Existence of periodic solutions for a mechanical system with piecewise constant forces

Transcription

1 Hacettepe Journal of Mathematics and Statistics Volume , Existence of periodic solutions for a mechanical system with piecewise constant forces Duygu Aru aslan and Nur Cengiz Abstract In this study, we consider spring-mass systems subjected to piecewise constant forces. We investigate sucient conditions for the existence of periodic solutions of homogeneous and nonhomogeneous damped spring-mass systems with the help of the Floquet theory. In addition to determining conditions for the existence of periodic solutions, stability analysis is performed for the solutions of the homogeneous system. The Floquet multipliers are taen into account for the stability analysis [3. The results are stated in terms of the parameters of the systems. These results are illustrated and supported by simulations for dierent values of the parameters. Keywords: Generalized piecewise constant forces, spring-mass system, stability and periodic solutions AMS Classication: 34A36, 34K13, 34K20, 37N05 Received : Accepted : Doi : /HJMS Introduction In the literature, many mathematical models have been suggested to examine real world processes. In some cases, state of the process in previous time inuences the present and future states of the process signicantly. Thus, earlier suggested models have been improved by taing time delays into consideration. These models including some of the past states of the processes are expressed by delay dierential equations. These equations are very important in various problems of economics, ecology, biology, engineering, medicine and physics [17, 20. Department of Mathematics, Süleyman Demirel University, 32260, Isparta, Turey, duyguarugaslan@sdu.edu.tr Corresponding Author. Graduate School of Natural and Applied Sciences, Süleyman Demirel University, 32260, Isparta, Turey, nurcengiz90@gmail.com

2 Delay dierential equations are generally given in the following form x t = ft, xt, xt τ, where τ is a positive delay value, t R, x R n. It is clear that evolution of the system 1.1 depends both on the present and the previous time. If the greatest integer function [t instead of t τ is taen as deviating argument in equation 1.1, then dierential equations with piecewise constant argument of the form 1.2 x t = ft, xt, x[t, are obtained. Dierential equations with piecewise constant argument can be considered in the class of delay dierential equations since the rate of change of the system is determined by both the present value and the previously memorized values of the state variable x. This class of dierential equations was described by Cooe and Wiener at the beginning of the 1980s [15. In recent years, several studies about their theory such as existence and uniquness of solutions [15, 29, oscillations [1, 16, 26, 30, 31, existence of periodic and almost periodic solutions [1, 24, 27, 32, integral manifolds [23 and applications in the elds of biology, control theory, neural networs and biomedical models of disease [9, 10, 11, 12, 19, 21, 22, 28, 33, stability [6, 7 are available. This type of equations has been usually examined by reducing them into discrete equations and applying numerical methods [11, 12, 15, 16, 17, 18, 22, 25, 29. In 1.2, the distance between the switching moments is xed as 1 unit. Taing any piecewise constant function instead of the greatest integer function, dierential equations with piecewise constant argument were generalized by Ahmet in [4. Later, dierential equations with piecewise constant argument of generalized type have been studied by many authors [2, 5, 6, 8, 11, 13, 14. A classical homogeneous damped spring-mass system is given by mẍ + cẋ + x = 0, where xt is the displacement of the mass, m > 0 is the mass, c > 0 is the damping coecient and > 0 is the spring constant. This system exhibits damped harmonic motion. In other words, if friction force acts on an oscillating system, then the oscillation amplitude approaches to zero. According to the sign of the discriminant of the characteristic equation = c2 m 2 4 m r 2 + c m r + m = 0, the behavior of the system is determined. If > 0, = 0 and < 0, then the system exhibits motion with over damped, critical damped and under damped, respectively. The motions with over damped and critical damped are not oscillating, but the motion with under damped is oscillating. If this system has an external force F, it is described as follows mẍ + cẋ + x = F. In [16, taing the greatest integer function as deviating argument for the force and using the method of reduction to discrete equations, Dai and Singh studied the oscillatory motion of the following spring-mass system mẍ + cẋ + x = Ax[t, where A denotes the magnitude of the force and the last system has discontinuities at the moments [t because of the piecewise constant force F = Ax[t.

3 523 In this study, we tae external forces with piecewise constant argument of generalized type instead of greatest integer function for the damped spring-mass system. We determine sucient conditions for stability of solutions of the linear homogeneous system using Floquet multipliers. Additionally, we deal with existence and uniqueness of periodic solutions for both homogeneous and nonhomogeneous spring-mass systems. 2. Preliminaries Let Z, N and R be the sets of all integers, natural and real numbers, respectively. Fix two real-valued sequences θ i, ζ i, i Z, such that θ i < θ i+1, θ i ζ i θ i+1 for all i Z, θ i as i, and assume that there exists a number θ > 0 such that θ i+1 θ i θ, i Z. The most general form of dierential equations with piecewise constant argument of generalized type can be expressed as follows [3 x t = ft, xt, xγt, where t R, x R n, γt = ζ i for t [θ i, θ i+1, i Z. It is clear that the greatest integer function [t is a special case of the argument function γt of alternate type: retarded and advanced. In fact, γt = [t for θ i = ζ i = i, i Z. In the present paper, we shall consider the following damped spring-mass systems with piecewise constant argument of generalized type 2.1 and 2.2 mx + cx + x = Axγt, mx + cx + x = Axγt + gt, where x R, t R, gt is a scalar, continuous real valued function, and γt = ζ i, if t [θ i, θ i+1, i Z. The systems 2.1 and 2.2 have discontinuities at the moments θ i, i Z, since the piecewise function γt is not continuous. But the solutions of the systems 2.1 and 2.2 present a continuous and continuously dierentiable dynamic motion within intervals t [θ i, θ i+1, i Z. Let z 1 = x and z 2 = x. Then, spring-mass systems 2.1 and 2.2 can be reduced to the following rst order dierential equations, respectively, 2.3 and z t = A 0zt + A 1zγt, 2.4 z t = A 0zt + A 1zγt + ft. [ [ Here the matrices A 0 = m c and A 1 = A m m 0 depend on the parameters [ 0 of the spring-mass systems 2.1 and 2.2. Moreover, ft =. The systems 2.3 gt and 2.4 are linear homogeneous and nonhomogeneous systems with argument-function γt, respectively. In this study, our purpose is to examine damped spring-mass systems with generalized piecewise constant forces without transforming them into discrete equations and we shall assume that the systems exhibit harmonic motion with under damped, i.e., = c2 m 4 < 0. While doing this examination, we use Floquet theory for dierential equations with piecewise constant argument of generalized type and obtain sucient 2 m conditions required for the spring-mass systems 2.3 and 2.4 to have periodic solutions [3. Furthermore, we determine under which conditions the solutions of the system 2.3

4 524 exhibit stable, asymptotically stable and unstable behaviour with the help of the Floquet multipliers. Later, we give a result about the situation expressing that the system 2.4 has a unique periodic solution corresponding to the initial value. Additionally, we give numerical simulations, which show that the spring-mass system 2.3 has asymptotically stable or unstable solutions and the spring-mass system 2.4 has a periodic solution corresponding to special values chosen for the sequences and the parameters Denition. [3 A continuous function zt is a solution of on R if: i the derivative z t exists at each point t R with the possible exception of the points θ i, i Z, where one-sided derivatives exist; ii the equation is satised for zt on each interval θ i, θ i+1, i Z, and it holds for the right derivative of zt at the points θ i, i Z. 3. The matrix-function of the linear homogeneous system Let us consider the linear homogenous part of 2.3 and 2.4 without piecewise constant argument as follows [ x t = A 0xt = m c xt. m Let I be a 2 2 identity matrix. Thus, the state transition matrix Xt, s, t, s R, such that Xs, s =I, of the system 3.1 is given by Xt, s = e A 0t s = e αt s where α = c 2m, = m c2 4m =. 2 2 Consider the following matrix-function M it = Xt, ζ i + cost s + α sint s 1 sint s sint s m cost s α t ζ i Xt, sa 1sds, i Z, sint s dened in [6. This matrix-function is important for the results obtained below. For our systems 2.3 and 2.4 under consideration, we nd that [ M it = e αt ζ i Ki L i, M i where indices K i, L i, M i, N i are evaluated as K i = 1 A cost ζ i + α sint ζi + Aeαt ζ i, N i L i = 1 sint ζi, M i = 1 A sint ζ i m and N i = cost ζ i α sint ζi.

5 525 We shall need the following assumptions listed below: A1 A 1 e αt ζ i cost ζ i α sint ζi 1, t [θ i, θ i+1, i Z; A2 g is ω-periodic; A3 There exists a number p N such that θ i+p = θ i + ω, ζ i+p = ζ i + ω, for all i Z. A1 implies that the matrix-function M it is invertible. A2 and A3 guarantee that the systems 2.3 and 2.4 are periodic Theorem. [6 For every t 0, z 0 R R 2 there exists a unique solution zt = zt, t 0, z 0 of 2.3 in the sense of Denition 2.1. such that zt 0 = z 0 if and only if condition A1 is valid Theorem. [6 A number t 0 R, θ i t 0 < θ i+1, is xed. For every z 0 R 2 there exists a unique solution zt = zt, t 0, z 0 of 2.3 in the sense of Denition 2.1. such that zt 0 = z 0 if and only if det[m it 0 0 and det[m jt 0 for t = θ j, θ j+1, j Z. 4. The fundamental matrix of the homogeneous spring-mass system 2.3 Let us x t 0 R. Then we can nd an i Z such that θ i t 0 < θ i+1. Then there exists a fundamental matrix Zt = Zt, t 0 satisfying Zt 0 = Zt 0, t 0 =I of the springmass system 2.3 [3, 5. Without loss of generality assume that θ i < t 0 < ζ i, i Z. For interval t [t 0, θ i+1 which includes the initial point t 0, the fundamental matrix is Zt, t 0 = Zt = M itm 1 i t 0. Moreover, the fundamental matrix Zt = Zt, t 0 of 2.3 is constructed in two different ways for increasing and decreasing t as explained below [3, 5. If θ i t 0 < θ i+1, t [θ l, θ l+1, l > i, then Zt = M l t [ i+1 =l M 1 θ M 1 θ M 1 i t 0. If θ i t 0 θ i+1, t [θ j, θ j+1, j < i, then Zt = M jt [ i 1 =j M 1 θ +1 M +1 θ +1 M 1 i t 0. It can be shown that Zt, s = ZtZ 1 s, t, s R, and a solution zt, zt 0 = z 0, t 0, z 0 R R 2, of 2.3 is equal to zt = Zt, t 0z 0, t R. It's dicult to give a direct formula for the fundamental matrix of the spring-mass system 2.3. For illustration, we evaluate this matrix for the intervals t [θ i, θ i+1, t [θ i+1, θ i+2 where t is increasing and for the interval t [θ i 1, θ i where t is decreasing. First, consider the interval [θ i, θ i+1 which includes the initial value t 0, θ i ζ i θ i+1. Then, for t [t 0, θ i+1 fundamental matrix of 2.3 has the form 4.1 Zt = M itm 1 i t 0 = e αt t 0 s i1 [ Ki1 L i1 M i1 N i1 and its indices are as follows K i1 = 1 A cost t 0 + α sint t0 + Aeαt ζ i cost 0 ζ i α sint0 ζi,,

6 526 L i1 = A sint t 0 Ae αζ i e αt 0 sint ζ i e αt sint 0 ζ i, M i1 = 1 A sint t 0, m N i1 = 1 A cost t 0 α sint t0 + Aeαt 0 ζ i cost ζ i α sint ζi, s i1 = 1 A 1 e αt 0 ζ i cost 0 ζ i α sint0 ζi. Second, for t [θ i 1, θ i, θ i 1 ζ i 1 θ i, i Z, we nd the fundamental matrix of 2.3 as follows [ 4.2 Zt = M i 1tMi 1θ im iθ im 1 i t 0 = e αt t 0 Ki2 L i2, s i1 s i2 M i2 N i2 where K i2 = 1 A 2 cost t 0 + α sint t0 + A 1 A e αt ζi 1 cosζ i 1 t 0 + α sinζi 1 t0 + e αθ i ζ i cost t 0 θ i + ζ i + α sint t0 θi + ζi sint θ i sint m 2 0 ζ i + A 1 A e αθ i ζ i 1 sint ζ i 1 sint m 2 0 θ i + A2 2 eαt ζ i+θ i ζ i 1 cost 0 + θ i ζ i ζ i 1 α sint0 + θi ζi ζi 1 + m 2 sinθi ζi 1 sint0 ζi, L i2 = 1 1 A 2 sint t 0 + A 1 A e αt ζi 1 sinζ i 1 t 0 + e αt 0 ζ i sint ζ i e αθ i ζ i sint 0 ζ icost θ i + α sint θi + e αθ i ζ i 1 sint ζ i 1cost 0 θ i + α sint0 θi + A2 e αt ζ i 1+θ i ζ i sint 2 0 ζ icosθ i ζ i 1 α sinθi ζi 1 + e αt ζ i 1+t 0 ζ i sinζ i 1 ζ i + e αt 0 ζ i 1 +θ i ζ i sint ζ i 1cosθ i ζ i α sinθi ζi,

7 M i2 = m + A m 1 A 2 sint t 0 1 A e αθ i ζ i sint θ icost 0 ζ i α sint0 ζi e αθ i ζ i 1 sinθ i ζ icost ζ i 1 + t 0 ζ i α sint ζi 1 + t0 ζi + e αθ i ζ i 1 sint 0 ζ icost ζ i 1 + θ i ζ i α sint ζi 1 + θi ζi, N i2 = 1 A 2 cost t 0 α sint t0 + A 1 A e αt 0 ζ i cost ζ i α sint ζi + e αθ i ζ i sint 0 ζ i sint θ i m 2 + e αθ i ζ i 1 cost t 0 + θ i ζ i 1 α sint t0 + θi ζi 1 e αθ i ζ i 1 sint ζ i 1 cost 0 θ i m 2 + A2 2 eαθ i ζ i 1 +t 0 ζ i cost ζ i 1 + θ i ζ i α sint ζi 1 + θi ζi + sint ζ i 1 sinθ i ζ i, m 2 s i1 = 1 A s i2 = 1 A 1 e αt 0 ζ i cost 0 ζ i α sint0 ζi, 1 e αθ i ζ i 1 cosθ i ζ i 1 α sinθi ζi 1. Finally, for t [θ i+1, θ i+2, θ i+1 ζ i+1 θ i+2, i Z, where t is increasing, fundamental matrix of 2.3 is [ 4.3 Zt = M i+1tmi+1θ 1 i+1m iθ i+1m 1 i t 0 = e αt t 0 Ki3 L i3, s i1 s i3 M i3 N i3 where the indices are given by K i3 = 1 A 2 cost t 0 + α sint t0 + A 1 A e αt ζi+1 cost 0 ζ i+1 α sint0 ζi+1 + e αθ i+1 ζ i cost t 0 θ i+1 + ζ i + α sint t0 θi+1 + ζi m sint θi+1 2 sint0 ζi + A 1 A e αθ i+1 ζ i+1 sint ζ i+1 sint m 2 0 θ i+1 + A2 2 eαt ζ i+θ i+1 ζ i+1 cost 0 + θ i+1 ζ i ζ i+1 α sint0 + θi+1 ζi ζi+1 + sinθi+1 ζi+1 sint0 ζi, m 2 527

8 528 L i3 = 1 1 A 2 sint t 0 + A 1 A e αt ζi+1 sint 0 ζ i+1 + e αθ i+1 ζ i+1 sint ζ i+1cost 0 θ i+1 + α sint0 θi+1 e αθ i+1 ζ i sint 0 ζ icost θ i+1 + α sint θi+1 + e αt 0 ζ i sint ζ i + A2 e αt 0 ζ i+1 +θ i+1 ζ i sint ζ i+1cosθ 2 i+1 ζ i α sinθi+1 ζi e αt ζ i+1+t 0 ζ i sinζ i ζ i+1 e αt ζ i+1+θ i+1 ζ i sint 0 ζ icosθ i+1 ζ i+1 α sinθi+1 ζi+1, M i3 = m + A m 1 A sint t 0 1 A e αθ i+1 ζ i sint θ i+1cost 0 ζ i α sint0 ζi + e αθ i+1 ζ i+1 sint 0 θ i+1cost ζ i+1 α sint ζi+1, N i3 = 1 A 2 cost t 0 α sint t0 + A 1 A e αt 0 ζ i cost ζ i α sint ζi + e αθ i+1 ζ i sint 0 ζ i sint θ i+1 m 2 + e αθ i+1 ζ i+1 cost t 0 + θ i+1 ζ i+1 α sint t0 + θi+1 ζi+1 sint ζ i+1 sint 0 θ i+1 m 2 + A2 2 eαθ i+1 ζ i+1 +t 0 ζ i cost ζ i+1 + θ i+1 ζ i α sint ζi+1 + θi+1 ζi + sint ζ i+1 sinθ i+1 ζ i, m 2 s i1 = 1 A and s i3 = 1 A 1 e αt 0 ζ i cost 0 ζ i α sint0 ζi 1 e αθ i+1 ζ i+1 cosθ i+1 ζ i+1 α sinθi+1 ζi+1. Clearly, it is not easy to obtain a complete form for the fundamental matrix. So here, we have created the fundamental matrix of the system 2.3 in three intervals for illustration. The following theorem about existence and uniqueness of solutions for the spring-mass system 2.4 can be given.

9 Theorem. [3 Assume that the condition A1 is fullled. Then for every t 0, z 0 R R 2 there exists a unique solution zt = zt, t 0, z 0 of 2.4 in the sense of Denition 2.1. such that zt 0 = z 0, and 5. Main results zt = Zt, t 0z 0 + t t 0 Zt, sfsds. Let t 0 = 0 and denote by Zω = Zω, t 0 = Zω, 0 the matrix of monodromy. The eigenvalues ρ of the monodromy matrix Zω are called Floquet multipliers or simply multipliers of system 2.3. The eigenvalues of the matrix P = 1 ln Zω, λ, are called ω the Floquet exponents or simply exponents [3. In this section, for the ω periodic spring-mass systems 2.3 and 2.4, we shall state the results depending on the multipliers ρ 1 and ρ 2. ω can be located in one of the intervals [θ i, θ i+1 which includes t 0, [θ l, θ l+1, l > i where t is increasing or [θ j, θ j+1, j < i where t is decreasing. The monodromy matrix can be found using the constructions presented in the Section 4. Thus, the periodicity of the solutions for the systems 2.3, 2.4 and stability of the solutions for the system 2.3 can be investigated using the multipliers. These results about periodicity and stability are stated by the following theorems [ Theorem. If ρ is a multiplier then there exists a nontrivial solution zt of 2.3 such that zt + ω = ρzt. Conversely, if there exists a nontrivial solution zt of 2.3 such that zt + ω = ρzt, then ρ is a multiplier Theorem. System 2.3 has a σω periodic solution if and only if there exists a multiplier ρ such that ρ σ = 1, σ Z. We can say that the spring-mass system 2.4 has a unique ω-periodic solution under the assumptions given in the next theorem Theorem. If unity is not one of the multipliers, then 2.4 has a unique ω periodic solution zt such that 5.1 ω z0 = z 0 = [I Zω 1 Zω, sfsds. Moreover, for the solutions of 2.3, it is possible to analyze stability by Floquet multipliers as given below Theorem. The solutions of 2.3 are i stable if ρ j 1 for j = 1, 2 and ρ j is simple when ρ j = 1; ii asymptotically stable if and only if ρ j < 1, j = 1, 2; iii unstable if there exists a multiplier for some j = 1, 2 such that ρ j > 1. If, for example, ω belongs to one of the intervals [θ i, θ i+1, [θ i 1, θ i or [θ i+1, θ i+2, i Z, then the monodromy matrix and thus multipliers can be computed by using the equations 4.1, 4.2 and 4.3, respectively. It is possible to nd the multipliers for other cases similarly. 6. The numerical simulations In this section, we give examples with simulations to see more concrete results. Tae the initial value t 0 = 0 for all of these examples. 0

10 Example. First, let the parameters be m = c = A = 1, = 40 and the sequences be θ i = ζ i = 1 i. In other words, consider the damped spring mass system x + x + 40x = xγt. Considering z 1 = x and z 2 = x, spring-mass system 6.1 can be reduced to the following rst order dierential equation 6.2 [ z t = [ 0 0 zt zγt. Since the conditions A1 and A3 are satised, the equation 6.1 and so the system 6.2 are 0.1 periodic. t = ω = 0.1 [θ 0, θ 1 = [0, 0.1. Then, for t [0, 0.1, the monodromy matrix is in the following form Z0.1 = [ Thus, we nd the multipliers ρ 1 = i and ρ 2 = i. So, ρ 1 = ρ 2 = < 1. According to Theorem 5.1., there exist nontrivial solutions zt+0.1 = ρ 1zt and zt+0.1 = ρ 2zt of the system 6.2. It is seen that the solutions of 6.2 are asymptotically stable, in particular, the xed point 0, 0 of the system 6.2 is asymptotically stable. The simulations showing asymptotic stability of the trivial solution of 6.2 are given in Figure 1.

11 531 Figure 1. Simulations showing the asymptotic stability of the trivial solution of 6.2 for t [0, 18 with m = c = A = 1, = 40, θ i = ζ i = 1 i, i Z starting at z10, z20 = 1, 0 10 If we continue for more intervals, we obtain the simulations given in Figure Example. Second, let m = 1, c = , = , A = 1 and the sequences be θ i = ζ i = 1 i. So, consider the spring mass system x x x = xγt. Considering z 1 = x and z 2 = x, spring-mass system 6.3 can be reduced to the following rst order dierential equation [ [ z t = zt + zγt. 1 0 It can be shown that the conditions A1 and A3 are satised. Hence, the equation 6.3 and so the system 6.4 are 0.1 periodic. t = ω = 0.1 [θ 0, θ 1 = [0, 0.1. Then,

12 532 Figure 2. Simulations showing the asymptotic stability of the trivial solution of 6.2 for t [0, 80 with m = c = A = 1, = 40, θ i = ζ i = 1 i, i Z starting at z10, z20 = 1, 0 10 for t [0, 0.1, the monodromy matrix is Z0.1 = [ Thus, the multipliers are ρ 1 = i and ρ 2 = i. So, ρ 1 = ρ 2 = < 1. Then Theorem 5.1. implies that there exist nontrivial solutions zt = ρ 1zt and zt = ρ 2zt of the system 6.4. It is seen that the solutions of 6.4 are asymptotically stable, in particular, the xed point 0, 0 of the system 6.4 is asymptotically stable. The simulations showing asymptotic stability of the trivial solution of 6.4 are given in Figure 3.

13 533 Figure 3. Simulations showing the asymptotic stability of the trivial solution of 6.4 for t [0, 20 with m = 1, c = , = , A = 1, θ i = ζ i = 1 i, i Z starting at 10 z 10, z 20 = , If we continue for more intervals, we see that the solution z 1t is asymptotically stable with the simulation in Figure 4. Similarly, for more intervals, it can be seen that the other solution is also asymptotically stable.

14 534 Figure 4. Simulation showing the asymptotic stability of the trivial solution of 6.4 for t [0, 350 with m = 1, c = , = , A = 1, θ i = ζ i = 1 i, i Z starting at 10 z 10, z 20 = , Example. For = 10, m = 1, c = 0.01, A = and the sequences θ i = ζ i = 1 i, consider the linear nonhomogeneous spring-mass system with piecewise 10 constant argument of generalized type 6.5 x x + 10x = 10 8 xγt + 4 cos20πt. Taing z 1 = x and z 2 = x, spring-mass system 6.5 can be reduced to the following nonhomogeneous dierential equation 6.6 [ z t = [ zt [ zγt + 0 4cos20πt. Since the conditions A1 and A3 are satised, the equation 6.5 and so the system 6.6 are 0.1 periodic. t = ω = 0.1 [θ 0, θ 1 = [0, 0.1. Then, for t [0, 0.1, the monodromy matrix is in the form Z0.1 = [ From monodromy matrix, the multipliers are found as ρ 1 = i and ρ 2 = i, so, ρ 1 = ρ 2 = < 1. Since unity is not one of the multipliers, the spring-mass system 6.6 has a unique 0.1 periodic solution with the initial value z 10, z 20 = , evaluated by using the equation 5.1. The simulations showing 0.1 periodic solution of 6.6 are given in Figure 5.

15 535 Figure 5. Simulations showing the 0.1 periodic solutions of 6.6 for t [0, 10 with = 10, m = 1, c = 0.01, A = , θ i = ζ i = 1 i, i Z starting at z10, z20 = , Besides, the simulation according to the solutions z 1t and z 2t in Figure 6 shows the existence of periodic solution.

16 536 Figure 6. Simulation according to the solutions z 1t and z 2t showing the 0.1 periodic solutions of 6.6 for t [0, 10 with = 10, m = 1, c = 0.01, A = , θ i = ζ i = 1 i, i Z starting at 10 z 10, z 20 = , Example. Let m = 1, c = 1, = 100, A = 101 and the sequences θ i = ζ i = 1 10 i. In other words, consider the spring mass system 6.7 x + x + 100x = 101xγt Taing z 1 = x and z 2 = x, spring-mass system 6.7 can be reduced to the following form 6.8 [ z t = [ 0 0 zt zγt. Since the conditions A1 and A3 are satised, the equation 6.7 and so the system 6.8 are 0.1 periodic. t = ω = 0.1 [θ 0, θ 1 = [0, 0.1. Then, for t [0, 0.1, the monodromy matrix is in the form Z0.1 = [ Thus, we nd the multipliers ρ 1 = and ρ 2 = So, according to Theorem 5.1., there exists nontrivial solutions zt+0.1 = ρ 1zt and zt+0.1 = ρ 2zt of the system 6.8. It is seen that the spring-mass system 6.8 has unstable solutions, in particular, the trivial solution of the system 6.8 is unstable. The simulations showing instability of the trivial solution of 6.8 are given in Figure 7.

17 537 Figure 7. Simulation showing the unstable solutions of 6.8 for t [0, 10 with m = 1, c = 1, = 100, A = 101, θ i = ζ i = 0.1i, i Z starting at z 10, z 20 = 2, 2 References [1 Aftabizadeh, A.R., Wiener, J. and Xu, J.-M. Oscillatory and periodic solutions of delay dierential equations with piecewise constant argument, Proc. Amer. Math. Soc. 99, , [2 Ahmet, M.U. Integral manifolds of dierential equations with piecewise constant argument of generalized type, Nonlinear Anal. 66, , [3 Ahmet, M.U. Nonlinear Hybrid Continuous/Discrete Time Models, Amsterdam, Paris, Atlantis Press, [4 Ahmet, M.U. On the integral manifolds of the dierential equations with piecewise constant argument of generalized type, Proceedings of the Conference on Dierential and Dierence Equations at the Florida Institute of Technology, August 1-5, 2005, Melbourne, Florida, Editors: R.P. Agarval and K. Perera, Hindawi Publishing Corporation, 1120, [5 Ahmet, M.U. Quasilinear retarded dierential equations with functional dependence on piecewise constant argument, Communications On Pure And Applied Analysis 13 2, , [6 Ahmet, M.U. Stability of dierential equations with piecewise constant arguments of generalized type, Nonlinear Anal. 68, , [7 Ahmet, M.U. and Aru aslan, D. Lyapunov-Razumihin method for dierential equations with piecewise constant argument, Discrete and Continuous Dynamical Systems, Series A 25 2, , [8 Ahmet, M.U., Aru aslan, D. and Liu, X. Permanence of nonautonomous ratio-dependent predator-prey systems with piecewise constant argument of generalized type, Dynamics of Continuous Discrete and Impulsive Systems Series A. Mathematical Analysis 15 1, 3751, [9 Ahmet, M.U., Aru aslan, D. and Ylmaz, E. Stability analysis of recurrent neural networs with piecewise constant argument of generalized type, Neural Networs 23, , [10 Ahmet, M.U., Aru aslan, D. and Ylmaz, E. Stability in cellular neural networs with a piecewise constant argument, Journal of Computational and Applied Mathematics 233, , 2010.

18 538 [11 Ahmet, M.U., Ötem, H., Picl, S.W. and Weber, G.-W. An anticipatory extension of Malthusian model, CASYS'05-Seventh International Conference, AIP Conference Proceedings 839, , [12 Busenberg, S. and Cooe, K.L. Models of vertically transmitted diseases with sequentialcontinuous dynamics, Nonlinear Phenomena in Mathematical Sciences, Academic Press, New Yor, , [13 Chiu, K.-S., Pinto, M. Periodic solutions of dierential equations with a general piecewise constant argument and applications, Electron. J. Qual. Theory Dier. Equ. 46, 19 pp, [14 Chiu, K.-S., Pinto, M. Variation of parameters formula and Gronwall inequality for dierential equations with a general piecewise constant argument, Acta Math. Appl. Sin. Engl. Ser. 27 4, , [15 Cooe, K.L. and Wiener, J. Retarded dierential equations with piecewise constant delays, J. Math. Anal. Appl. 99, , [16 Dai, L. and Singh, M.C. On oscillatory motion of spring-mass systems subjected to piecewise constant forces, Journal of Sound and Vibration 173 2, , [17 Gopalsamy, K. Stability and Oscillation in Delay Dierential Equations of Population Dynamics, Kluwer Academic Publishers, Dordrecht, [18 Györi, I. On approximation of the solutions of delay dierential equations by using piecewise constant argument, Int. J., Math. Math. Sci., 14, , [19 Gopalsamy, K. and Liu, P. Persistence and global stability in a population model, J. Math. Anal. Appl. 224, 5980, [20 Gyori, I. and Ladas, G. Oscillation Theory of Delay Dierential Equations with Applications, Oxford University Press, New Yor, [21 Matsunaga, H., Hara, T. and Saata, S. Global attractivity for a logistic equation with piecewise constant argument, Nonlinear Dierential Equations Appl. 8, 4552, [22 Muroya, Y. Persistence, contractivity and global stability in logistic equations with piecewise constant delays, J. Math. Anal. Appl. 270, , [23 Papaschinopoulos, G. On the integral manifold for a system of dierential equations with piecewise constant argument, J. Math. Anal. Appl. 201, 7590, [24 Seifert, G. Almost periodic solutions of certain dierential equations with piecewise constant delays and almost periodic time dependence, J. Dierential Equations 164, , [25 Shah, S.M. and Wiener, J. Advanced dierential equations with piecewise constant argument deviations, Int. J. Math. Math. Sci. 6, , [26 Shen, J.H. and Stavroulais, I.P. Oscillatory and nonoscillatory delay equation with piecewise constant argument, J. Math. Anal. Appl. 248, , [27 Wang, G. Periodic solutions of a neutral dierential equation with piecewise constant arguments, J. Math. Anal. Appl. 326, , [28 Wang, Z. and Wu, J. The stability in a logistic equation with piecewise constant arguments, Dierential Equations Dynam. Systems 14, , [29 Wiener, J. Generalized Solutions of Functional Dierential Equations, World Scientic, Singapore, [30 Wiener, J. and Cooe, K.L. Oscillations in systems of dierential equations with piecewise constant argument, J. Math. Anal. Appl. 137, , [31 Wiener, J. and Lashmiantham, V. A damped oscillator with piecewise constant time delay, Nonlinear Stud. 7, 7884, [32 Xia, Y., Huang, Z. and Han, M. Existence of almost periodic solutions for forced perturbed systems with piecewise constant argument, J. Math. Anal. Appl. 333, , [33 Yang, X. Existence and exponential stability of almost periodic solution for cellular neural networs with piecewise constant argument, Acta Math. Appl. Sin. 29, , 2006.