Об устойчивости решений систем разностных уравнений в одном критическом случае

Рассматривается система нелинейных разностных уравнений, допускающая нулевое решение. С помощью метода функций Ляпунова изучается его устойчивость. Наряду с полной системой уравнений рассматривается линеаризованная система разностных уравнений. Известно, что если все корни характеристического уравне...

Повний опис

Збережено в:
Бібліографічні деталі
Дата:2008
Автор: Игнатьев, А.О.
Формат: Стаття
Мова:Російська
Опубліковано: Інститут прикладної математики і механіки НАН України 2008
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/10948
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:Об устойчивости решений систем разностных уравнений в одном критическом случае / А.О. Игнатьев // Український математичний вісник. — 2008. — Т. 5, № 4. — С. 488-506. — Бібліогр.: 33 назв. — рос.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859621187727917056
author Игнатьев, А.О.
author_facet Игнатьев, А.О.
citation_txt Об устойчивости решений систем разностных уравнений в одном критическом случае / А.О. Игнатьев // Український математичний вісник. — 2008. — Т. 5, № 4. — С. 488-506. — Бібліогр.: 33 назв. — рос.
collection DSpace DC
description Рассматривается система нелинейных разностных уравнений, допускающая нулевое решение. С помощью метода функций Ляпунова изучается его устойчивость. Наряду с полной системой уравнений рассматривается линеаризованная система разностных уравнений. Известно, что если все корни характеристического уравнения линеаризованной системы по модулю меньше единицы, то нулевое решение полной системы асимптотически устойчиво. Если хотя бы один из корней характеристического уравнения по модулю больше единицы, то нулевое решение полной системы неустойчиво. В случае, когда часть корней характеристического уравнения по модулю меньше единицы, а часть равна единице, задача устойчивости не решается рассмотрением лишь линейных членов, и для ее решения нужно привлечь нелинейные слагаемые. Такой случай называется критическим. В настоящей работе рассмотрен критический случай одного корня, равного единице, когда задача устойчивости решается членами до третьего порядка малости в разложении правых частей исходных уравнений в ряды Маклорена. A system of nonlinear difference equations which admits the zero solution is considered. Its stability is studied by means of Lyapunov’s direct method. Side by side with this system, a linearized system of difference equations is also considered. It is well known that if all roots of the characteristic equation of a linearized system lie within the unit circle on the complex plane, then the zero solution of the original full system is asymptotically stable. If at least one eigenvalue lies outside the unit disk, then the zero solution of the original system is unstable. In the case where the moduli of some eigenvalues are equal to unity, and the moduli of others are less than unity, the stability problem cannot be solved by considering only the linear terms. To solve this problem, it is necessary to use the terms of higher orders in expansions of the righthand sides of the original system of difference equations in Maclaurin series. Such case is called a critical one. In this paper, we consider the critical case where one eigenvalue is equal to unity, and the stability problem can be solved by involving terms up to the third order in expansions of the right-hand sides of the initial equations in Maclaurin series.
first_indexed 2025-11-29T03:24:09Z
format Article
fulltext Ukrainian Mathematical Bulletin Volume 5 (2008), № 4, 479 – 497 UMB On the stability of solutions of systems of difference equations in a critical case Aleksandr O. Ignatyev (Presented by A. E. Shishkov) Abstract. A system of nonlinear difference equations which admits the zero solution is considered. Its stability is studied by means of Lya- punov’s direct method. Side by side with this system, a linearized system of difference equations is also considered. It is well known that if all roots of the characteristic equation of a linearized system lie within the unit circle on the complex plane, then the zero solution of the original full system is asymptotically stable. If at least one eigenvalue lies outside the unit disk, then the zero solution of the original system is unstable. In the case where the moduli of some eigenvalues are equal to unity, and the moduli of others are less than unity, the stability problem cannot be solved by considering only the linear terms. To solve this problem, it is necessary to use the terms of higher orders in expansions of the right- hand sides of the original system of difference equations in Maclaurin series. Such case is called a critical one. In this paper, we consider the critical case where one eigenvalue is equal to unity, and the stabil- ity problem can be solved by involving terms up to the third order in expansions of the right-hand sides of the initial equations in Maclaurin series. 2000 MSC. 39A11, 34K20. Key words and phrases. Difference equations, Lyapunov function, stability. 1. Introduction and basic definitions The theory of discrete dynamical systems has grown tremendously in the last decade. Difference equations can arise in a number of ways. They may be the natural model of a discrete process (in combinatorics, for ex- ample), or they can be a discrete approximation of a continuous process. The development of the theory of difference systems has been strongly promoted by advanced technologies in the scientific computation and by a Received 25.07.2008 ISSN 1812 – 3309. c© Institute of Mathematics of NAS of Ukraine 480 On the stability of solutions... large number of applications to models in biology, engineering, and other sciences. For example, in papers [5,6,8,9,14], systems of difference equa- tions were used as natural models of the dynamics of populations; in [10], the difference equations were applied to a simulation in genetics; in [25], the dynamics of an ecological system was also described by a system of difference equations. The method of construction of difference schemes for systems of differential equations is proposed in [2]. This method pro- vides the consistency between differential and difference equations in the sense of the stability of the zero solution (we note that, as in the case of ordinary differential equations, the stability problem for any solution of a difference equation is reduced to that of the zero solution). Many evolution processes are characterized by the fact that, at certain time moments, they experience abruptly a change of the state. These processes are subjected to short-term perturbations, whose duration is negligible in comparison with that of the process. Consequently, it is natural to assume that these perturbations act instantaneously, that is, in the form of impulses. Papers [31, 32] were the first articles in this direction. The early works on differential equations with impulse effect were summarized in monograph [33] in which the foundations of this theory were described. In recent years, the study of impulsive systems has attracted the increasing interest [16–18, 20–22, 28]. An impulsive system consists of a continuous system which is governed by ordinary differential equations and a discrete system which is governed by difference equations. So the dynamics of impulsive systems essentially depends on properties of the corresponding difference systems, and this confirms the importance of studying the qualitative properties of difference systems. The stability of a discrete process is the ability of the process to resist the a priori unknown small influences. A process is said to be stable if such disturbances do not change it. This property turns out to be of utmost importance since an individual predictable process can be physically realized, in general, only if it is stable in the corresponding natural sense. One of the most powerful tools used in stability theory is Lyapunov’s direct method. This method consists in the use of an auxiliary function (the Lyapunov function). Consider the system of difference equations x(n + 1) = f(n, x(n)), f(n, 0) = 0, (1.1) where n = 0, 1, 2, . . . is the discrete time, x(n) = (x1(n), . . . , xk(n))T ∈ A. O. Ignatyev 481 R k, and f = (f1, . . . , fk) T ∈ R k. System (1.1) admits the trivial solution x(n) = 0. (1.2) By x(n, n0, x 0), we denote the solution of system (1.1) coinciding with x0 = (x0 1, x 0 2, . . . , x0 k) T for n = n0. Let Z+ be the set of nonnegative real integers, and let Nn0 be the set of nonnegative real integers satisfying the inequality n ≥ n0, Br = {x ∈ R k : ‖x‖ ≤ r}. By analogy with ordinary differential equations, we introduce the fol- lowing definitions. Definition 1.1. The trivial solution of system (1.1) is said to be stable if, for any ε > 0, n0 ∈ Z+, there exists δ = δ(ε, n0) > 0 such that ‖x0‖ < δ implies ‖x(n, n0, x 0)‖ < ε for n > n0. Otherwise, the trivial solution of system (1.1) is called unstable. If δ in this definition can be chosen to be independent of n0 (i.e. δ = δ(ε)), then the zero solution of system (1.1) is said to be uniformly stable. Definition 1.2. Solution (1.2) of system (1.1) is said to be attract- ing if, for any n0 ∈ Z+, there exists an η = η(n0) > 0 such that, for any ε > 0 and x0 ∈ Bη, there exists a σ = σ(ε, n0, x 0) > 0 such that ‖x(n, n0, x 0)‖ < ε for all n ≥ n0 + σ. In other words, solution (1.2) of system (1.1) is called attracting if lim n→∞ ‖x(n, n0, x 0)‖ = 0. (1.3) Definition 1.3. The trivial solution of system (1.1) is said to be uni- formly attracting if, for some η > 0 and for each ε > 0, there exists a σ = σ(ε) ∈ N such that ‖x(n, n0, x 0)‖ < ε for all n0 ∈ Z+, x0 ∈ Bη, and n ∈ Nn0+σ. In other words, solution (1.2) of system (1.1) is called uniformly at- tracting if (1.3) holds uniformly in n0 ∈ Z+, x0 ∈ Bη. Definition 1.4. The zero solution of system (1.1) is called • asymptotically stable if it is both stable and attracting; • uniformly asymptotically stable if it is both uniformly stable and uniformly attracting. 482 On the stability of solutions... Definition 1.5. The trivial solution of system (1.1) is said to be exponentially stable if there exist M > 0 and η ∈ (0, 1) such that ‖x(n, n0, x 0)‖ < M‖x0‖ηn−n0 for n ∈ Nn0 . A great number of papers is devoted to the investigation of the sta- bility of solution (1.2) of system (1.1). The general theory of differ- ence systems and the foundations of the stability theory are presented in monographs [1, 7, 11, 13, 15, 19, 29, 30]. In paper [23], it is shown that if system (1.1) is autonomous (i.e. f does not depend on n explicitly) or periodic (i.e. there exists ω ∈ N such that f(n, x) ≡ f(n + ω, x)), then the stability of solution (1.2) implies its uniform stability, and the asymp- totic stability implies its uniform asymptotic stability. The asymptotic stability of perturbed linear difference systems with periodic coefficients was studied in [3]. Papers [4, 12, 24, 26, 27] considered the stability of solutions of periodic and almost periodic systems. Let us formulate the main theorems of Lyapunov’s direct method about the stability of the zero solution of the system of autonomous difference equations x(n + 1) = f(x(n)). (1.4) These statements have been mentioned in [13, Theorems 4.20 and 4.27]. They are related to the existence of an auxiliary function V (x); and the analog of its derivative is the variation of V relative to (1.4) which is defined as ∆V (x) = V (f(x)) − V (x). Theorem A. If there exists a positive definite continuous function V (x) such that ∆V (x) relative to (1.4) is a negative semidefinite function or identically equals zero, then the trivial solution of system (1.4) is stable. Theorem B. If there exists a positive definite continuous function V (x) such that ∆V (x) relative to (1.4) is negative definite, then the trivial solution of system (1.4) is asymptotically stable. Theorem C. If there exists a continuous function V (x) such that ∆V (x) relative to (1.4) is negative definite, and the function V is not positive semidefinite, then the trivial solution of system (1.4) is unstable. Consider the autonomous system x(n + 1) = Ax(n) + X(x(n)), (1.5) where A is a k × k nonsingular matrix, and X is a function such that A. O. Ignatyev 483 lim ‖x‖→0 ‖X(x)‖ ‖x‖ = 0. (1.6) According to [13, p. 175], we denote ρ(A) = max1≤i≤k |λi|, where λi (i = 1, . . . , k) are the roots of the characteristic equation det(A − λIk) = 0. (1.7) Here, Ik is the unit k × k matrix. In [1, Corollary 5.6.3 and Theorem 5.6.4], the following theorem was proved. Theorem 1.1. If ρ(A) < 1, then the zero solution of system (1.5) is asymptotically stable (moreover, the exponential stability holds in this case). If ρ(A) > 1, then the zero solution of system (1.5) is unstable. If ρ(A) ≤ 1 and the moduli of some eigenvalues of A are equal to unity, then a function X(x) in system (1.5) can be chosen to be such that the zero solution of system (1.5) is either stable or unstable. Thus, it follows from Theorem 1.1 that the problem of stability of the zero solution of system (1.5) can be solved by means of the system of the linear approximation x(n + 1) = Ax(n) (1.8) (when ρ(A) < 1 or ρ(A) > 1). In the case ρ(A) = 1, we have a critical case where the solution of the stability problem requires to use the terms of higher degrees. For studying the stability of the zero solution of system (1.5), Elaydi [13] proposed to employ Lyapunov functions as a quadratic form V (x) = ∑ i1+i2+···+ik=2, ij≥0 (j=1,...,k) bi1,i2,...,ikxi1 1 xi2 2 . . . x ik k , (1.9) where bi1,i2,...,ik are constants. He formulated the following statement [13, Corollary 4.31] without any proof. Proposition 1.1. If ρ(A) > 1, then there exist a quadratic form V (x) which is not positive semidefinite and a negative definite quadratic form W (x) such that W (x) = V (Ax) − V (x). We will show here that Proposition 1.1 is not true and study the stability problem in the critical case where one eigenvalue of the matrix A is equal to unity. First, we show that Proposition 1.1 is false. To do 484 On the stability of solutions... this, let us consider the system x(n + 1) = Ax(n), where x = ( x1 x2 ) ∈ R 2, A = ( 1 0 0 2 ). Numbers 1 and 2 are the roots of its characteristic equation, ρ(A) = 2 > 1. But, for any quadratic form V (x) = b2,0x 2 1 + b1,1x1x2 + b0,2x 2 2, we have W (x) = V (Ax) − V (x) = b1,1x1x2 + 3b0,2x 2 2. (1.10) The quadratic form (1.10) can be neither positive definite nor negative definite. This example shows that Proposition 1.1 is not true. 2. Critical case where one eigenvalue is equal to unity In this section, we consider the critical case where one root of the char- acteristic equation (1.7) is equal to unity, i.e. we assume that Eq. (1.7) has one root λ1 = 1, and other roots satisfy the conditions |λi| < 1 (i = 2, 3, . . . , k). The function X = (X1, . . . , Xk) T is supposed to be holomor- phic, and its expansion into a Maclaurin series begins with terms of the second order of smallness. So, system (1.5) has the form xj(n + 1) = aj1x1(n) + aj2x2(n) + · · · + ajkxk(n) + Xj(x1(n), . . . , xk(n)) (j = 1, . . . , k). (2.1) Henceforth, we consider the critical case where the characteristic equation of the system in the first approximation xj(n + 1) = aj1x1(n) + aj2x2(n) + · · · + ajkxk(n) (j = 1, . . . , k) (2.2) has one root equal to unity, and other k − 1 roots have moduli which are less than unity. In system (2.2), we introduce the variable y instead of one of the variables xj by means of the substitution y = β1x1 + β2x2 + · · · + βkxk, (2.3) where βj (j = 1, . . . , k) are some constants which we choose to be such that y(n + 1) = y(n). (2.4) Relations (2.3) and (2.4) yield A. O. Ignatyev 485 y(n + 1) = β1x1(n + 1) + β2x2(n + 1) + · · · + βkxk(n + 1) = β1[a11x1(n) + a12x2(n) + · · · + a1kxk(n)] + β2[a21x1(n) + a22x2(n) + · · · + a2kxk(n)] + · · · + βk[ak1x1(n) + ak2x2(n) + · · · + akkxk(n)] = β1x1(n) + β2x2(n) + · · · + βkxk(n). Equating the coefficients corresponding to xj(n) (j = 1, 2, . . . , k), we obtain the system of linear homogeneous algebraic equations for βj (j = 1, . . . , k), a1jβ1 + a2jβ2 + · · · + akjβk = βj , (2.5) or, in the matrix form, (AT − Ik)β = 0, where β = (β1, . . . , βk) T . Since the equation det(AT − λIk) = 0 has the root λ = 1, the determinant of system (2.5) is equal to zero. Therefore, this system has a solution in which not all constants are equal to zero. To be definite, we assume that βk 6= 0. Then we can use the variable y instead of the variable xk. Other variables xj (j = 1, . . . , k − 1) are preserved without any change. Denoting cji = aji − βi βk ajk, cj = ajk βk (i, j = 1, 2, . . . , k − 1), we transform Eqs. (2.2) to the form xj(n + 1) = cj1x1(n) + cj2x2(n) + · · · + cj,k−1xk−1(n) + cjy(n) (j = 1, . . . , k − 1), (2.6) y(n + 1) = y(n), (2.7) where cji and cj are constants. The characteristic equation of system (2.6), (2.7) is reduced to two equations: λ − 1 = 0 and det(C − λIk−1) = 0, (2.8) where C = (cij) k−1 i,j=1. Since the characteristic equation is invariant with 486 On the stability of solutions... respect to linear transformations and, in this case, has k − 1 roots whose moduli are less than unity, Eq. (2.8) has k − 1 roots, and their moduli are less than unity. We denote xj = yj + ljy (j = 1, . . . , k − 1), (2.9) where lj (j = 1, . . . , k − 1) are constants which we choose to be such that the right-hand sides of system (2.6) do not contain y(n). In these designations with regard for (2.7), system (2.6) takes the form: yj(n + 1) = cj1y1(n) + cj2y2(n) + · · · + cj,k−1yk−1(n) + [cj1l1 + cj2l2 + · · · + (cjj − 1)lj + · · · + cj,k−1lk−1 + cj ]y(n) (j = 1, . . . , k − 1). We choose the constants lj to be such that cj1l1 + cj2l2 + · · ·+(cjj −1)lj + · · ·+ cj,k−1lk−1 = −cj (j = 1, . . . , k−1). (2.10) Unity is not a root of the characteristic equation (2.8); hence the determi- nant of system (2.10) is not equal to zero. Therefore, this system has the unique solution (l1, . . . , lk−1). As a result of change (2.9), system (2.6), (2.7) transforms to the form yj(n + 1) = cj1y1(n) + cj2y2(n) + · · · + cj,k−1yk−1(n) (j = 1, . . . , k − 1), y(n + 1) = y(n), and nonlinear system (2.1) takes the form yj(n + 1) = cj1y1(n) + cj2y2(n) + · · · + cj,k−1yk−1(n) + Yj(y1(n), . . . , yk−1(n), y(n)) (j = 1, . . . , k − 1), y(n + 1) = y(n) + Y (y1(n), . . . , yk−1(n), y(n)), (2.11) where Yj (j = 1, . . . , k − 1) and Y are holomorphic functions of y1, . . . , yk−1, y. Their expansions in power series lack constant and first-degree terms: Yj(y1, y2, . . . , yk−1, y) = ∞ ∑ i1+i2+···+ik−1+ik=2 v (j) i1,i2,...,ik−1,ik yi1 1 yi2 2 . . . y ik−1 k−1 yik (j = 1, . . . , k − 1), Y (y1, y2, . . . , yk−1, y) = ∞ ∑ i1+i2+···+ik−1+ik=2 vi1,i2,...,ik−1,ikyi1 1 yi2 2 . . . y ik−1 k−1 yik . A. O. Ignatyev 487 By virtue of (2.9), it is clear that the stability problem for the trivial solution of system (2.1) is equivalent to that for the zero solution of system (2.11). Further, the form of (2.11) will be basic for studying the stability of the zero solution in the case where this problem can be solved with the use of terms of the first and second orders. Theorem 2.1. If the function Y is such that the coefficient v0,0,...,0,2 is not equal to zero, then the solution y1 = 0, y2 = 0, . . . , yk−1 = 0, y = 0 of system (2.11) is unstable. Proof. Let V1(y1, . . . , yk−1) = ∑ s1+s2+···+sk−1=2 Bs1,s2,...,sk−1 ys1 1 ys2 2 . . . y sk−1 k−1 be a quadratic form such that V1(c11y1 + · · · + c1,k−1yk−1, . . . , ck−1,1y1 + · · · + ck−1,k−1yk−1) − V1(y1, . . . , yk−1) = y2 1 + y2 2 + · · · + y2 k−1. (2.12) Since the moduli of all eigenvalues of the matrix C = (cij) k−1 I,j=1 are less than unity, such quadratic form is unique according to [13, Theorem 4.30] and negative definite. Consider the Lyapunov function V (y1, . . . , yk−1, y) = V1(y1, . . . , yk−1) + αy, (2.13) where α = const. Let us find ∆V : ∆V ∣ ∣ ∣ ∣ (2.11) = ∑ s1+···+sk−1=2 Bs1,...,sk−1 {[c11y1 + · · ·+c1,k−1yk−1 +Y1(y1, . . . , yk−1, y)]s1 · · · × [ck−1,1y1 + · · · + ck−1,k−1yk−1 + Yk−1(y1, . . . , yk−1, y)]sk−1 − ys1 1 . . . y sk−1 k−1 } + αY (y1, . . . , yk−1, y). Taking (2.12) into account, ∆V can be written in the form ∆V ∣ ∣ ∣ ∣ (2.11) = W (y1, . . . , yk−1, y) + W∗(y1, . . . , yk−1, y), where 488 On the stability of solutions... W = (y2 1 + y2 2 + · · · + y2 k−1) + αv0,0,...,0,2y 2 + α(v2,0,...,0y 2 1 + v1,1,...,0y1y2 + · · · + v1,0,...,1,0y1yk−1 + v1,0,...,0,1y1y + v0,2,...,0y 2 2 + · · · + v0,0,...,1,1yk−1y), and W∗ is a holomorphic function whose Maclaurin series begins with terms of the third power in y1, . . . , yk−1, y. We choose the sign of α such that αv0,...,0,2 > 0. We now show that α can be chosen so small that the quadratic form W is positive definite. To do this, we show that α can be chosen so that the principal minors of the matrix            1 + αv2,0,...,0 1 2αv1,1,...,0 . . . 1 2αv1,0,...,1,0 1 2αv1,0,...,0,1 1 2αv1,1,...,0 1 + αv0,2,...,0 . . . 1 2αv0,1,...,1,0 1 2αv0,1,...,0,1 1 2αv1,0,1,...,0 1 2αv0,1,1,...,0 . . . 1 2αv0,0,1,...,1,0 1 2αv0,0,1,...,0,1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2αv1,0,...,1,0 1 2αv0,1,...,1,0 . . . 1 + αv0,...,0,2,0 1 2αv0,...,0,1,1 1 2αv1,0,...,0,1 1 2αv0,1,...,0,1 . . . 1 2αv0,0,...,1,1 1 2αv0,0,...,0,2            are positive. In fact, any principal minor ∆s of this matrix is a continuous function of α: ∆s = ∆s(α). Note that ∆s(0) = 1 for s = 1, 2, . . . , k − 1. Thus, there exists α∗ > 0 such that, for |α| < α∗, we have ∆s(α) ≥ 1 2 (s = 1, 2, . . . , k−1). We now prove that the inequality ∆k > 0 holds for sufficiently small |α|. To do this, we expand ∆k in terms of the elements of the last row. We obtain ∆k = 1 2αv0,0,...,0,2∆k−1 +α2∆∗, where ∆∗ is a polynomial in α and vi1,i2,...,ik (i1 + i2 + · · ·+ ik = 2, ij ≥ 0). Hence, we have ∆k > 0 for sufficiently small |α|. That is, the quadratic form W is positive definite for α, whose absolute value is small enough and whose sign coincides with the sign of v0,0,...,2. Therefore, the sum W +W∗ is also positive definite in a sufficiently small neighborhood of the origin. At the same time, the function V of form (2.13) is alternating. Hence, the zero solution of system (2.11) is unstable. The proof is completed. Thus, in the case where v0,0,...,2 6= 0, the stability problem has been solved independently of the terms, whose degrees are higher than two. Consider now the case where v0,0,...,2 = 0. We transform system (2.11) to the form where v (j) 0,0,...,2 = 0 (j = 1, 2, . . . , k − 1). We denote yj = ξj + mjy 2 (j = 1, 2, . . . , k − 1), (2.14) where mj are constants. In these designations, system (2.11) takes the A. O. Ignatyev 489 form ξj(n + 1) = cj1ξ1(n) + cj2ξ2(n) + · · · + cj,k−1ξk−1(n) + y2(n)(cj1m1 + cj2m2 + · · · + cj,k−1mk−1) + Yj(ξ1(n) + m1y 2(n), . . . , ξk−1(n) + mk−1y 2(n), y(n)) − mj [y 2(n) + 2y(n)Y (ξ1(n) + m1y 2(n), . . . , ξk−1(n) + mk−1y 2(n), y(n)) + Y 2(ξ1(n) + m1y 2(n), . . . , ξk−1(n) + mk−1y 2(n), y(n))], (2.15) y(n + 1) = y(n) + Y (ξ1(n) + m1y 2(n), . . . , ξk−1(n) + mk−1y 2(n), y(n)). (2.16) We choose constants m1, . . . , mk−1 to be such that the coefficients cor- responding to y2(n) on the right-hand sides of system (2.15) are equal to zero. Equating the corresponding coefficients to zero, we obtain the system of linear algebraic equations for m1, . . . , mk−1: cj1m1 + cj2m2 + · · ·+ cj,k−1mk−1 = mj −v (j) 0,0,...,2 (j = 1, 2, . . . , k−1). This system has a unique solution, because unity is not an eigenvalue of the matrix C. Substituting the obtained values m1, . . . , mk−1 to (2.15) and (2.16), we get the system ξj(n + 1) = cj1ξ1(n) + cj2ξ2 + · · · + cj,k−1ξk−1(n) + Ξj(ξ1(n), . . . , ξk−1(n), y(n)) (j = 1, . . . , k − 1), (2.17) y(n + 1) = y(n) + Y∗(ξ1(n), . . . , ξk−1(n), y(n)), (2.18) where Ξj(ξ1, . . . , ξk−1, y) = Yj(ξ1 + m1y 2, . . . , ξk−1 + mk−1y 2, y) − 2mjyY (ξ1 + m1y 2, . . . , ξk−1 + mk−1y 2, y) − mjY 2(ξ1 + m1y 2, . . . , ξk−1 + mk−1y 2, y) − v (j) 0,0,...,2y 2, Y∗(ξ1, . . . , ξk−1, y) = Y (ξ1 + m1y 2, . . . , ξk−1 + mk−1y 2, y). Expansions of Ξj and Y∗ in power series begin with terms of the second degree, and the coefficients corresponding to y2 in expansions of Ξj and Y∗ are equal to zero. System (2.17) and (2.18) will be basic in our further investigation of the stability of the zero solution 490 On the stability of solutions... ξ1 = 0, ξ2 = 0, . . . , ξk−1 = 0, y = 0. (2.19) By Ξ (0) j (y) (j = 1, . . . , k − 1) and Y (0) ∗ (y), we denote, respectively, the sum of all terms in the functions Ξj and Y∗ which do not include ξ1, . . . , ξk−1, so that Ξ (0) j (y) = Ξj(0, . . . , 0, y) = hjy 3 + ∞ ∑ s=4 h (s) j ys, Y (0) ∗ (y) = Y∗(0, . . . , 0, y) = hy3 + ∞ ∑ s=4 h(s)ys, where h, hj , h (s), h (s) j (j = 1, . . . , k − 1; s = 4, 5, . . . ) are constants. Theorem 2.2. Solution (2.19) of system (2.17), (2.18) is asymptotically stable for h < 0 and unstable for h > 0. Proof. We now show that there exists a Lyapunov function V such that it depends on ξ1, . . . , ξk−1, y, and ∆V is positive definite. Consider the system of linear equations ξj(n + 1) = cj1ξ(n) + cj2ξ2(n) + · · · + cj,k−1ξk−1(n) (j = 1, . . . , k − 1). (2.20) Let W = ∑ i1+···+ik−1=2 wi1,...,ik−1 ξi1 1 . . . ξ ik−1 k−1 be a quadratic form of the variables ξ1, . . . , ξk−1 and such that ∆W ∣ ∣ ∣ ∣ (2.20) = ξ2 1 + · · · + ξ2 k−1. (2.21) Since all eigenvalues of the matrix C are inside of the unit disk, the form W satisfying (2.21) exists and is unique and negative definite [13, Theorem 4.30]. If the functions Ξj (j = 1, . . . , k − 1) do not depend on y, then a variation ∆W of the function W along system (2.17), i.e. the expression ∑ i1+···+ik−1=2 wi1,··· ,ik−1 {[c11ξ1 + c12ξ2 + · · · + c1,k−1ξk−1 + Ξ1] i1 × · · · [ck−1,1ξ1 + · · · + ck−1,k−1ξk−1 + Ξk−1] ik−1 − ξi1 1 · · · ξ ik−1 k−1 }, (2.22) is a positive definite function of the variables ξ1, . . . , ξk−1 for sufficiently small ξ1, . . . , ξk−1. A. O. Ignatyev 491 On the other hand, if the function Y∗ does not depend on ξ1, . . . , ξk−1 (i.e. if Y∗ = Y (0) ∗ ), then the variation of 1 2hy2 is equal to ∆ (1 2 hy2 ) = 1 2 h [ 2yY (0) ∗ + Y (0) ∗ 2] = h2y4 + hh(4)y5 + o(y5), (2.23) and this variation is a positive definite function of y for sufficiently small |y|. Therefore, under these conditions, the variation of the function V1 = 1 2hy2 +W (ξ1, . . . , ξk−1) along the total system (2.17), (2.18) is a positive definite function of all variables ξ1, . . . , ξk−1, y in some neighborhood of the origin. In view of (2.21) and (2.23), this variation can be represented in the form (h2 + g1)y 4 + ξ2 1 + · · · + ξ2 k−1 + k−1 ∑ i,j=1 g (1) ij ξiξj , (2.24) where g1 is a holomorphic function of the variable y, vanishing for y = 0, and g (1) ij are holomorphic functions of the variables ξ1, . . . , ξk−1, vanishing for ξ1 = · · · = ξk−1 = 0. But since the functions Ξj (j = 1, . . . , k − 1) include y, and the function Y∗ includes ξ1, . . . , ξk−1, the variation of the function V1 along system (2.17), (2.18) is not, in general, positive definite. In this variation, there appear the terms breaking the positive definiteness. Note that expression (2.24) remains positive definite if the function g1 contains not only the variable y, but also the variables ξ1, . . . , ξk−1, and the functions g (1) ij contain not only variables ξ1, . . . , ξk−1, but also the variable y. It is only important that the functions g1 and g (1) ij vanish for ξ1 = · · · = ξk−1 = y = 0. Taking into account this fact, we write the second variation of the function V1 along system (2.17), (2.18) in the form ∆V1 = ∆ (1 2 hy2 ) + ∆W = hyY∗ + 1 2 hY 2 ∗ + ∑ i1+···+ik−1=2 wi1,...,ik−1 {[c11ξ1 + c12ξ2 + · · · + c1,k−1ξk−1 + Ξ1] i1 × · · · [ck−1,1ξ1 + · · · + ck−1,k−1ξk−1 + Ξk−1] ik−1 − ξi1 1 · · · ξ ik−1 k−1 } = [h2 + g1(ξ1, . . . , ξk−1, y)]y4 + ξ2 1 + · · · + ξ2 k−1 + k−1 ∑ i,j=1 g (1) ij (ξ1, . . . , ξk−1, y)ξiξj + Q(ξ1, . . . , ξk−1, y), (2.25) 492 On the stability of solutions... where the functions g1 and g (1) ij (i, j = 1, . . . , k− 1) vanish for ξ1 = · · · = ξk−1 = y = 0, and Q is the sum of all terms which can be included neither to the expression g1(ξ1, . . . , ξk−1, y)y4 (2.26) nor to the expression k−1 ∑ i,j=1 g (1) ij (ξ1, . . . , ξk−1, y)ξiξj . (2.27) All terms which are included into Q can be divided into four follow- ing groups: the terms free of ξ1, . . . , ξk−1, the terms linear with respect to ξ1, . . . , ξk−1, the terms quadratic with respect to ξ1, . . . , ξk−1, and the terms having degree higher than two with respect to ξ1, . . . , ξk−1. It is evident that all terms of the last group can be included into expres- sion (2.27); therefore, we consider only three first groups of terms. All terms free of ξ1, . . . , ξk−1 are obviously included in expressions (2.23) (where they have been written explicitly) and in ∑ i1+···+ik−1=2 wi1,...,ik−1 Ξ (0) 1 i1 · · ·Ξ (0) k−1 ik−1 (where there are summands of the sixth and higher degrees with respect to y). All these summands can be included into expression (2.26). Hence, the function Q does not include the terms free of ξ1, . . . , ξk−1. Terms linear with respect to ξ1, . . . , ξk−1 are included into expres- sion (2.25) both by means of summands from hyY∗ + 1 2hY 2 ∗ and from (2.22). If these terms have order not less than the fourth one with re- spect to y, then it is clear that they can be included into expression (2.26). Thus, the function Q has only those terms linear with respect to ξ1, . . . , ξk−1 which have degrees two and three with respect to y. Finally, consider the terms quadratic with respect to ξ1, . . . , ξk−1. If these terms have the total degree higher than two, then they can be included into expression (2.27); and, therefore, they are not included in the function Q. All quadratic terms with respect to ξ1, . . . , ξk−1 having the second degree (i.e. the terms with constant coefficients) are included into the expression ∑ i1+···+ik−1=2 wi1,··· ,ik−1 {[c11ξ1 + c12ξ2 + · · · + c1,k−1ξk−1] i1 A. O. Ignatyev 493 × · · · [ck−1,1ξ1 + · · · + ck−1,k−1ξk−1] ik−1 − ξi1 1 . . . ξ ik−1 k−1 } = ξ2 1 + · · · + ξ2 k−1 and, hence, are not included into the function Q. Thus, the function Q has the form Q = y2Q2(ξ1, . . . , ξk−1) + y3Q3(ξ1, . . . , ξk−1), (2.28) where Q2 and Q3 are linear forms with respect to ξ1, . . . , ξk−1: Q2 = q (2) 1 ξ1 + q (2) 2 ξ2 + · · · + q (2) k−1ξk−1, Q3 = q (3) 1 ξ1 + q (3) 2 ξ2 + · · · + q (3) k−1ξk−1. The presence of summand (2.28) in (2.25) breaks the positive defi- niteness of ∆V1. To get rid of the summand y2Q2(ξ1, . . . , ξk−1), we add the summand y2P2(ξ1, . . . , ξk−1) = y2(p (2) 1 ξ1 + p (2) 2 ξ2 + · · · + p (2) k−1ξk−1) to the function V1. Here, p (2) j (j = 1, . . . , k − 1) are constants. In other words, consider the function V2 = 1 2 hy2 + W (ξ1, . . . , ξk−1) + y2P2(ξ1, . . . , ξk−1) (2.29) instead of the function V1. The term y2P2(ξ1, . . . , ξk−1) brings the fol- lowing summands to ∆V2: ∆(y2P2(ξ1, . . . , ξk−1)) = [y2 + 2yY∗(ξ1, . . . , ξk−1, y) + Y 2 ∗ (ξ1, . . . , ξk−1, y)] × k−1 ∑ j=1 p (2) j [cj,1ξ1 + cj,2ξ2 + · · · + cj,k−1ξk−1 + Ξj(ξ1, . . . , ξk−1, y)] −y2[p (2) 1 ξ1 + p (2) 2 ξ2 + · · · + p (2) k−1ξk−1] = y2   k−1 ∑ j=1 p (2) j (cj1ξ1 + cj2ξ2 + · · · + cj,k−1ξk−1 − ξj)  +G(ξ1, . . . , ξk−1, y), where the function G is the sum of summands every of which can be included either to expression (2.26) or to (2.27). Let us choose constants p (2) 1 , . . . , p (2) k−1 such that the equality k−1 ∑ j=1 p (2) j (cj1ξ1 + cj2ξ2 + · · · + cj,k−1ξk−1 − ξj) = − k−1 ∑ j=1 q (2) j ξj (2.30) 494 On the stability of solutions... holds. To do this, let us equate the coefficients corresponding to ξj (j = 1, . . . , k − 1) on the right- and left-hand sides of equality (2.30). We obtain the system of linear equations for p (2) j (j = 1, . . . , k − 1): c1jp (2) 1 + c2jp (2) 2 + · · · + (cjj − 1)p (2) j + · · · + ck−1,jp (2) k−1 = −q (2) j (j = 1, . . . , k − 1). (2.31) The determinant of this system is not equal to zero, because all eigen- values of C are inside the unit disk. Therefore, system (2.31) has the unique solution. Substituting the obtained values p (2) 1 , . . . , p (2) k−1 in the expression P2(ξ1, . . . , ξk−1), we get ∆V2 = [h2 + g2(ξ1, . . . , ξk−1, y)]y4 + (ξ2 1 + · · · + ξ2 k−1) + k−1 ∑ i,j=1 g (2) ij (ξ1, . . . , ξk−1, y)ξiξj + y3Q3(ξ1, . . . , ξk−1), (2.32) where g2 and g (2) ij are functions vanishing for ξ1 = ξ2 = · · · = ξk−1 = y = 0. Similarly, we can show that it is possible to get rid of the summand y3Q3(ξ1, . . . , ξk−1) in expression (2.32). To do this, all we need is to add the summand y3P3(ξ1, . . . , ξk−1) = y3(p (3) 1 ξ1 + p (3) 2 ξ2 + · · · + p (3) k−1ξk−1) to the function V2, where p (3) j (j = 1, . . . , k − 1) are constants. In other words, consider the function V = 1 2 hy2 + W (ξ1, . . . , ξk−1) + y2P2(ξ1, . . . , ξk−1) + y3P3(ξ1, . . . , ξk−1) (2.33) instead of the function V2. Its variation along system (2.17), (2.18) is equal to ∆V = [h2 + g(ξ1, . . . , ξk−1, y)]y4 + (ξ2 1 + · · · + ξ2 k−1) + k−1 ∑ i,j=1 gij(ξ1, . . . , ξk−1, y)ξiξj , (2.34) where g and gij are functions vanishing for ξ1 = ξ2 = · · · = ξk−1 = y = 0. It follows from (2.34) that ∆V is positive definite in a sufficiently small neighborhood of the origin, and the function V of form (2.33) is A. O. Ignatyev 495 negative definite for h < 0 and changes its sign for h > 0. Hence, according to Theorems B and C, we can conclude that solution (2.19) of system (2.17), (2.18) is asymptotically stable for h < 0 and unstable for h > 0. This completes the proof. Remark 2.1. Obviously, substitutions (2.3), (2.9), and (2.14) are such that the investigation of the stability of solution (2.19) of system (2.17), (2.18) is equivalent to the investigation of the stability of the zero solution of system (2.1). Remark 2.2. In Theorems 2.1 and 2.2, there are the conditions under which the problem of the stability of the zero solution of system (2.1) can be solved in the critical case where one eigenvalue of the linearized system is equal to unity. The obtained criteria do not depend on nonlinear terms with degrees of smallness more than three. If we obtain h = 0, then the stability problem cannot be solved by terms of the first, second, and third degrees of smallness in the expansions of the right-hand sides of the system of difference equations. To solve this problem in this case, it is necessary to consider also the terms of higher degrees. References [1] R. P. Agarwal, Difference Equations and Inequalities, Marcel Dekker, New York, 1992. [2] A. Yu. Aleksandrov, A. P. Zhabko, On stability of solutions to one class of non- linear difference systems // Siberian Math. J., 44:6 (2003), 951–958. [3] K. Audin, H. Bulgak, and G. V. Demidenko, Asymptotic stability of solutions to perturbed linear difference equations with periodic coefficients // Siberian Math. J., 43:3 (2002), 389–401. [4] K. Audin, H. Bulgak and G. V. Demidenko, Numerical characteristics for asymp- totic stability of solutions to linear difference equations with periodic coeffi- cients // Siberian Math. J., 41:6 (2000), 1005–1014. [5] F. Brauer, C. Castillo-Chavez, Mathematical Models in Population Biology and Epidemiology, Springer, New York, 2001. [6] E. Braverman, On a discrete model of population dynamics with impulsive har- vesting or recruitment // Nonlinear Analysis, 63:5–7 (2005), e751–e759. [7] P. V. Bromberg, Matrix Methods in Theory of Relay and Impulse Regulation, Nauka, Moscow, 1967 (in Russian). [8] M. L. Castro, J. A. L. Silva, D. A. R. Justo, Stability in an age-structured metapop- ulation model // J. Math. Biol., 52:2 (2006), 183–208. [9] F. Chen, Permanence and global attractivity of a discrete multispecies Lotka– Volterra competition predador-prey systems // Applied Mathematics and Com- putation, 182:1 (2006), 3–12. [10] R. Continho, B. Fernandez, R. Lima, A. Meyroneinc, Discrete time piecewise affine models of genetic regulatory networks // J. Math. Biol., 52:4 (2006), 524– 570. 496 On the stability of solutions... [11] P. Cull, M. Flahive, R. Robson, Difference Equations. From Rabbits to Chaos, Springer, New York, 2005. [12] S. M. Dobrovol’skii and A. V. Rogozin, The direct Lyapunov method for an almost periodic difference system on a compact // Siberian Math. J., 46:1 (2005), 77–82. [13] S. Elaydi, An Introduction to Difference Equations, Springer, New York, 2005. [14] J. E. Franke, A. A. Yakubu, Globally attracting attenuant versus resonant cycles in periodic compensatory Leslie models // Math. Biosci., 204:1 (2006), 1–20. [15] I. V. Gaishun, Systems with Discrete Time, Inst. of Math. of the NAS of Belarus’, Minsk, 2001. [16] R. I. Gladilina, A. O. Ignatyev, On the necessary and sufficient conditions of the asymptotic stability for impulsive systems // Ukr. Math. J., 55:8 (2003), 1035– 1043. [17] W. M. Haddad, V. Chellaboina, S. G. Nersesov, Hybrid Nonnegative and Com- partmental Dynamical Systems // Math. Problems in Engin., 8:3, 493–515. [18] W. M. Haddad, V. Chellaboina, S. G. Nersesov, Impulsive and Hybrid Dynamical Systems, Princeton Univ. Press, Princeton and Oxford, 2006. [19] A. Halanay, D. Wexler, Qualitative Theory of Impulsive Systems, Edit. Acad., Bucharest, 1968 (in Romanian); Mir, Moscow, 1971 (in Russian). [20] A. O. Ignatyev, Method of Lyapunov functions in problems of stability of solutions of systems of differential equations with impulse action // Sbornik: Mathematics, 194:10 (2003), 1543–1558. [21] A. O. Ignatyev, Asymptotic stability and instability with respect to part of variables for solutions to impulsive systems // Siberian Math. J., 49:1 (2008), 102–108. [22] A. O. Ignatyev, O. A. Ignatyev, A. A. Soliman, On the asymptotic stability and instability of solutions of systems with impulse effect // Math. Notes, 80:4 (2006), 516–525. [23] A. O. Ignatyev, O. A. Ignatyev, On the stability in periodic and almost periodic difference systems // Journal of Mathematical Analysis and Applications, 313:2 (2006), 678–688. [24] A. O. Ignatyev, O. A. Ignatyev, On the stability of discrete systems // in Integral Methods in Science and Engineering, Birkhäuser, Boston, (2006), pp. 105–116. [25] V. A. A. Jansen, A. L. Lloyd, Local stability analysis of spatially homogeneous solutions of multi-patch systems // J. Math. Biol., 41:3 (2000), 232-252. [26] O. V. Kirichenova, On stability of solutions to nonlinear almost periodic systems of difference equations // Siberian Math. J., 39:1 (1998), 39–41. [27] O. V. Kirichenova, A. S. Kotyurgina and R. K. Romanovskii, The method of Lyapunov functions for systems of linear difference equations with almost periodic coefficients // Siberian Math. J., 37:1 (1996), 147–150. [28] V. Lakshmikantham, D. D. Bainov, P. S. Simeonov, Theory of Impulsive Differ- ential Equations, World Scientific, Singapore, 1989. [29] V. Lakshmikantham, D. Trigiante, Theory of Difference Equations: Numerical Methods and Applications, Marcel Dekker, New York, 2002. [30] D. I. Martynyuk, Lectures in Qualitative Theory of Difference Equations, Naukova Dumka, Kiev, 1972 (in Russian). A. O. Ignatyev 497 [31] V. D. Mil’man, A. D. Myshkis, On the stability of motion under the presence of pushes // Sibirskii Matematicheskii Zhurnal, 1:2 (1960), 233–237 (in Russian). [32] A. D. Myshkis and A. M. Samoilenko, Systems with pushes at fixed instants // Matematicheskii sbornik, 74:2 (1967), 202–208 (in Russian). [33] A. M. Samoilenko, N. A. Perestyuk, Impulsive Differential Equations, World Sci- entific, Singapore, 1995. Contact information Aleksandr Olegovich Ignatyev Institute of Applied Mathematics and Mechanics of NASU, 74, R. Luxemburg Str., Donetsk 83114, Ukraine E-Mail: ignat@iamm.ac.donetsk.ua, aoignat@mail.ru
id nasplib_isofts_kiev_ua-123456789-10948
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1810-3200
language Russian
last_indexed 2025-11-29T03:24:09Z
publishDate 2008
publisher Інститут прикладної математики і механіки НАН України
record_format dspace
spelling Игнатьев, А.О.
2010-08-10T10:14:36Z
2010-08-10T10:14:36Z
2008
Об устойчивости решений систем разностных уравнений в одном критическом случае / А.О. Игнатьев // Український математичний вісник. — 2008. — Т. 5, № 4. — С. 488-506. — Бібліогр.: 33 назв. — рос.
1810-3200
https://nasplib.isofts.kiev.ua/handle/123456789/10948
Рассматривается система нелинейных разностных уравнений, допускающая нулевое решение. С помощью метода функций Ляпунова изучается его устойчивость. Наряду с полной системой уравнений рассматривается линеаризованная система разностных уравнений. Известно, что если все корни характеристического уравнения линеаризованной системы по модулю меньше единицы, то нулевое решение полной системы асимптотически устойчиво. Если хотя бы один из корней характеристического уравнения по модулю больше единицы, то нулевое решение полной системы неустойчиво. В случае, когда часть корней характеристического уравнения по модулю меньше единицы, а часть равна единице, задача устойчивости не решается рассмотрением лишь линейных членов, и для ее решения нужно привлечь нелинейные слагаемые. Такой случай называется критическим. В настоящей работе рассмотрен критический случай одного корня, равного единице, когда задача устойчивости решается членами до третьего порядка малости в разложении правых частей исходных уравнений в ряды Маклорена.
A system of nonlinear difference equations which admits the zero solution is considered. Its stability is studied by means of Lyapunov’s direct method. Side by side with this system, a linearized system of difference equations is also considered. It is well known that if all roots of the characteristic equation of a linearized system lie within the unit circle on the complex plane, then the zero solution of the original full system is asymptotically stable. If at least one eigenvalue lies outside the unit disk, then the zero solution of the original system is unstable. In the case where the moduli of some eigenvalues are equal to unity, and the moduli of others are less than unity, the stability problem cannot be solved by considering only the linear terms. To solve this problem, it is necessary to use the terms of higher orders in expansions of the righthand sides of the original system of difference equations in Maclaurin series. Such case is called a critical one. In this paper, we consider the critical case where one eigenvalue is equal to unity, and the stability problem can be solved by involving terms up to the third order in expansions of the right-hand sides of the initial equations in Maclaurin series.
ru
Інститут прикладної математики і механіки НАН України
Об устойчивости решений систем разностных уравнений в одном критическом случае
On the stability of solutions of systems of difference equations in a critical case
Article
published earlier
spellingShingle Об устойчивости решений систем разностных уравнений в одном критическом случае
Игнатьев, А.О.
title Об устойчивости решений систем разностных уравнений в одном критическом случае
title_alt On the stability of solutions of systems of difference equations in a critical case
title_full Об устойчивости решений систем разностных уравнений в одном критическом случае
title_fullStr Об устойчивости решений систем разностных уравнений в одном критическом случае
title_full_unstemmed Об устойчивости решений систем разностных уравнений в одном критическом случае
title_short Об устойчивости решений систем разностных уравнений в одном критическом случае
title_sort об устойчивости решений систем разностных уравнений в одном критическом случае
url https://nasplib.isofts.kiev.ua/handle/123456789/10948
work_keys_str_mv AT ignatʹevao obustoičivostirešeniisistemraznostnyhuravneniivodnomkritičeskomslučae
AT ignatʹevao onthestabilityofsolutionsofsystemsofdifferenceequationsinacriticalcase