Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices

In recent results on representation of solutions of systems of delayed differential equations the condition that the linear parts are given by pairwise permutable matrices was assumed. In this paper it is shown how this strong condition can be avoided, and representation of solutions of systems of d...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Нелінійні коливання
Datum:2016
Hauptverfasser: Medveď, M., Pospíšil, M.
Format: Artikel
Sprache:English
Veröffentlicht: Інститут математики НАН України 2016
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/177287
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices / M. Medveď, M. Pospíšil // Нелінійні коливання. — 2016. — Т. 19, № 4. — С. 521-532 — Бібліогр.: 10 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-177287
record_format dspace
spelling Medveď, M.
Pospíšil, M.
2021-02-14T08:17:58Z
2021-02-14T08:17:58Z
2016
Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices / M. Medveď, M. Pospíšil // Нелінійні коливання. — 2016. — Т. 19, № 4. — С. 521-532 — Бібліогр.: 10 назв. — англ.
1562-3076
https://nasplib.isofts.kiev.ua/handle/123456789/177287
517.9
In recent results on representation of solutions of systems of delayed differential equations the condition that the linear parts are given by pairwise permutable matrices was assumed. In this paper it is shown how this strong condition can be avoided, and representation of solutions of systems of differential equations with nonconstant coefficients and variable delays is derived. The results are applied to a system with two constant delays. Also the nonexistence of blow-up solutions is proved for nonlinear systems.
Останнi результати про зображення розв’язкiв систем диференцiальних рiвнянь iз запiзненням були отриманi за умови комутування матриць, що визначають лiнiйнi частини. У статтi показано як можна позбутися цiєї сильної умови та отримано зображення розв’язкiв систем диференцiальних рiвнянь з несталими коефiцiєнтами та запiзненнями, що змiнюються. Цi результати застосовано до системи зi сталими запiзненнями. Також для нелiнiйних систем доведено вiдсутнiсть вибухових розв’язкiв.
en
Інститут математики НАН України
Нелінійні коливання
Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
Зображення розв’язків систем лінійних диференціальних рівнянь з багатьма запізненнями та лінійними частинами, визначеними некомутуючими матрицями
Представление решений систем линейных дифференциальных уравнений со многими запаздываниями и линейными частями, определенными некоммутирующими матрицами
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
spellingShingle Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
Medveď, M.
Pospíšil, M.
title_short Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
title_full Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
title_fullStr Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
title_full_unstemmed Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
title_sort representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices
author Medveď, M.
Pospíšil, M.
author_facet Medveď, M.
Pospíšil, M.
publishDate 2016
language English
container_title Нелінійні коливання
publisher Інститут математики НАН України
format Article
title_alt Зображення розв’язків систем лінійних диференціальних рівнянь з багатьма запізненнями та лінійними частинами, визначеними некомутуючими матрицями
Представление решений систем линейных дифференциальных уравнений со многими запаздываниями и линейными частями, определенными некоммутирующими матрицами
description In recent results on representation of solutions of systems of delayed differential equations the condition that the linear parts are given by pairwise permutable matrices was assumed. In this paper it is shown how this strong condition can be avoided, and representation of solutions of systems of differential equations with nonconstant coefficients and variable delays is derived. The results are applied to a system with two constant delays. Also the nonexistence of blow-up solutions is proved for nonlinear systems. Останнi результати про зображення розв’язкiв систем диференцiальних рiвнянь iз запiзненням були отриманi за умови комутування матриць, що визначають лiнiйнi частини. У статтi показано як можна позбутися цiєї сильної умови та отримано зображення розв’язкiв систем диференцiальних рiвнянь з несталими коефiцiєнтами та запiзненнями, що змiнюються. Цi результати застосовано до системи зi сталими запiзненнями. Також для нелiнiйних систем доведено вiдсутнiсть вибухових розв’язкiв.
issn 1562-3076
url https://nasplib.isofts.kiev.ua/handle/123456789/177287
citation_txt Representation of solutions of systems of linear differential equations with multiple delays and linear parts given by nonpermutable matrices / M. Medveď, M. Pospíšil // Нелінійні коливання. — 2016. — Т. 19, № 4. — С. 521-532 — Бібліогр.: 10 назв. — англ.
work_keys_str_mv AT medvedm representationofsolutionsofsystemsoflineardifferentialequationswithmultipledelaysandlinearpartsgivenbynonpermutablematrices
AT pospisilm representationofsolutionsofsystemsoflineardifferentialequationswithmultipledelaysandlinearpartsgivenbynonpermutablematrices
AT medvedm zobražennârozvâzkívsistemlíníinihdiferencíalʹnihrívnânʹzbagatʹmazapíznennâmitalíníinimičastinamiviznačeniminekomutuûčimimatricâmi
AT pospisilm zobražennârozvâzkívsistemlíníinihdiferencíalʹnihrívnânʹzbagatʹmazapíznennâmitalíníinimičastinamiviznačeniminekomutuûčimimatricâmi
AT medvedm predstavlenierešeniisistemlineinyhdifferencialʹnyhuravneniisomnogimizapazdyvaniâmiilineinymičastâmiopredelennyminekommutiruûŝimimatricami
AT pospisilm predstavlenierešeniisistemlineinyhdifferencialʹnyhuravneniisomnogimizapazdyvaniâmiilineinymičastâmiopredelennyminekommutiruûŝimimatricami
first_indexed 2025-11-25T22:45:17Z
last_indexed 2025-11-25T22:45:17Z
_version_ 1850570973972004864
fulltext UDC 517.9 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL EQUATIONS WITH MULTIPLE DELAYS AND LINEAR PARTS GIVEN BY NONPERMUTABLE MATRICES* ЗОБРАЖЕННЯ РОЗВ’ЯЗКIВ СИСТЕМ ЛIНIЙНИХ ДИФЕРЕНЦIАЛЬНИХ РIВНЯНЬ З БАГАТЬМА ЗАПIЗНЕННЯМИ ТА ЛIНIЙНИМИ ЧАСТИНАМИ, ВИЗНАЧЕНИМИ НЕКОМУТУЮЧИМИ МАТРИЦЯМИ M. Medved’ Dep. Math., Anal. and Numer. Math., Fac. Math., Phys. and Inform., Comenius Univ. Bratislava Mlynská dolina, 842 48 Bratislava, Slovak Republic e-mail: Milan.Medved@fmph.uniba.sk M. Pospı́šil Dep. Math., Anal. and Numer. Math., Fac. Math., Phys. and Inform., Comenius Univ. Bratislava Mlynská dolina, 842 48 Bratislava, Slovak Republic and Math. Inst. Slovak Acad. Sci. Štefánikova 49, 814 73 Bratislava, Slovak Republic e-mail: Michal.Pospisil@fmph.uniba.sk In recent results on representation of solutions of systems of delayed differential equations the condition that the linear parts are given by pairwise permutable matrices was assumed. In this paper it is shown how this strong condition can be avoided, and representation of solutions of systems of differential equations with nonconstant coefficients and variable delays is derived. The results are applied to a system with two constant delays. Also the nonexistence of blow-up solutions is proved for nonlinear systems. Останнi результати про зображення розв’язкiв систем диференцiальних рiвнянь iз запiзненням були отриманi за умови комутування матриць, що визначають лiнiйнi частини. У статтi по- казано як можна позбутися цiєї сильної умови та отримано зображення розв’язкiв систем ди- ференцiальних рiвнянь з несталими коефiцiєнтами та запiзненнями, що змiнюються. Цi резуль- тати застосовано до системи зi сталими запiзненнями. Також для нелiнiйних систем доведено вiдсутнiсть вибухових розв’язкiв. 1. Introduction and preliminaries. Systems of linear differential equations with one or multi- ple constant delays were considered in [5, 6] and solutions were represented using matrix polynomials of a time-dependent degree. In the case of multiple delays, pairwise permutabi- lity of matrices representing coefficients of linear terms was a necessary condition for deriving the representation. That means that the matrices A,B1, . . . , Bn in the equation ẋ(t) = Ax(t) +B1x(t− τ1) + . . .+Bnx(t− τn) + f(t), t ≥ 0, ∗ The first author was supported by the Slovak Research and Development Agency under the contract No. APVV-14-0378. The second author was supported by the Grant VEGA-SAV 2/0153/16. c© M. Medved’, M. Pospı́šil, 2016 ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 521 522 M. MEDVED’, M. POSPÍŠIL had to satisfyABi = BiA, BiBj = BjBi for each i, j = 1, . . . , n. For now the most general case with nonconstant coefficients and variable delays was investigated in [9]. We recall this result. Theorem 1.1. Let n ∈ N, Bi ∈ C([0,∞), L(RN ,RN )), gi ∈ G0 for i = 1, . . . , n and Gs := { g ∈ C([s,∞),R) | g(t) < t on [s,∞), g is increasing } , f ∈ C([0,∞),RN ), γ := min{g1(0), . . . , gn(0)}, ϕ ∈ C([γ, 0],RN ). Then the solution of the equation ẋ(t) = B1(t)x(g1(t)) + . . .+Bn(t)x(gn(t)) + f(t), t ≥ 0, (1.1) satisfying the initial condition x(t) = ϕ(t), γ ≤ t ≤ 0, (1.2) has the form x(t) =  ϕ(t), γ ≤ t < 0, XB1,...,Bn g1,...,gn (t, 0)ϕ(0) + ∫ t 0 XB1,...,Bn g1,...,gn (t, s)× × [B1(s)ψ(g1(s)) + . . .+Bn(s)ψ(gn(s))] ds+ + ∫ t 0 XB1,...,Bn g1,...,gn (t, s)f(s)ds, 0 ≤ t, (1.3) where ψ(t) = { ϕ(t), t ∈ [γ, 0), θ, t /∈ [γ, 0), (1.4) with the N -dimensional vector θ of zeros, and XB1,...,Bm g1,...,gm (t, s) :=  Θ, t < s, Y (t, s), s ≤ t < g−1m (s), Y (t, s) + ∫ t g−1 m (s) Y (t, q1)Bm(q1)Y (gm(q1), s)dq1 + . . . . . .+ ∫ t g−km (s) Y (t, q1)Bm(q1)× × ∫ gm(q1) g −(k−1) m (s) Y (gm(q1), q2)Bm(q2)× . . . . . .× ∫ gm(qk−1) g−1 m (s) Y (gm(qk−1), qk)× ×Bm(qk)Y (gm(qk), s)dqk . . . dq1, g−km (s) ≤ t < g −(k+1) m (s), k ∈ N, (1.5) ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL . . . 523 where Y (t, s) = X B1,...,Bm−1 g1,...,gm−1 (t, s), for m = 2, . . . , n, the N ×N zero matrix Θ, and XB g (t, s) :=  Θ, t < s, I, s ≤ t < g−1(s), I + ∫ t g−1(s) B(q1)dq1 + . . .+ ∫ t g−k(s) B(q1)× × ∫ g(q1) g−(k−1)(s) B(q2) . . . ∫ g(qk−1) g−1(s) B(qk)dqk . . . dq1, g−k(s) ≤ t < g−(k+1)(s), k ∈ N. Note that the pairwise permutability of B1(t), . . . , Bn(t) was not needed. However, a result from [9] on the equation involving a nondelayed term on the right-hand side was stated only for the case of pairwise permutable constant coefficients. In the present paper we show how to tackle this problem without permutability condition and even for nonconstant coefficients. So this is the most general case for the delay functions from the class G0. To illustrate our results, we consider a system of delayed differential equations with constant coefficients and two constant delays in Section 3, and derive a formula for its solution. A known result for permutable matrices [10] (see also [3]) is obtained as a particular case. In the final section we apply results of Section 2 to prove a criterion on a nonexistence of blow-up solutions for differential equations with variable coefficients and nonconstant delays. By such a solution we understand a function x : [a, b) → RN with a ∈ R, a < b ≤ ∞ such that limt→T− ‖x(t)‖ = ∞ for some a < T < b and a vector norm ‖ · ‖. We note that the main result of this section (Theorem 4.1) is a generalization of a weaker result from [7]. Further applications of results of this paper can be achieved, e.g., in stability theory [6 – 9] or Fredholm boundary-value problems [2]. 2. Representation of solutions of general time dependent systems. In this section we shall investigate systems of differential equations with multiple variable delays. For the simplicity we start to consider the equation with constant coefficients, ẋ(t) = Ax(t) +B1x(g1(t)) + . . .+Bnx(gn(t)) + f(t), t ≥ 0. (2.1) Theorem 2.1. Let n ∈ N, A, Bi, i = 1, . . . , n, be N ×N matrices, gi ∈ G0 for i = 1, . . . , n, γ := min{g1(0), . . . , gn(0)}, f ∈ C([0,∞),RN ) and ϕ ∈ C([γ, 0],RN ) be given functions. Then the solution of the initial value problem (2.1), (1.2) has the form x(t) =  ϕ(t), γ ≤ t < 0, X̃(t, 0)ϕ(0) + ∫ t 0 X̃(t, s)[B1ψ(g1(s)) + . . . . . .+Bnψ(gn(s))]ds+ ∫ t 0 X̃(t, s)f(s)ds, 0 ≤ t, for ψ given by (1.4), and X̃(t, s) = eAtXB̃1,...,B̃n g1,...,gn (t, s)e−As for XB1,...,Bn g1,...,gn (t, s) defined by (1.5), and B̃i(t) = e−AtBie Agi(t) for i = 1, . . . , n. ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 524 M. MEDVED’, M. POSPÍŠIL Proof. Let x(t) = eAty(t). Then y(t) satisfies ẏ(t) = B̃1(t)y(g1(t)) + . . .+ B̃n(t)y(gn(t)) + f̃(t), t ≥ 0, y(t) = ϕ̃(t), γ ≤ t ≤ 0, (2.2) for f̃(t) = e−Atf(t), ϕ̃(t) = e−Atϕ(t). Theorem 1.1 implies y(t) =  ϕ̃(t), γ ≤ t < 0, XB̃1,...,B̃n g1,...,gn (t, 0)ϕ̃(0) + ∫ t 0 XB̃1,...,B̃n g1,...,gn (t, s)× × [ B̃1(s)ψ̃(g1(s)) + . . .+ B̃n(s)ψ̃(gn(s)) ] ds+ + ∫ t 0 XB̃1,...,B̃n g1,...,gn (t, s)f̃(s)ds, 0 ≤ t, (2.3) where ψ̃(t) = { ϕ̃(t), t ∈ [γ, 0), θ, t /∈ [γ, 0). (2.4) Note that ϕ̃(0) = ϕ(0), and ψ̃(gi(t)) = e−Agi(t)ψ(gi(t)) for i = 1, . . . , n. The statement is obtained when one returns back to x(t). Now, we turn to variable coefficients. So we shall consider the equation ẋ(t) = A(t)x(t) +B1(t)x(g1(t)) + . . .+Bn(t)x(gn(t)) + f(t), t ≥ 0. (2.5) Theorem 2.2. Let n ∈ N, A,Bi ∈ C([0,∞), L(RN ,RN )), gi ∈ G0 for i = 1, . . . , n, γ := := min{g1(0), . . . , gn(0)}, f ∈ C([0,∞),RN ) and ϕ ∈ C([γ, 0],RN ) be given functions. Then the solution of the initial value problem (2.5), (1.2) has the form x(t) =  ϕ(t), γ ≤ t < 0, X̃(t, 0)ϕ(0) + ∫ t 0 X̃(t, s)[B1(s)ψ(g1(s)) + . . . . . .+Bn(s)ψ(gn(s))]ds+ ∫ t 0 X̃(t, s)f(s)ds, 0 ≤ t, (2.6) for ψ given by (1.4), and X̃(t, s) = Φ(t)XB̃1,...,B̃n g1,...,gn (t, s)Φ−1(s) for XB1,...,Bn g1,...,gn (t, s) defined by (1.5), B̃i(t) = Φ−1(t)Bi(t)Φ(gi(t)) for i = 1, . . . , n and Φ(t) is a fundamental matrix satisfying Φ̇(t) = A(t)Φ(t), t ≥ γ, Φ(0) = I (2.7) with A(t) = { A(t), t ≥ 0, A(0), t < 0, ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL . . . 525 where Φ̇ at 0 is considered one-sidedly. Proof. Let x(t) = Φ(t)y(t).Then y(t) solves (2.2) with f̃(t) = Φ−1(t)f(t), ϕ̃(t) = Φ−1(t)ϕ(t) and B̃i(t) as in the statement of the theorem. So, y(t) has the form (2.3). When one returns to x(t) the statement immediately follows. Remark 2.1. Note that from (1.5) it follows that XB1,...,Bn g1,...,gn (t, s) does not depend on the values of Bi(q) on [0, g−1i (s)) for each i = 1, . . . , n. Consequently, we can take any extension of A(t) onto [γ,∞) instead of A(t) and the same solution x(t) given by (2.6) results. Next, we consider a particular case of matrix functions A,B1, . . . , Bn. Corollary 2.1. Let n ∈ N, Bi ∈ C([0,∞), L(RN ,RN )) for i = 1, . . . , n, Q ∈ C1(R, L(RN , RN )) be a nonsingular T -periodic matrix with Q(0) = I, R be a constant N ×N matrix, gi ∈ G0 for i = 1, . . . , n, γ := min{g1(0), . . . , gn(0)}, f ∈ C([0,∞),RN ) and ϕ ∈ C([γ, 0],RN ) be given functions. Then the solution of the initial value problem consisting of the equation ẋ(t) = ( Q̇(t) +Q(t)R ) Q−1(t)x(t) +B1(t)x(g1(t)) + . . .+Bn(t)x(gn(t)) + f(t), t ≥ 0, and initial condition (1.2) has the form (2.6) for ψ given by (1.4), X̃(t, s) = Φ(t)XB̃1,...,B̃n g1,...,gn (t, s)× ×Φ−1(s) for XB1,...,Bn g1,...,gn (t, s) defined by (1.5), B̃i(t) = Φ−1(t)Bi(t)Φ(gi(t)), i = 1, . . . , n, and Φ(t) = { Q(t)eRt, t ≥ 0, e(Q̇(0)+R)t, t < 0, Proof. Noting that Φ(t) is a C1 function satisfying (2.7) with A(t) = { ( Q̇(t) +Q(t)R ) Q−1(t), t ≥ 0, Q̇(0) +R, t < 0, the statement follows from Theorem 2.2. 3. Derivation of a formula for solutions of time independent systems. Here, we apply our results to find a solution of a system with linear terms represented by nonpermutable constant matrices and constant delays. Let us consider the initial value problem consisting of the equation ẋ(t) = B1x(t− τ1) +B2x(t− τ2) + f(t), t ≥ 0, (3.1) and initial condition (1.2) where τ1, τ2 > 0, B1, B2 are constant N × N matrices and f ∈ ∈ C([0,∞),RN ), ϕ ∈ C([γ, 0],RN ) are given functions with γ = −max{τ1, τ2}. The assumpti- ons of Theorem 2.1 are satisfied with A = Θ and gi(t) = t− τi for i = 1, 2. Hence the theorem gives a solution of (3.1), (1.2). In this case, X̃(t, s) = XB1,B2 g1,g2 (t, s) and the formula for the soluti- on can be simplified. Note that XB1 g1 (·, s) is a matrix solution of Ẋ(t) = B1X(t− τ1), t ≥ s, X(s) = I. ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 526 M. MEDVED’, M. POSPÍŠIL Thus we can use [5] to write XB1 g1 (t, s) = ⌊ t−s τ1 ⌋∑ i=0 Bi 1 (t− iτ1 − s)i i! (3.2) with the floor function b·c. Here we used the property of empty sum, i.e., b∑ i=a z(i) = ∑ i∈∅ z(i) = 0 for any function z provided that a > b. We shall compute XB1,B2 g1,g2 (t, s) for s + kτ2 ≤ t < < s+ (k + 1)τ2, k ∈ N. In the following we use the step function σ(t) = { 0, t < 0, 1, t ≥ 0. Let 1 ≤ K ≤ k be an arbitrary fixed integer to extend the sums to infinity. Then XK(t, s) := t∫ s+Kτ2 XB1 g1 (t, q1)B2 q1−τ2∫ s+(K−1)τ2 XB1 g1 (q1 − τ2, q2)B2 . . . . . . qK−1−τ2∫ s+τ2 XB1 g1 (qK−1 − τ2, qK)B2X B1 g1 (qK − τ2, s)dqK . . . dq1 = = t∫ s+Kτ2 ∞∑ i0=0 Bi0 1 (t− i0τ1 − q1)i0 i0! σ(t− i0τ1 − q1)B2 . . . . . . qK−2−τ2∫ s+2τ2 ∞∑ iK−2=0 B iK−2 1 (qK−2 − τ2 − iK−2τ1 − qK−1)iK−2 iK−2! × × σ(qK−2 − τ2 − iK−2τ1 − qK−1)B2× × qK−1−τ2∫ s+τ2 ∞∑ iK−1=0 B iK−1 1 (qK−1 − τ2 − iK−1τ1 − qK)iK−1 iK−1! × × σ(qK−1 − τ2 − iK−1τ1 − qK)B2 ∞∑ iK=0 BiK 1 (qK − τ2 − iKτ1 − s)iK iK ! × × σ(qK − τ2 − iKτ1 − s)dqK . . . dq1. (3.3) ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL . . . 527 The last integral on the right-hand side is equal to ∞∑ iK−1,iK=0 B iK−1 1 B2B iK 1 σ(qK−1 − 2τ2 − (iK−1 + iK)τ1 − s)× × qK−1−τ2−iK−1τ1∫ s+τ2+iKτ1 (qK−1 − τ2 − iK−1τ1 − qK)iK−1 iK−1! × × (qK − τ2 − iKτ1 − s)iK iK ! dqK = = ∞∑ iK−1,iK=0 B iK−1 1 B2B iK 1 (qK−1 − 2τ2 − (iK−1 + iK)τ1 − s)iK−1+iK+1 (iK−1 + iK + 1)! × × σ(qK−1 − 2τ2 − (iK−1 + iK)τ1 − s) = = ∞∑ jK−1=0 (qK−1 − 2τ2 − jK−1τ1 − s)jK−1+1 (jK−1 + 1)! σ(qK−1 − 2τ2 − jK−1τ1 − s)× × jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 , where we used the substitution qK = s+ τ2 + iKτ1 + ξ(qK−1− 2τ2− (iK−1 + iK)τ1− s) leading to a multiple of Euler beta function, and then we changed iK−1 + iK → jK−1, iK → jK . Consequently, the last double integral on the right-hand side of (3.3) is equal to ∞∑ iK−2,jK−1=0 B iK−2 1 B2 jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 σ(qK−2 − 3τ2 − (iK−2 + jK−1)τ1 − s)× × qK−2−τ2−iK−2τ1∫ s+2τ2+jK−1τ1 (qK−2 − τ2 − iK−2τ1 − qK−1)iK−2 iK−2! × × (qK−1 − 2τ2 − jK−1τ1 − s)jK−1+1 (jK−1 + 1)! dqK−1 = = ∞∑ iK−2,jK−1=0 B iK−2 1 B2 jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 × × (qK−2 − 3τ2 − (iK−2 + jK−1)τ1 − s)iK−2+jK−1+2 (iK−2 + jK−1 + 2)! × × σ(qK−2 − 3τ2 − (iK−2 + jK−1)τ1 − s) = ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 528 M. MEDVED’, M. POSPÍŠIL = ∞∑ jK−2=0 (qK−2 − 3τ2 − jK−2τ1 − s)jK−2+2 (jK−2 + 2)! σ(qK−2 − 3τ2 − jK−2τ1 − s)× × ∞∑ jK−1=0 B jK−2−jK−1 1 B2 jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 for jK−2 = iK−2 + jK−1. Analogously for other multiple integrals. Finally, the right-hand side of (3.3) is equal to ∞∑ i0,j1=0 Bi0 1 B2 j1∑ j2=0 Bj1−j2 1 B2 . . . jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 σ(t−Kτ2 − (i0 + j1)τ1 − s)× × t−i0τ1∫ s+Kτ2+j1τ1 (t− i0τ1 − q1)i0 i0! (q1 −Kτ2 − j1τ1 − s)j1+K−1 (j1 +K − 1)! dq1 = = ∞∑ i0,j1=0 Bi0 1 B2 j1∑ j2=0 Bj1−j2 1 B2 . . . jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 × × (t−Kτ2 − (i0 + j1)τ1 − s)i0+j1+K (i0 + j1 +K)! σ(t−Kτ2 − (i0 + j1)τ1 − s) = = ⌊ t−Kτ2−s τ1 ⌋∑ j0=0 (t−Kτ2 − j0τ1 − s)j0+K (j0 +K)! × × j0∑ j1=0 Bj0−j1 1 B2 j1∑ j2=0 Bj1−j2 1 B2 . . . jK−1∑ jK=0 B jK−1−jK 1 B2B jK 1 , where j0 = i0 + j1. Now changing j0 → i0, j0 − j1 → i1, . . . , jK−1 − jK → iK we get XK(t, s) = ⌊ t−Kτ2−s τ1 ⌋∑ i0=0 (t−Kτ2 − i0τ1 − s)i0+K (i0 +K)! i0∑ i1=0 Bi1 1 B2 i0−i1∑ i2=0 Bi2 1 B2 . . . . . . i0−(i1+...+iK−1)∑ iK=0 BiK 1 B2B i0−(i1+...+iK) 1 = ⌊ t−Kτ2−s τ1 ⌋∑ i0=0 (t−Kτ2 − i0τ1 − s)i0+K (i0 +K)! × × ∑ i1,...,iK≥0 i1+...+iK≤i0 Bi1 1 B2B i2 1 B2 . . . B iK 1 B2B i0−(i1+...+iK) 1 for each K ∈ N such that 1 ≤ K ≤ k and s + kτ2 ≤ t < s + (k + 1)τ2. So, we obtain the following result. ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL . . . 529 Proposition 3.1. The solution of the initial value problem (3.1), (1.2) has the form x(t) =  ϕ(t), γ ≤ t < 0, X(t, 0)ϕ(0) +B1 ∫ τ1 0 X(t, s)ϕ(s− τ1)ds+ +B2 ∫ τ2 0 X(t, s)ϕ(s− τ2)]ds+ ∫ t 0 X(t, s)f(s)ds, 0 ≤ t, (3.4) where X(t, s) = ⌊ t−s τ2 ⌋∑ K=0 ⌊ t−Kτ2−s τ1 ⌋∑ i0=0 (t−Kτ2 − i0τ1 − s)i0+K (i0 +K)! × × ∑ i1,...,iK≥0 i1+...+iK≤i0 Bi1 1 B2B i2 1 B2 . . . B iK 1 B2B i0−(i1+...+iK) 1 . Proof. From the previous arguments it follows that the solution has the form x(t) =  ϕ(t), γ ≤ t < 0, X(t, 0)ϕ(0) + ∫ t 0 X(t, s)[B1ψ(s− τ1)+ +B2ψ(s− τ2)]ds+ ∫ t 0 X(t, s)f(s)ds, 0 ≤ t, with X(t, s) = ⌊ t−s τ2 ⌋∑ K=0 XK(t, s) forXK(t, s) defined asX0(t, s) := XB1 g1 (t, s) of (3.2) and by (3.3) forK = 1, . . . , ⌊ t− s τ2 ⌋ .Note that X(t, s) = Θ whenever t < s. That gives formula (3.4). In particular, we obtain the known result (see [10]). Corollary 3.1. If B1B2 = B2B1, then the solution of (3.1), (1.2) has the form (3.4) where X(t, s) = ∑ i,j≥0 iτ1+jτ2≤t−s Bi 1B j 2 (t− iτ1 − jτ2 − s)i+j i!j! . Proof. Applying ∑ i1,...,iK≥0 i1+...+iK≤i0 1 = (i0 +K)! i0!K! the statement follows immediately. ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 530 M. MEDVED’, M. POSPÍŠIL 4. Nonexistence of blow-up solutions. The results of Section 2 may be applied, e.g., in stabi- lity or controllability theory. In this section we illustrate the results by proving a nonexistence of blow-up solutions. First we recall the following estimation from [9]. Lemma 1. Let s ∈ R, n ∈ N, Bi ∈ C([s,∞), L(RN ,RN )) and gi ∈ Gs for i = 1, . . . , n. Then ‖XB1,...,Bn g1,...,gn (t, s)‖ ≤ exp  t∫ s n∑ i=1 ‖Bi(q)‖dq  for any t ≥ s. Theorem 4.1. Let n ∈ N, A,Bi ∈ C([0,∞), L(RN ,RN )), gi ∈ G0 for i = 1, . . . , n, γ := := min{g1(0), . . . , gn(0)}, f ∈ C([0,∞)× R(n+1)N ,RN ) and ϕ ∈ C([γ, 0],RN ) be given functi- ons. Let x : [γ, b) → RN with 0 < b ≤ ∞ be a continuous solution of the equation ẋ(t) = A(t)x(t) + n∑ i=1 Bi(t)x(gi(t)) + f(t, x(t), x(g1(t)), . . . , x(gn(t))), t ≥ 0, (4.1) satisfying the initial condition (1.2). If ‖f(t, u0, . . . , un)‖ ≤ n∑ i=0 Ri(t)ωi(‖ui‖), (t, u0, . . . , un) ∈ R× R(n+1)N where Ri, ωi, i = 0, . . . , n, are continuous nonnegative functions defined on [0,∞), and ωi, i = 0, . . . , n, are nondecreasing such that ω0(0) + . . .+ ωn(0) > 0 and ∞∫ 0 du ω0(u) + ∑n i=1 ωi(2u) = ∞, then limt→T− ‖x(t)‖ < ∞ for all T ∈ (0, b). Proof. Let us suppose in contrary that there exists a smallest T ∈ (0, b) such that lim t→T− ‖x(t)‖ = ∞. By Theorem 2.2 and in its notation, x(t) satisfies x(t) = X̃(t, 0)ϕ(0) + n∑ i=1 t∫ 0 X̃(t, s)Bi(s)ψ(gi(s))ds+ t∫ 0 X̃(t, s)F (s)ds (4.2) for t ≥ 0, where F (t) = f(t, x(t), x(g1(t)), . . . , x(gn(t))). Now, since Φ(t) = I + t∫ 0 A(s)Φ(s)ds, t ≥ 0, ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 REPRESENTATION OF SOLUTIONS OF SYSTEMS OF LINEAR DIFFERENTIAL . . . 531 Gronwall lemma [4] yields ‖Φ(t)‖ ≤ exp  t∫ 0 ‖A(s)‖ds  , t ≥ 0, for an induced matrix norm ‖ · ‖. Similarly, Φ−1(t) = I− t∫ 0 Φ−1(s)A(s)ds, t ≥ 0, i.e., ‖Φ−1(t)‖ ≤ exp  t∫ 0 ‖A(s)‖ds  , t ≥ 0. Therefore, along with Lemma 4.1, ‖X̃(t, s)‖ ≤ exp  t∫ 0 ‖A(q)‖dq + t∫ s n∑ i=1 ‖B̃i(q)‖dq + s∫ 0 ‖A(q)‖dq  =: M(t, s) for any 0 ≤ s ≤ t. Hence, denoting ‖ϕ‖ := maxγ≤t≤0 ‖ϕ(t)‖, from (4.2) one obtains ‖x(t)‖ ≤ M(t, 0)‖ϕ‖+ n∑ i=1 min{t,g−1 i (0)}∫ 0 M(t, s)‖Bi(s)‖‖ϕ‖ds+ + t∫ 0 M(t, s) ( R0(s)ω0(‖x(s)‖) + n∑ i=1 Ri(s)ωi(‖x(gi(s))‖) ) ds for any t ≥ 0. Let us denote m1 := ‖ϕ‖ max 0≤t≤T M(t, 0) + n∑ i=1 min{t,g−1 i (0)}∫ 0 M(t, s)‖Bi(s)‖ds  , m2 :=  max 0≤s≤t 0≤t≤T M(t, s)  max 0≤t≤T i=0,...,n Ri(s), η(s) := ω0(s) + n∑ i=1 ω(2s). Then the last inequality implies ‖x(t)‖ ≤ m1 +m2 t∫ 0 ω0(‖x(s)‖) + n∑ i=1 ωi(‖x(gi(s))‖)ds =: z(t) ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4 532 M. MEDVED’, M. POSPÍŠIL for any 0 ≤ t ≤ T. Note that the right-hand side z(t) is nondecreasing, m1 ≥ ‖ϕ‖ and ‖x(gi(t))‖ ≤ sup 0≤s≤t ‖x(gi(s))‖ ≤ sup 0≤s≤g−1 i (0) ‖x(gi(s))‖+ sup g−1 i (0)≤s≤t ‖x(gi(s))‖ ≤ 2z(t). Consequently, ‖x(t)‖ ≤ z(t) ≤ m1 +m2 t∫ 0 ω0(z(s)) + n∑ i=1 ωi(2z(s))ds = m1 +m2 t∫ 0 η(z(s))ds for any 0 ≤ t ≤ T. The well-known Bihari inequality [1] implies Ω(‖x(t)‖) ≤ Ω(z(t)) ≤ Ω(m1) +m2t ≤ Ω(m1) +m2T, 0 ≤ t ≤ T, where Ω(z) = ∫ z 0 ds η(s) . A contradiction follows from the limit Ω(‖x(t)‖) → ∞∫ 0 du η(u) = ∞, t → T−. References 1. Bihari I.A. A generalization of a lemma of Bellman and its application to uniqueness problem of differential equation // Acta Math. Acad. Sci. Hung. — 1956. — 7. — P. 81 – 94. 2. Boichuk A. A., Medved’ M., Zhuravliov V. P. Fredholm boundary-value problems for linear delay systems defined by pairwise permutable matrices // Electron. J. Qual. Theory Different. Equat. — 2015. — 23. — P. 1 – 9. 3. Diblı́k J., Morávková B. Representation of the solutions of linear discrete systems with constant coefficients and two delays // Abstr. Appl. Anal. — 2014. — 2014. — Article ID 320476. — P. 1 – 19. 4. Hartman P. Ordinary differential equations. — New York: John Wiley & Sons, Inc., 1964. 5. Khusainov D. Ya., Shuklin G. V. Linear autonomous time-delay system with permutation matrices solving // Stud. Univ. Žilina, Math. Ser. — 2003. — 17. — P. 101 – 108. 6. Medved’ M., Pospı́šil M. Sufficient conditions for the asymptotic stability of nonlinear multidelay differential equations with linear parts defined by pairwise permutable matrices // Nonlinear Anal.-Theory Methods Appl. — 2012. — 75. — P. 3348 – 3363. 7. Medved’ M., Pospı́šil M., Škripková L. Stability and the nonexistence of blowing-up solutions of nonlinear delay systems with linear parts defined by permutable matrices // Nonlinear Anal.-Theory Methods Appl. — 2011. — 74. — P. 3903 – 3911. 8. Medved’ M., Pospı́šil M., Škripková L. On exponential stability of nonlinear fractional multidelay integro- differential equations defined by pairwise permutable matrices // Appl. Math. and Comput. — 2014. — 227. — P. 456 – 468. 9. Pospı́šil M. Representation and stability of solutions of systems of functional differential equations with multiple delays // Electron. J. Qual. Theory Different. Equat. — 2012. — 54. — P. 1 – 30. 10. Pospı́šil M., Jaroš F. On the representation of solutions of delayed differential equations via Laplace trans- form // Electron. J. Qual. Theory Different. Equat. (to appear). Received 16.05.16 ISSN 1562-3076. Нелiнiйнi коливання, 2016, т . 19, N◦ 4