2 and 3-periodic solutions

In this section we give a method of finding periodic solutions of (2) and their existence conditions. Let f be n-periodic continuous function. We consider two cases n = 2 and n = 3 in this section before considering the general case n in Section 4.

The case n = 2. We seek a function x as a 2-periodic function that solves (2).

Equation (2) is equivalent to

x ( t + 1 ) + p x ( t ) = q x ( [ t + 1 ] ) + f ( t + 1 ) .

Since x″(t + 1) = x″(t − 1), this becomes

x ( t 1 ) = p x ( t ) + q x ( [ t + 1 ] ) + f ( t + 1 ) . (5)

Substitute this into (2) and assuming p2 ≠ 1, we get

x ( t ) = q 1 p 2 ( x ( [ t ] p x ( [ t + 1 ] ) ) + 1 1 p 2 ( f ( t ) p f ( t + 1 ) ) . (6)

Integrating (6) twice on [0, t], t < 2, we obtain

x ( t ) = x ( 0 ) + x ( 0 ) t + q 1 p 2 Q ( t ) + F 2 ( t ) , (7)

where

Q ( t ) = 0 t 0 t 1 ( x ( [ s ] ) p x ( [ s + 1 ] ) d s d t 1 , F 2 ( t ) = 1 1 p 2 0 t 0 t 1 ( f ( s ) p f ( s + 1 ) ) d s d t 1 . (8)

The function Q on [0, 2) can be represented as

Q ( t ) = ( x ( 0 ) p x ( 1 ) ) t 2 2 for t [ 0 , 1 )

and

Q ( t ) = 0 1 0 t 1 X ( s ) d s d t 1 + 1 t 0 1 X ( s ) d s d t 1 + 1 t 1 t 1 X ( s ) d s d t 1 = Q ( 1 0 ) + ( x ( 0 ) p x ( 1 ) ) ( t 1 ) + ( x ( 1 ) p x ( 0 ) ) ( t 1 ) 2 2 for t [ 1 , 2 ) ,

where X(s) = x([s]) − px([s + 1] and Q(a − 0) denotes a left limit of Q(t) at the point t = a.

This shows that the right-hand side of (7) contains only unknown numbers x(0), x(1) and x′(0). Since x and x′ are continuous and periodic, they should satisfy x(0) = x(2), x′(0) = x′(2). To find x(0), x(1) and x′(0) we apply (7) to get the system of equations

x ( 1 ) = x ( 0 ) + x ( 0 ) + q 1 p 2 Q ( 1 ) + F 2 ( 1 ) , x ( 2 ) = x ( 0 ) + 2 x ( 0 ) + q 1 p 2 Q ( 2 ) + F 2 ( 2 ) , x ( 0 ) = x ( 0 ) + q 1 p 2 Q ( 2 ) + F 2 ( 2 ) .

Taking into account

Q ( 1 ) = 1 2 ( x ( 0 ) p x ( 1 ) ) , Q ( 2 ) = 3 2 ( x ( 0 ) p x ( 1 ) ) + 1 2 ( x ( 1 ) p x ( 0 ) ) ,

the last system of equations yields

( 1 + 1 2 q 1 p 2 ) x ( 0 ) ( 1 + 1 2 q p 1 p 2 ) x ( 1 ) + x ( 0 ) = F 2 ( 1 ) , 1 2 q 1 p 2 ( 3 p ) x ( 0 ) + 1 2 q 1 p 2 ( 1 3 p ) x ( 1 ) + 2 x ( 0 ) = F 2 ( 2 ) , q 1 + p x ( 0 ) + q 1 + p x ( 1 ) = F 2 ( 2 ) . (9)

The determinant D(p, q) of this system is

D ( p , q ) := 1 + 1 2 q 1 p 2 1 1 2 q p 1 p 2 1 1 2 q 1 p 2 ( 3 p ) 1 2 q 1 p 2 ( 1 3 p ) 2 q 1 + p q 1 + p 0 = 4 q 1 + p .

Since D(p, q) ≠ 0, we get

## Theorem 3.1

Let p2 − 1 ≠ 0 and f be 2-periodic continuous function. Then equation (2) has a unique 2-periodic solution having the form (7), where (x(0), x(1), x′(0)) is the unique solution of (9).

## Example 3

We consider equation (2) with p = 3, q = 1 and 2-periodic function f ( t ) = t , t [ 0 , 1 ) , 2 t , t [ 1 , 2 ] . It can be shown from (8) that

F 2 ( t ) = 1 24 t 2 ( 9 + 2 t ) , t [ 0 , 1 ) , 1 8 ( 5 3 + 4 t + 1 3 ( 1 + t ) 2 ( 1 + 2 t ) ) , t [ 1 , 2 ] , Q ( t ) = 47 t 2 24 , t [ 0 , 1 ) , 1 24 ( 2 4 t + 49 t 2 ) , t [ 1 , 2 ] .

Solving (9), we obtained x ( 0 ) = 97 48 , x ( 1 ) = 95 48 , x ( 0 ) = 1 192 . The graph of 2-periodic solution (7) is shown in Figure 3.

### Figure 3

The graph of 2-periodic solution x(t) of equation (2) for Example 3.

## Example 4

Let p = 2i, q = 1 and f(t) = eiπt. For this case, it can be shown that the 2-periodic solution of equation (2) is

x ( t ) = 1 50 π 2 ( ( 10 + 20 i ) e i π t ( 3 4 i ) ( 1 + t ) t ) , t [ 0 , 1 ) , 1 50 π 2 ( ( 10 + 20 i ) e i π t + ( 3 4 i ) ( 2 3 t + t 2 ) ) , t [ 1 , 2 ] .

The case n = 3. Let a function x be a 3-periodic function. From (2) we have

x ( t ) + p x ( t 1 ) = q x ( [ t ] ) + f ( t ) , x ( t + 1 ) + p x ( t ) = q x ( [ t + 1 ] ) + f ( t + 1 ) , x ( t 1 ) + p x ( t + 1 ) = q x ( [ t + 2 ] ) + f ( t + 2 ) . (10)

In the last equation we have used the fact that x″(t + 2) = x″(t − 1). The system of equations (10) of x″(t − 1), x″(t) and x″(t + 1) is solvable iff

p 1 0 0 p 1 1 0 p = 1 + p 3 0.

From this system of equations, assuming p3 ≠ −1, we get

x ( t ) = q 1 + p 3 ( p 2 x ( [ t + 1 ] ) p x ( [ t + 2 ] ) + x ( [ t ] ) ) + 1 1 + p 3 ( p 2 f ( t + 1 ) p f ( t + 2 ) + f ( t ) ) . (11)

Integrating (11) two times on [0, t], t < 3, we have

x ( t ) = x ( 0 ) + x ( 0 ) t + q 1 + p 3 Q 3 ( t ) + F 3 ( t ) , (12)

where

Q 3 ( t ) = 0 t 0 t 1 ( p 2 x ( [ s + 1 ] ) p x ( [ s + 2 ] ) + x ( [ s ] ) ) d s d t 1 , F 3 ( t ) = 1 1 + p 3 0 t 0 t 1 ( p 2 f ( s + 1 ) p f ( s + 2 ) + f ( s ) ) d s d t 1 .

The function Q3 on [0, 3) can be presented by

Q 3 ( t ) = ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) t 2 2 for t [ 0 , 1 ) , Q 3 ( t ) = 0 1 0 t 1 Y ( s ) d s d t 1 + 1 t 0 1 Y ( s ) d s d t 1 + 1 t 1 t 1 Y ( s ) d s d t 1 = Q 3 ( 1 0 ) + ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) ( t 1 ) + ( p 2 x ( 2 ) p x ( 0 ) + x ( 1 ) ) ( t 1 ) 2 2 for t [ 1 , 2 ) ,

where Y(s) = p2x([s + 1]) − px([s + 2]) + x([s]),

Q 3 ( t ) = 0 2 0 t 1 Y ( s ) d s d t 1 + 2 t 0 1 Y ( s ) d s d t 1 + 2 t 1 2 Y ( s ) d s d t 1 + 2 t 2 t 1 Y ( s ) d s d t 1 = Q ( 2 0 ) + ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) ( t 2 ) + ( p 2 x ( 2 ) p x ( 0 ) + x ( 1 ) ) ( t 2 ) + ( p 2 x ( 0 ) p x ( 1 ) + x ( 2 ) ) ( t 2 ) 2 2 for t [ 2 , 3 ) .

Hence the right-hand side of (12) contains only unknown variables x(0), x(1), x(2) and x′(0). Since x and x′ are continuous and periodic, they should satisfy

x ( 0 ) = x ( 3 ) , x ( 0 ) = x ( 3 ) .

Using these equations and (12) we have

x ( 1 ) = x ( 0 ) + x ( 0 ) + q 1 + p 3 Q 3 ( 1 ) + F 3 ( 1 ) , x ( 2 ) = x ( 0 ) + 2 x ( 0 ) + q 1 + p 3 Q 3 ( 2 ) + F 3 ( 2 ) , x ( 3 ) = x ( 0 ) + 3 x ( 0 ) + q 1 + p 3 Q 3 ( 3 ) + F 3 ( 3 ) , x ( 0 ) = x ( 0 ) + q 1 + p 3 Q 3 ( 3 ) + F 3 ( 3 ) . (13)

Note that

Q 3 ( 1 ) = 1 2 ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) , Q 3 ( 2 ) = 3 2 ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) + 1 2 ( p 2 x ( 2 ) p x ( 0 ) + x ( 1 ) ) , Q 3 ( 3 ) = 5 2 ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) + 3 2 ( p 2 x ( 2 ) p x ( 0 ) + x ( 1 ) ) + 1 2 ( p 2 x ( 0 ) p x ( 1 ) + x ( 2 ) ) , Q 3 ( 3 ) = 3 2 ( p 2 x ( 1 ) p x ( 2 ) + x ( 0 ) ) + 1 2 ( p 2 x ( 2 ) p x ( 0 ) + x ( 1 ) ) .

Therefore, we have

( 1 + 1 2 q 1 + p 3 ) x ( 0 ) + ( 1 2 q 1 + p 3 1 ) x ( 1 ) 1 2 q p 1 + p 3 x ( 2 ) + x ( 0 ) = F 2 ( 1 ) , ( 1 + q 1 + p 3 3 p 2 ) x ( 0 ) + q 1 + p 3 3 p 2 + 1 2 x ( 1 ) + ( q 1 + p 3 3 p + p 2 2 1 ) x ( 2 ) + 2 x ( 0 ) = F 2 ( 2 ) , q 1 + p 3 5 3 p + p 2 2 x ( 0 ) + q 1 + p 3 5 p 2 + 3 p 2 x ( 1 ) + q 1 + p 3 5 p + 3 p 2 + 1 2 x ( 2 ) + 3 x ( 0 ) = F 2 ( 3 ) , q 1 + p 3 ( 1 p + p 2 ) x ( 0 ) + q 1 + p 3 ( 1 p + p 2 ) x ( 1 ) + q 1 + p 3 ( 1 p + p 2 ) x ( 2 ) = F 2 ( 3 ) . (14)

Let D3(p, q) be the determinant of the matrix

A = 1 + 1 2 q 1 + p 3 1 2 q 1 + p 3 1 1 2 q p 1 + p 3 1 1 + q 1 + p 3 3 p 2 q 1 + p 3 3 p 2 + 1 2 q 1 + p 3 3 p + p 2 2 1 2 q 1 + p 3 5 3 p + p 2 2 q 1 + p 3 5 p 2 + 3 p 2 q 1 + p 3 5 p + 3 p 2 + 1 2 3 q 1 + p 3 ( 1 p + p 2 ) q 1 + p 3 ( 1 p + p 2 ) q 1 + p 3 ( 1 p + p 2 ) 0 .

It can be shown that

D 3 ( p , q ) = q ( 36 + 36 p 2 + 6 p ( 6 + q ) + 6 q + q 2 ) 4 ( 1 + p 3 ) .

Summarizing these results, we get

## Theorem 3.2

Let p3 + 1 ≠ 0 and f be a 3-periodic continuous function.

1. If D3(p, q) ≠ 0, then equation (2) has a unique 3-periodic solution having the form (12), where (x(0), x(1), x(2), x′(0)) is the unique solution of (14).

2. If D3(p, q) = 0 and F3(1) = F3(2) = F3(3) = F 3 (3) = 0, then equation (2) has infinite number of 3-periodic solutions having the form

x ( t ) = α ( x ( 0 ) + x ( 0 ) t + q 1 + p 3 Q 3 ( t ) ) + F 3 ( t ) ,

where (x(0), x(1), x(2), x′(0)) is an eigenvector of A corresponding to 0, α is any number.

3. If D3(p, q) = 0 and the rank[A] < rank[A, FT], F = (F3(1), F3(2), F3(3), F 3 (3)), then equation (2) does not have any 3-periodic solution.

## Remark 3.1

We emphasize again that in the definition of solution, it is important to have the continuity condition of its derivative. If we omit this condition, we must remove the equality x′(0) = x′(3) in (13). Let p = 1. Then the equation (13) is equivalent to 3 × 3 system

F 2 ( 1 ) α = ( 1 + 1 2 q 2 ) x ( 0 ) + ( 1 2 q 2 1 ) x ( 1 ) 1 2 q 2 x ( 2 ) , F 2 ( 2 ) 2 α = ( 1 + q 2 2 2 ) x ( 0 ) + q 2 4 2 x ( 1 ) + ( q 2 1 2 1 ) x ( 2 ) , F 2 ( 3 ) 3 α = q 2 3 2 x ( 0 ) + q 2 7 2 x ( 1 ) + q 2 1 2 x ( 2 ) , (15)

where α is any number. Since the determinant D(q) of the system of equations (15) is

D ( q ) = 1 4 q ( 9 + 3 q + q 2 4 ) ,

(15) has a unique solution (x(0), x(1), x(2)) when D(q) ≠ 0. Therefore, for any continuous 3-periodic f and q with q ( 9 + 3 q + q 2 4 ) 0 the continuous function

x α ( t ) = x ( 0 ) + α t + q 2 Q 3 ( t ) + F 3 ( t )

satisfies (2). In particular, the functions {xα} are well-defined for q and 3- periodic functions f satisfying the conditions of the Theorem 2.1 in [12], which claim on uniqueness of almost periodic solutions. This example shows that the Theorem 2.1 in [12] is incorrect.

We denote by 𝓛0 the class of continuous 3-periodic functions f with F3(1) = F3(2) = F3(3) = F 3 (3) = 0, where

F 3 ( t ) = 1 1 + p 3 0 t 0 t 1 ( p 2 f ( s + 1 ) p f ( s + 2 ) + f ( s ) ) d s d t 1 .

## Remark 3.2

Theorem (3.2) (ii) shows that the differentiability of a solution of (2) does not ensure uniqueness of the 3-periodic and hence almost periodic solutions of this equation. We can observe the existence of infinite number of 3-periodic solutions for the case when f belongs to 𝓛0. Let f(t) = fk(t), fk(t) = cos 2kπt, k = 1, 2, … and p = 1, q = −6. Then F 3 ( t ) = 1 1 + p 2 1 ( 2 k π ) 2 ( 1 cos 2 k π t ) and hence fk ∈ 𝓛0. The matrix A is represented by

A = 1 2 5 2 3 2 1 2 6 2 2 9 2 21 2 3 2 3 3 3 3 0 .

One can cheek that, det(A) = D3(1, −6) = 0 and the number 0 is an eigenvalue of A with multiplicity 2. The corresponding eigenvectors are x 1 = ( 1 2 , 1 2 , 0 , 1 ) , x2 = (−2, 1, 1, 0). By Theorem 3.2 the functions

x 1 ( t ) = ( 1 2 + t 3 Q 31 ( t ) ) α + 1 8 1 ( k π ) 2 ( 1 cos 2 k π t ) , (16)

x 2 ( t ) = ( 2 3 Q 32 ( t ) ) α + 1 8 1 ( k π ) 2 ( 1 cos 2 k π t ) , (17)

are 3-periodic solutions of (2), where α is any nonzero constant,

Q 31 ( t ) = 0 a s t [ 0 , 1 ) , ( t 1 ) 2 2 a s t [ 1 , 2 ) , 3 2 + t ( t 2 ) 2 2 a s t [ 2 , 3 ] , Q 32 ( t ) = t 2 a s t [ 0 , 1 ) , 1 2 t + 2 ( t 1 ) 2 a s t [ 1 , 2 ) , 5 + 2 t ( t 2 ) 2 a s t [ 2 , 3 ] .

## 4 n-periodic solutions

We next solve equation (2), where f is periodic with positive integer period n ≥ 3. It is clear that to seek a function x as a periodic function, we assume that x(t) = x(t + n).

It follows from (2) and periodicity of x(t) that

( x ( t ) + p x ( t 1 ) ) = q x ( [ t ] ) + f ( t ) , ( x ( t + 1 ) + p x ( t ) ) = q x ( [ t + 1 ] ) + f ( t + 1 ) , ( x ( t + n 2 ) + p x ( t + n 3 ) ) = q x ( [ t + n 2 ] ) + f ( t + n 2 ) , ( x ( t 1 ) + p x ( t + n 2 ) ) = q x ( [ t + n 1 ] ) + f ( t + n 1 ) . (18)

Assuming the right-hand sides of (18) are known, we consider this system of equations with respect to

x ( t 1 ) , x ( t ) , , x ( t + n 1 ) .

It is solvable if and only if Δ(p) ≠ 0, where Δ(p) = det P and P is the n × n matrix

P = p 1 0 0 0 0 p 1 0 0 0 0 0 p 1 1 0 0 0 p .

Observe that

Δ ( p ) = det P = p n ( 1 ) n .

Assuming pn ≠ (−1)n, we can find x″(t) from (18), i.e.,

x ( t ) = Δ ( p ; t ) Δ ( p ) , (19)

where Δ(p; t) = det Q and Q is the n × n matrix

Q = p Q 1 ( t ) 0 0 0 0 Q 2 ( t ) 1 0 0 0 Q n 1 ( t ) 0 p 1 1 Q n 0 0 p ,

Q k ( t ) = q x ( [ t + k 1 ] ) + f ( t + k 1 ) , k = 1 , 2 , , n .

Using the properties of determinant, we have

det Q = Q 1 ( t ) 0 1 0 0 0 0 p 1 0 0 0 0 0 p 1 1 0 0 0 p + Q 2 ( t ) p 0 0 0 0 0 p 1 0 0 0 0 0 p 1 1 0 0 0 p + + Q n 1 ( t ) ( 1 ) n 1 p 0 0 0 0 0 1 0 0 0 0 0 0 1 0 1 0 0 0 p + Q n ( t ) ( 1 ) n p 0 0 0 0 0 1 0 0 0 0 0 0 1 0 1 0 0 0 1 = ( 1 ) n + 1 Q 1 ( t ) + Q 2 ( t ) p n 1 + + ( 1 ) n 1 Q n 1 ( t ) p 2 + ( 1 ) n Q n ( t ) p ) .

Taking into Q1(t) = Qn+1(t), the equation (19) represents

x ( t ) = 1 Δ ( p ) k = 1 n ( 1 ) k + 1 Q k + 1 ( t ) p n k = q Δ ( p ) k = 1 n ( 1 ) k + 1 x ( [ t + k ] ) p n k + 1 Δ ( p ) k = 1 n ( 1 ) k + 1 f ( t + k ) p n k . (20)

Integrating (20) we get

x ( t ) = x ( 0 ) + x ( 0 ) t + q Δ ( p ) Q ~ n ( t ) + F n ( t ) , (21)

where

Q ~ n ( t ) = 0 t 0 t 1 R ( s ) d s d t 1 , R ( s ) = k = 1 n ( 1 ) k + 1 x ( [ s + k ] ) p n k , F n ( t ) = 1 Δ ( p ) 0 t 0 t 1 k = 1 n ( 1 ) k + 1 f ( s + k ) p n k d s d t 1 .

We set

X [ s ] = k = 1 n ( 1 ) k + 1 x ( [ s + k ] ) p n k .

Then the function n on [0, n) can be represented by

Q ~ n ( t ) = X [ 0 ] t 2 2 , t [ 0 , 1 ) , Q ~ n ( t ) = 0 1 0 t 1 R ( s ) d s d t 1 + 1 t 0 1 R ( s ) d s d t 1 + 1 t 1 t 1 R ( s ) d s d t 1 = Q ~ n ( 1 0 ) + X [ 0 ] ( t 1 ) + X [ 1 ] ( t 1 ) 2 2 , t [ 1 , 2 ) , Q ~ n ( t ) = 0 2 0 t 1 R ( s ) d s d t 1 + 2 t 0 1 R ( s ) d s d t 1 + 2 t 1 2 R ( s ) d s d t 1 + 2 t 2 t 1 R ( s ) d s d t 1 = Q ~ n ( 2 0 ) + ( X [ 0 ] + X [ 1 ] ) ( t 2 ) + X [ 2 ] ( t 2 ) 2 2 , t [ 2 , 3 ) , Q ~ n ( t ) = 0 n 1 0 t 1 R ( s ) d s d t 1 + n 1 t l = 1 n 1 l 1 l R ( s ) d s d t 1 + n 1 t n 1 t R ( s ) d s d t 1 = Q ~ n ( n 1 0 ) + l = 1 n 1 X [ l 1 ] ( t n + 1 ) + X [ n 1 ] ( t n + 1 ) 2 2 , t [ n 1 , n ) .

We remark that the functions n and Fn are twice differentiable on (n − 1, n) and there exist one-sided second derivatives at t = n − 1 and t = n.

The right-hand side of (21) contains only unknown numbers x(0), …, x(n − 1), x′(0). Using the periodicity conditions x and continuity x′ from (21) we have n + 1 system of equations

x ( 1 ) = x ( 0 ) + x ( 0 ) + q Δ ( p ) Q n ( 1 ) + F n ( 1 ) , x ( n 1 ) = x ( 0 ) + ( n 1 ) x ( 0 ) + q Δ ( p ) Q n ( n 1 ) + F n ( n 1 ) , x ( n ) = x ( 0 ) + n x ( 0 ) + q Δ ( p ) Q n ( n ) + F n ( n ) , x ( n ) = x ( 0 ) + q Δ ( p ) Q n ( n ) + F n ( n ) . (22)

The last system of equations can be written as

x ( 1 ) = x ( 0 ) + x ( 0 ) + 1 2 q Δ ( p ) X ( 0 ) + F n ( 1 ) , x ( 2 ) = x ( 0 ) + 2 x ( 0 ) + q Δ ( p ) k = 1 2 ( r = 1 k 1 X ( r 1 ) + 1 2 X ( k 1 ) ) + F n ( 2 ) , x ( n 1 ) = x ( 0 ) + ( n 1 ) x ( 0 ) + q Δ ( p ) k = 1 n 1 ( r = 1 k 1 X ( r 1 ) + 1 2 X ( k 1 ) ) + F n ( n 1 ) , 0 = n x ( 0 ) + q Δ ( p ) k = 1 n ( r = 1 k 1 X ( r 1 ) + 1 2 X ( k 1 ) ) + F n ( n ) , 0 = q Δ ( p ) l = 1 n X [ l 1 ] + F n ( n ) . (23)

Since

k = 1 m ( r = 1 k 1 X ( r 1 ) + 1 2 X ( k 1 ) ) = k = 0 m 1 ( m k 1 2 ) X ( k ) ,

we have

x ( 1 ) = x ( 0 ) + x ( 0 ) + 1 2 q Δ ( p ) X ( 0 ) + F n ( 1 ) , x ( 2 ) = x ( 0 ) + 2 x ( 0 ) + q Δ ( p ) k = 0 1 ( 2 k 1 2 ) X ( k ) + F n ( 2 ) , x ( n 1 ) = x ( 0 ) + ( n 1 ) x ( 0 ) + q Δ ( p ) k = 0 n 2 ( n 1 k 1 2 ) X ( k ) + F n ( n 1 ) , 0 = n x ( 0 ) + q Δ ( p ) k = 0 n 1 ( n k 1 2 ) X ( k ) + F n ( n ) , 0 = q Δ ( p ) l = 1 n X [ l 1 ] + F n ( n ) . (24)

To evaluate the above summation, we apply the following lemma.

## Lemma 4.1

For any positive integer number l ≥ 2 the equality

k = 0 l 1 ( l k 1 2 ) X ( k ) = r = 1 n P l r ( p ) x ( r )

holds, where

P l r ( p ) = k = 0 r 1 ( l k 1 2 ) ( 1 ) r k + 1 p n r + k + k = r l 1 ( l k 1 2 ) ( 1 ) n + r + 1 k p k r .

## Proof

Let α l k = l k 1 2 .

Δ ( p ) q k = 0 l 1 α l k X ( k ) = k = 0 l 1 α l k r = 1 n ( 1 ) r + 1 x ( k + r ) p n r = k = 0 l 1 α l k s = k + 1 k + n ( 1 ) s k + 1 x ( s ) p n s + k = k = 0 l 1 α l k ( s = k + 1 n ( 1 ) s k + 1 x ( s ) p n s + k + s = n + 1 k + n ( 1 ) s k + 1 x ( s ) p n s + k ) = k = 0 l 1 α l k ( s = k + 1 n ( 1 ) s k + 1 x ( s ) p n s + k + r = 1 k ( 1 ) r + n k + 1 x ( r + n ) p r + k ) = k = 0 l 1 α l k ( r = 1 k ( 1 ) r + n k + 1 x ( r ) p r + k + r = k + 1 n ( 1 ) r k + 1 x ( r ) p n r + k ) = α l 0 r = 1 n ( 1 ) r + 1 x ( r ) p n r + α l 1 ( r = 1 1 ( 1 ) r + n x ( r ) p 1 r + r = 2 n ( 1 ) r x ( r ) p n r + 1 ) + + α l , l 1 ( r = 1 l 1 ( 1 ) r + n l + 2 x ( r ) p r + l 1 + r = l n ( 1 ) r l + 2 x ( r ) p n r + l 1 ) = ( α l 0 p n 1 + α l 1 ( 1 ) 1 + n p 0 + + α l , l 1 ( 1 ) n l + 3 p l 2 ) x ( 1 ) + + ( α l 0 ( 1 ) l p n l + 1 + α l 1 ( 1 ) l 1 p n l + 2 + + α l , l 2 ( 1 ) 2 p n 1 + α l , l 1 ( 1 ) n + 1 p 0 ) x ( l 1 ) + ( α l 0 ( 1 ) l + 1 p n l + α l 1 ( 1 ) l p n l + 1 + + α l , l 1 ( 1 ) 2 p n 1 ) x ( l ) + + ( α l 0 ( 1 ) n + 1 p 0 + α l 1 ( 1 ) n p + + α l , l 1 ( 1 ) n l + 2 p l 1 ) x ( n ) = r = 1 n x ( r ) ( k = 0 r 1 α l k ( 1 ) r k + 1 p n r + k + k = r l 1 α l k ( 1 ) n + r + 1 k p k r ) .

By Lemma 4.1, equation (24) can be written as

x ( 1 ) = x ( 0 ) + x ( 0 ) + q Δ ( p ) r = 1 n P 1 r ( p ) x ( r ) + F n ( 1 ) , x ( 2 ) = x ( 0 ) + 2 x ( 0 ) + q Δ ( p ) r = 1 n P 2 r ( p ) x ( r ) + F n ( 2 ) , x ( n 1 ) = x ( 0 ) + ( n 1 ) x ( 0 ) + q Δ ( p ) r = 1 n P n 1 , r ( p ) x ( r ) + F n ( n 1 ) , 0 = n x ( 0 ) + q Δ ( p ) r = 1 n P n r ( p ) x ( r ) + F n ( n ) , 0 = q Δ ( p ) r = 1 n P 0 ( p ) x ( r ) F n ( n ) , (25)

where

P 0 ( p ) = k = 1 n ( 1 ) k + 1 p n k .

We rewrite (25) as

( 1 + q Δ ( p ) P 1 n ( p ) ) x ( 0 ) + ( q Δ ( p ) P 11 ( p ) 1 ) x ( 1 ) + q Δ ( p ) r = 2 n 1 P 1 r ( p ) x ( r ) + x ( 0 ) = F ( 1 ) , ( 1 + q Δ ( p ) P 2 n ( p ) ) x ( 0 ) + ( q Δ ( p ) P 22 ( p ) 1 ) x ( 2 ) + q Δ ( p ) r 2 n 1 P 1 r ( p ) x ( r ) + 2 x ( 0 ) = F ( 1 ) , ( 1 + q Δ ( p ) P n 1 , n ( p ) ) x ( 0 ) + ( q Δ ( p ) P n 1 , n 1 ( p ) 1 ) x ( n 1 ) + q Δ ( p ) r = 1 n 2 P 1 r ( p ) x ( r ) + ( n 1 ) x ( 0 ) = F ( 1 ) , q Δ ( p ) P n n ( p ) x ( 0 ) + q Δ ( p ) r = 1 n 1 P n r ( p ) x ( r ) + n x ( 0 ) = F n ( n ) , q Δ ( p ) P 0 ( p ) r = 1 n x ( r ) = F n ( n ) . (26)

We denote by Dn(p, q) the determinant of the equation (26). Then, Dn(p, q) is the determinant of the matrix

D = 1 + q Δ ( p ) P 1 n ( p ) q Δ ( p ) P 11 ( p ) 1 q Δ ( p ) P 12 ( p ) q Δ ( p ) P 1 , n 1 ( p ) 1 1 + q Δ ( p ) P 2 n ( p ) q Δ ( p ) P 11 ( p ) q Δ ( p ) P 22 ( p ) 1 q Δ ( p ) P 2 , n 1 ( p ) 2 1 + q Δ ( p ) P n 1 , n ( p ) q Δ ( p ) P n 1 , 1 ( p ) q Δ ( p ) P n 1 , 2 ( p ) q Δ ( p ) P n 1 , n 1 ( p ) 1 n 1 q Δ ( p ) P n n ( p ) q Δ ( p ) P n , 1 ( p ) q Δ ( p ) P n , 2 ( p ) q Δ ( p ) P n , n 1 ( p ) n q Δ ( p ) P 0 ( p ) q Δ ( p ) P 0 ( p ) q Δ ( p ) P 0 ( p ) q Δ ( p ) P 0 ( p ) 0

Summarizing these results, we get the following main result of the paper.

## Theorem 4.1

Let pn − (−1)n ≠ 0 and f be n-periodic continuous function.

1. If Dn(p, q) ≠ 0, then equation (2) has a unique n-periodic solution having the form (21), where (x(0), …, x(n − 1), x′(0)) is the unique solution of (26).

2. If Dn(p, q) = 0 and Fn(1) = … = Fn(n) = F n (n) = 0, then equation (2) has infinite number of n-periodic solutions having the form

x ( t ) = α ( x ( 0 ) + x ( 0 ) t + q Δ ( p ) Q ~ n ( t ) ) + F n ( t ) ,

where (x(0), …, x(n − 1), x′(0)) is an eigenvector of D corresponding to 0, α is any number.

3. if Dn(p, q) = 0 and rank[D] < rank[D, FT], F = (Fn(1), …, Fn(n), F n (n)), then equation (2) does not have any n-periodic solution.

## Example 5

Let p = 2i, q = 1 and f(t) = cos ( π t ) + i sin ( π 2 t ) . It can be shown that the unique 4-periodic solution x of equation (2) has the form

x ( t ) = 1 π 2 ( 19 2550 + 4 i 1275 ) ( ( 80 80 i ) + ( 131 188 i ) t ( 147 20 i ) t 2 + + ( 304 128 i ) cos ( π 2 t ) + ( 42 + 36 i ) cos ( π t ) ( 64 + 152 i ) sin ( π 2 t ) ) for t [ 0 , 1 ) , x ( t ) = 1 π 2 ( 19 2550 + 4 i 1275 ) ( ( 246 + 216 i ) ( 201 + 460 i ) t + ( 19 + 156 i ) t 2 + ( 304 128 i ) cos ( π 2 t ) + ( 42 + 36 i ) cos ( π t ) ( 64 + 152 i ) sin ( π 2 t ) ) for t [ 1 , 2 ) , x ( t ) = 1 π 2 ( 19 2550 + 4 i 1275 ) ( ( 734 392 i ) ( 689 148 i ) t + ( 141 + 4 i ) t 2 + ( 304 128 i ) cos ( π 2 t ) + ( 42 + 36 i ) cos ( π t ) ( 64 + 152 i ) sin ( π 2 t ) ) for t [ 2 , 3 ) , x ( t ) = 1 π 2 ( 19 2550 + 4 i 1275 ) ( ( 652 + 2048 i ) ( 235 + 1252 i ) t + ( 13 + 180 i ) t 2 ( 304 128 i ) cos ( π 2 t ) ( 42 + 36 i ) cos ( π t ) + ( 64 + 152 i ) sin ( π 2 t ) ) for t [ 3 , 4 ] .

# Acknowledgements

We thank the referee for valuable comments.

This work was supported by the Malaysian Ministry of Education through the Research Management Center (RMC), Universiti Teknologi Malaysia (FRGS, Votes: 5F235 and 5F198).

### References

[1] Jack K. Hale and Sjoerd M. Verduyn Lunel, Introduction to Functional Differential Equations, Springer, New York, 1993. Search in Google Scholar

[2] Arlington M. Fink, Almost Periodic Differential Equation, Springer, Berlin, 1974. Search in Google Scholar

[3] T. Yoshizawa, Stability Theory and the Existence of Periodic Solutions and Almost Periodic Solutions, Springer, New York, 1975. Search in Google Scholar

[4] Yoshiyuki Hino, Toshiki Naito, Nguyen VanMinh, and Jong Son Shin, Almost Periodic Solutions of Differential Equations in Banach Spaces, Stability and Control: Theory, Method and Applications, vol. 15, Taylor & Francis, London, 2002. Search in Google Scholar

[5] George Seifert, Second-order neutral delay-differential equations with piecewise constant time dependence, J. Math. Anal. Appl. 281 (2003), no. 1, 1–9. Search in Google Scholar

[6] Rong Yuan, Pseudo-almost periodic solutions of second-order neutral delay differential equations with piecewise constant argument, Nonlinear Anal. 41 (2000), 871–890. Search in Google Scholar

[7] Kenneth L. Cooke and Joseph Wiener, A survey of differential equations with piecewise continuous arguments, Delay Differential Equations and Dynamical Systems, Lecture Notes in Math., vol. 1475, Springer, Berlin, Heidelberg, 1991. Search in Google Scholar

[8] Joseph Wiener, Generalized Solutions of Functional Differential Equations, World Scientific, Singapore, 1993. Search in Google Scholar

[9] Mukhiddin I. Muminov, On the method of finding periodic solutions of second-order neutral differential equations with piecewise constant arguments, Adv. Difference Equ. (2017), 2017:336, 10.1186/s13662-017-1396-7. Search in Google Scholar

[10] Rong-Kun Zhuang, Existence of almost periodic solutions to n-th order neutral differential equation with piecewise constant arguments, Abstr. Appl. Anal. (2012), Article ID 186361, 10.1155/2012/186361. Search in Google Scholar

[11] Rong-Kun Zhuang and Rong Yuan, Weighted pseudo almost periodic solutions of N-th order neutral differential equations with piecewise constant arguments, Acta Math. Sin. (Engl. Ser.) 30 (2014), no. 7, 1259–1272. Search in Google Scholar

[12] Hong-Xu Li, Almost periodic solutions of second-order neutral delay-differential equations with piecewise constant argument, J. Math. Anal. Appl. 298 (2004), 693–709. Search in Google Scholar

[13] Elhadi Ait Dads and Lahcen Lhachimi, New approach for the existence of pseudo almost periodic solutions for some second-order differential equation with piecewise constant argument, Nonlinear Anal. 64 (2006), no. 6, 1307–1324. Search in Google Scholar