Columbia International Publishing Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 doi:10.7726/jams.2016.1012 Research Article (Invited) Necessary and Sufficient Condition for Existence of a Unique Invariant Torus of Singular Impulsive System Yurii Korol1* Received: 13 August 2016; Accepted: 1 October 2016; © The author(s) 2016. Published with open access at www.uscip.us Abstract Under the assumption that linear nonhomogeneous singular system of differential equation with impulse action defined on the direct product of torus and Euclidean space can be reduced to the central canonical form and the corresponding homogeneous system is exponential dichotomous on the semiaxes, we obtained necessary and sufficient condition for existence of a unique invariant torus of singular linear system. Keywords: Exponential Dichotomy; Invariant Torus; Bounded Solution; Central Canonical Form; Impulsive System 1. Introduction Some problems in control theory, radio physics, mathematical economics, and linear programming are known to be modeled by systems of differential equations with a singular matrix multiplying the derivative ππ₯ π΅(π‘) = π΄(π‘)π₯ + π(π‘). ππ‘ Such systems are usually referred to as singular systems of ordinary differential equations (see Boyarincev et al., 1989; Samoilenko et al., 2000). The notion of central canonical form introduced in Campbell and Petzold(1983) plays an important role in the theory of such systems. In this paper we consider the problem of existence a unique invariant torus of singular impulsive system. Similar problem for a system of equations defined on a direct product of an π-dimentional torus π π and the Euclidean space were investigated in Samoilenko(1987). Using these results let us find the necessary and sufficient condition for existence of a unique invariant torus under the assumption that singular system of differential equation can be reduced to the central canonical form. Let us consider system of impulse differential equation in central canonical form ______________________________________________________________________________________________________________________________ * Corresponding e-mail: [email protected] 1 Uzhhorod National University, Uzhhorod, Ukraine 149 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 ππ π(π) 0 πΈ 0 ππ₯ ] ] π₯ + π(π), = π(π), [ πβπ =[ (1) 0 πΈπ 0 πΌ ππ‘ ππ‘ Ξπ₯1 |πβΞ = π΅(π)π₯1 + πΌ(π), (2) π where π₯ β β , π₯ = πππ(π₯1 , π₯2 ), where π₯1 (π‘, π), π₯2 (π‘, π) are (π β π ) and π β dimentional vector functions respectively, π β π π , π π is π -dimentional torus, π(π), πΌ(π)are continuous(piecewise continuous with first kind discontinuities in the set Ξ) 2π-periodic with respect to each component ππ£ , π£ = 1, . . . , π and bounded for all π β π π vector functions, π2 (π) β βπ β1 (π π ); π(π), π΅(π) are continuous 2π-periodic with respect to each component ππ£ square matrices, function π(π) is bounded and satisfies Lipschitz condition for π β π π , πΈπβπ , πΈπ are identity matrix of order (π β π ) and π respectively, πΌ is quasidiagonal matrix, πππ‘(πΈπβπ + π΅(π)) β 0 for all π β π π . We assume that the set Ξ is a subset of the torus π π which is a manifold of dimension π β 1 defined by the equation Ξ¦(π) = 0 for some continuous scalar 2π-periodic with respect to each component ππ£ , π£ = 1, . . . , π, function. Denote by π‘π (π) the solutions of equation Ξ¦(ππ‘ (π)) = 0 that are moments of impulsive action in system (1),(2). Assume also that exist constant π > 0 such that π‘π (π) β π‘πβ1 (π) β₯ π, (3) for all π β β€, π β π π . System (1),(2) can be split on two independent systems ππ ππ₯1 = π(π), = π(π)π₯1 + π1 (π), Ξπ₯1 |πβΞ = π΅(π)π₯1 + πΌ(π), (4) ππ‘ ππ‘ ππ ππ₯2 = π(π), πΌ = π₯2 + π2 (π). (5) ππ‘ ππ‘ The general solution of the system (4) has the form π‘ π₯1 (π‘, π) = π(π‘, π)π + β« βπ(π‘, π)π β1 (π , π)π1 (ππ (π))ππ + 0 + βπ(π‘, π)π β1 (π‘π (π), π)πΌ (ππ‘π (π) (π)), β (6) 0<π‘π (π)<π‘ where π(π‘, π) is fundamental matrix of homogeneous impulsive system ππ₯1 = π(ππ‘ (π))π₯1 , ππ‘ Ξπ₯1 |πβΞ = π΅(ππ‘π (π) (π))π₯1 . (7) Solution of the system (5) we can write in the form (see Boyarincev et al., 1989) π β1 βπ π2 (ππ‘ (π)) π π π 1 (π). . . πππ (π). ππ 1 π1 βπ . . . βπ π 1 π=0 π1 +...+ππ =π By combining (6) and (8), we obtain the general solution of (1),(2) in form π₯2 (π‘, π) = π(ππ‘ (π)) = β β β β βπΌ π (8) π‘ π₯(π‘, π) = π(π‘, π)π + β« π(π‘, π , π)π1 (ππ (π))ππ + 0 β π(π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) , 0<π‘π (π)<π‘ π(ππ‘ (π)) ( β1 where π(π‘, π , π) = π(π‘, π)π (π , π) and π is (π β π )-dimentional constant vector. (9) ) 150 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 2. Bounded Solution on Whole Axis Let us find the conditions of existance bounded on whole axis solutions of system (1),(2) in assumption that linear homogeneous system (7) is exponential dichotomous in semiaxes π + and π β with projectors πΆ+ (π) and πΆβ (π) (πΆ±2 (π) = πΆ± (π)) respectively. It means that following inequalities are true: βπΊ + (π‘, π , π)β =β₯ π(π‘, π)πΆ+ (π)π β1 (π , π) β₯β€ πΎ1 expβπΌ1 (π‘βπ ) , π‘ β₯ π , βπΊΜ + (π‘, π , π)β =β₯ π(π‘, π)(πΌ β πΆ+ (π))π β1 (π , π) β₯β€ πΎ1 expβπΌ1 (π βπ‘) , π β₯ π‘, π‘, π β β+ , (10) βπΊ β (π‘, π , π)β =β₯ π(π‘, π)πΆβ (π)π β1 (π , π) β₯β€ πΎ2 expβπΌ2 (π‘βπ ) , π‘ β₯ π , βπΊΜ β (π‘, π , π)β =β₯ π(π‘, π)(πΌ β πΆβ (π))π β1 (π , π) β₯β€ πΎ2 expβπΌ2 (π βπ‘) , π β₯ π‘, π‘, π β ββ . The general solution of system(1),(2) that is bounded on semiaxes β+ and ββ has the form π₯(π‘, π, π) = π‘ πΊ + (π‘, 0, π)π + β« πΊ+ (π‘, π , π)π1 (ππ (π))ππ + 0<π‘π (π)β€π‘ 0 β β β« βπΊΜ+ (π‘, π , π)π1 (ππ (π))ππ β , π‘ β₯ 0, πΊΜ+ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β π‘β€π‘π (π)<β π‘ π(ππ‘ (π)) [ = πΊ + (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β β ] π‘ πΊΜβ (π‘, 0, π)π + β« πΊ β (π‘, π , π)π1 (ππ (π))ππ + β πΊβ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β ββ<π‘π (π)β€π‘ ββ 0 β β« βπΊΜβ (π‘, π , π)π1 (ππ (π))ππ β , π‘ β€ 0. πΊΜ β (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β π‘β€π‘π (π)<0 π‘ π(ππ‘ (π)) {[ ] (11) Solution (11) will be bounded on whole axis if and only if impulse action is fixed when π‘ = 0: π₯(0+, π, π) β π₯(0β, π, π) = πΌ(π). (12) It means that vector constant π = π(π) β βπβπ satisfy algebraic system which is received from (11) when π‘ = 0: β 0 π·(π)π = πΌ(π) + β« βπΊ β (0, π , π)π1 (ππ (π))ππ + β« βπΊΜ+ (0, π , π)π1 (ππ (π))ππ + ββ + β ββ<π‘π (π)β€0 βπΊ β (0, π‘π (π), π)πΌ (ππ‘π (π) (π)) + (13) 0 β Μ + (0, βπΊ π‘π (π), π)πΌ (ππ‘π (π) (π)) , 0β€π‘π (π)<β where π·(π) = πΆ+ (π) β (πΌ β πΆβ (π)) is (π β π ) × (π β π )-dimentional matrix. This algebraic system has solution if and only if the condition of orthogonality is true: 151 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 β 0 ππ·π΅ (π) { β« βπΊ β (0, π , π)π1 (ππ (π))ππ + β« βπΊΜ + (0, π , π)π1 (ππ (π))ππ + πΌ(π) + ββ + 0 (14) βπΊβ (0, π‘π (π), π)πΌ (ππ‘π (π) (π)) + β ββ<π‘π (π)β€0 β βπΊΜ + (0, π‘π (π), π)πΌ (ππ‘π (π) (π))} β‘ 0, 0β€π‘π (π)<β where ππ·π΅ (π) is (π β π ) × (π β π )-dimentional matrix which is ortoprojector to matrix π·π΅ (π). Then the bounded on β general solution of system (1),(2) has the form (11) where constant π = π(π) β βπβπ can be determine from the system (13) in the following way: π = ππ· (π)π + π· + (π)Ξ where π·+ (π) is Moore-Penrose pseudoinverse matrix (see Boichuk and Samoilenko, 2004) to matrix π·(π) and β 0 Ξ = { β« βπΊ β (0, π , π)π1 (ππ (π))ππ + β« βπΊΜ+ (0, π , π)π1 (ππ (π))ππ + πΌ(π) + ββ + β 0 βπΊβ (0, π‘π (π), π)πΌ (ππ‘π (π) (π)) + ββ<π‘π (π)β€0 β βπΊΜ + (0, π‘π (π), π)πΌ (ππ‘π (π) (π))}. 0β€π‘π (π)<β By substituting constant π into (11) we obtain the bounded on whole axis solution of (1),(2) in the form π₯(π‘, π, π) = πΊ+ (π‘, 0, π)ππ· (π)π + πΊ+ (π‘, 0, π)π·+ (π)Ξ + π‘ + β« βπΊ+ (π‘, π , π)π1 (ππ (π))ππ + 0 β 0<π‘π (π)β€π‘ β β β« βπΊΜ+ (π‘, π , π)π1 (ππ (π))ππ β β , π‘ β₯ 0, βπΊΜ+ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) π‘β€π‘π (π)<β π‘ π(ππ‘ (π)) πΊΜβ (π‘, 0, π)ππ· (π)π + πΊΜβ (π‘, 0, π)π·+ (π)Ξ + [ = βπΊ+ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β ] (15) π‘ + β« βπΊβ (π‘, π , π)π1 (ππ (π))ππ + β βπΊβ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β ββ<π‘π (π)β€π‘ ββ , π‘ β€ 0. 0 β β« βπΊΜβ (π‘, π , π)π1 (ππ (π))ππ β π‘ {[ β βπΊΜ β (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) π‘β€π‘π (π)<0 π(ππ‘ (π)) ] Since ππ·π΅ (π)π·(π) = 0 then ππ·π΅ (π)πΆβ (π) = ππ·π΅ (π)(πΌ β πΆ+ (π)). In this case necessary and sufficient condition for existence solution of system (1),(2) has the form 152 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 +β ππ·π΅ (π) { β« βπΊ β (0, π , π)π1 (ππ (π))ππ + βπΊβ (0, π‘π (π), π)πΌ (ππ‘π (π) (π)) + πΌ(π)} β‘ 0. (16) β ββ<π‘π (π)<β ββ Taking into account equality and condition (14) we obtain [πΆ+ (π) β (πΌ β πΆβ (π))]π·+ (π) = πΌ β ππ·π΅ (π), πΆ+ (π)π·+ (π)Ξ β πΌΞ = (πΌ β πΆβ (π))π·+ (π)Ξ. Further, since π·(π)ππ· (π) = [πΆ+ (π) β (πΌ β πΆβ (π))]ππ· (π) = 0, we obtain πΆ+ (π)ππ· (π) = (πΌ β πΆβ (π)ππ· (π)). Let us consider the case when the homogeneous system (7) don`t have nontrivial bounded on whole axis solutions. It means πΆ+ (π)ππ· (π) = (πΌ β πΆβ (π))ππ· (π) = 0. Then we can rewrite (15) in the form π₯(π‘, π, π) = π(π‘, π)πΆ+ (π)π· + (π)Ξ + π‘ + β« βπΊ+ (π‘, π , π)π1 (ππ (π))ππ + 0 β 0<π‘π (π)β€π‘ β β β« βπΊΜ+ (π‘, π , π)π1 (ππ (π))ππ β β , π‘ β₯ 0, βπΊΜ+ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) π‘β€π‘π (π)<β π‘ π(ππ‘ (π)) π(π‘, π)(πΆ+ (π)π· + (π) β πΌ)Ξ + [ = βπΊ+ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β ] (17) π‘ + β« βπΊβ (π‘, π , π)π1 (ππ (π))ππ + β βπΊβ (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) β ββ<π‘π (π)β€π‘ ββ , π‘ β€ 0. 0 β β« βπΊΜβ (π‘, π , π)π1 (ππ (π))ππ β β βπΊΜ β (π‘, π‘π (π), π)πΌ (ππ‘π (π) (π)) π‘β€π‘π (π)<0 π‘ π(ππ‘ (π)) {[ ] 3. Conditions of Existence of Invariant Torus Let us show that expression β π₯(0, π) = π’(π) = [ β« βπΊ0 (π , π)π1 (ππ (π))ππ + ββ β βπΊ0 (π‘π (π), π)πΌ (ππ‘π (π) (π)) ββ<π‘π (π)<β ], (18) π(π) where πΊ0 (π , π) = { πΆ+ (π)π· + (π)πΊβ (0, π , π), π < 0, (πΆ+ (π)π·+ (π) β πΌ)πΊΜ+ (0, π , π), π β₯ 0, (19) 153 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 which is obtained from (17) when π‘ = 0 defines for all π β π―π invariant torus of system (1),(2). As shown if condition (16) is fullfilled then system (1),(2) has bounded on β solution of the form (17) at fixed π β π―π . Let us show that the condition (16) on solutions ππ‘ (π) of corresponding Cauchy problem defines invariant set. Using notations presented in paper Korol(2016) we have for all π‘ β β and π β π―π +β ππ·π΅ (ππ‘ (π)) { β« βπΊβ (0, π , ππ‘ (π))π1 (ππ (ππ‘ (π)))ππ + ββ + β βπΊβ (0, π‘π (π), ππ‘ (π))πΌ (ππ‘π (π) (ππ‘ (π))) + πΌ(ππ‘ (π))} = ββ<π‘π (π)<β +β = π(π‘, π)ππ·π΅ (π) { β« βπΊ β (0, π + π‘, π)π1 (ππ +π‘ (π))ππ + ββ + β βπΊ β (0, π‘π (π) + π‘, π)πΌ (ππ‘π (π)+π‘ (π)) + πΌ(ππ‘ (π))} = 0. ββ<π‘π (π)<β Due to exponential dichotomy of system (7) on semiaxes and estimation (3) integral and sum are convergent. We will use the boundness of matrix-projector β₯ ππ·π΅ (ππ‘ (π)) β₯β€ πΎ0 for all π β π―π . +β βππ·π΅ (π) { β« βπΊ β (0, π , π)π1 (ππ (π))ππ + βπΊ β (0, π‘π (π), π)πΌ (ππ‘π (π) (π)) + πΌ(π)}β β€ β ββ<π‘π (Ο)<β ββ 0 β€ βππ·π΅ (π)β { β« ββπΊβ (0, π , π)π1 (ππ (π))ππ β + ββπΊβ (0, π‘π (π), π)πΌ(ππ‘π (π) (π))β + β ββ<π‘π (π)<0 ββ +β + β« ββπΊΜ+ (0, π , π)π1 (ππ (π))ππ β + β ββπΊΜ + (0, π‘π (π), π)πΌ(ππ‘π (π) (π))β + βπΌ(π)β} β€ 0<π‘π (π)<β 0 πΎ2 πΎ1 πΎ2 πΎ1 β€ πΎ0 ( + ) βπ1 (π)β + πΎ0 ( + + 1) βπΌ(π)β. βπΌ π πΌ2 πΌ1 1βπ 2 1 β π βπΌ1 π Now we will show that if condition (16) fullfills then expression (18) defines invariant torus of system (1),(2) for all π β π―π . Let us show that the integral and sum in (18) are convergent. Taking into account inequality (10) we consider the first (π β π ) rows of expression (18): 0 βπΆ+ (π)π· + (π) { β« βπΊ β (0, π , π)π1 (ππ (π))ππ + ββ β +[πΆ+ (π)π·+ (π) β πΌ ] {β« βπΊΜ+ (0, π , π)π1 (ππ (π))ππ 0 βπΊβ (0, π‘π (π), π)πΌ (ππ‘π (π) (π))} + β ββ<π‘π (π)<0 + β βπΊΜ + (0, π‘π (π), π)πΌ(ππ‘π (π) (π))}β β€ 0<π‘π (π)<β πΎ5 πΎ2 πΎ6 πΎ1 πΎ5 πΎ2 πΎ6 πΎ1 ] βπ1 β + ( ) βπΌ(π)β, β€[ + + βπΌ π 2 πΌ2 πΌ1 1βπ 1 β π βπΌ1 π where βπΆ+ (π)π·+ (π)β = πΎ5 , βπΆ+ (π)π· + (π) β πΌβ = πΎ6 and therefore as in Samoilenko at al.,(2000), from condition π1 (π) β πΆ(π―π ) it follows that π’(ππ‘ (π)) β πΆ(π―π ). Belonging the other π rows of 154 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 expression (18) to space πΆ(π―π ) follows from the fact that π2 (π) β πΆ π β1 (π―π ). From estimates (10) and well-known properties (see Samoilenko,1987) πΆ+ (ππ (π)) = π(π, π)πΆ+ (π)π β1 (π, π), πΆβ (ππ (π)) = π(π, π)πΆβ (π)π β1 (π, π), π(π‘, π, ππ (π)) = π(π‘ + π , π + π , π), π(π‘, π, π)π(π, π , π) = π(π‘, π , π), β1 (π(π‘, π, π)) we obtain ππ‘ (ππ (π)) = ππ‘+π (π), = π(π, π‘, π), β π’(ππ‘ (π)) = [ β« βπΊ0 (π , ππ‘ (π))π1 (ππ (ππ‘ (π)))ππ + β ββ<π‘π (π)<β ββ β« βπΊπ‘ (π , π)π1 (ππ (π))ππ + ββ ]= π(ππ‘ (π)) β =[ βπΊ0 (π‘π (π), ππ‘ (π))πΌ(ππ‘π (π) (ππ‘ (π))) β ββ<π‘π (π)<β βπΊπ‘ (π‘π (π), π)πΌ(ππ‘π (π) (π)) ], π(ππ‘ (π)) for all π‘ β β and π β π―π . It proves that π’(ππ‘ (π)) β πΆ 1 (π―π ) and set π’(π) defines invariant torus of system (1),(2). Thus we proved the following result. Theorem 1. Let the system (7) is exponential dichotomous on semiaxes π + and π β with projectors πΆ± (π), the condition (3) is executed and π2 (π) β βπ β1 (π π ). Then 1. system (1),(2) has invariant torus if and only if π1 (π) β πΆ(π―π ) satisfy the condition (16); 2. moreover if the homogeneous system (7) don`t have nontrivial solutions bounded on whole axis then expression (18) defines for all π β π―π a unique invariant torus of system (1),(2). 4. Conclusions In this paper we showed that if the linear nonhomogeneous singular system can be reduce to the central canonical form and the corresponding homogeneous system is exponential dichotomous on semiaxes then this system has a bounded on whole axis solution. The main result was presented in theorem 1 where we found the necessary and sufficient condition for existence of a unique invariant torus of singular impulsive system. References Boichuk A.A. (2007) A criterion for the existence of the unique invariant torus of a linear extension of dynamical systems. Ukrainian Mathematical Journal - 2007 - 59(1). - p.1-11 http://dx.doi.org/10.1007/s11253-007-0001-8 Boichuk A.A., Samoilenko A.M. (2004) Generalized inverse operators and fredholm boundary value problems. - Utrecht; Boston: VSP, 2004. - 317p. http://dx.doi.org/10.1515/9783110944679 Boyarincev Yu.E., Danilov V.A., Loginov A.A., Chistyakov V.F. (1989) Numerical methods for solving singular systems. β Novosibirsk: Nauka. 1989. β 223p. Campbell S.L., Petzold L.R. (1983) Canonical forms and solvable singular systems of differential equations. SIAM J. Alg. Discrete Methods. β 1983(4). β p.517-521. http://dx.doi.org/10.1137/0604051 155 Yurii Korol / Journal of Applied Mathematics and Statistics (2016) Vol. 3 No. 3 pp. 149-156 Korol Yu. Yu. (2016) Existence of an invariant torus for a degenerate linear extension of a dynamical system Nonlinear Oscillations. - 2016. - 19(2), p.217-226 Samoilenko A.M. (1987) Elements of mathematical theory of multi-frequency oscillations. Invariant tori. β M: Nauka, 1987. β 304 p. Samoilenko A.M., Perestyuk N.A. (1987) Differential equations with impulsive actions. β K.: Vyshcha shkola, 1987. β 288 p. Samoilenko A.M., Shkil M.I., Yakovets V.P. (2000) Linear systems of differential equations with singularities. β K.: Vyshcha shkola, 2000. β 294p. 156
© Copyright 2026 Paperzz