Integral manifolds for semilinear evolution equations and admissibility of function spaces

We prove the existence of integral (stable, unstable, and center) manifolds for the solutions to a semilinear integral equation in the case where the evolution family (U(t, s)) t≥s has an exponential trichotomy on a half line or on the whole line, and the nonlinear forcing term f satisfies the φ-Li...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Український математичний журнал
Datum:2012
Hauptverfasser: Vu Thi Ngoc Ha, Nguyen Thieu Huy, Ha Phi
Format: Artikel
Sprache:English
Veröffentlicht: Інститут математики НАН України 2012
Schlagworte:
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/164414
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:Integral manifolds for semilinear evolution equations and admissibility of function spaces / Vu Thi Ngoc Ha, Nguyen Thieu Huy, Ha Phi // Український математичний журнал. — 2012. — Т. 64, № 6. — С. 772-796. — Бібліогр.: 37 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-164414
record_format dspace
spelling Vu Thi Ngoc Ha
Nguyen Thieu Huy
Ha Phi
2020-02-09T14:52:06Z
2020-02-09T14:52:06Z
2012
Integral manifolds for semilinear evolution equations and admissibility of function spaces / Vu Thi Ngoc Ha, Nguyen Thieu Huy, Ha Phi // Український математичний журнал. — 2012. — Т. 64, № 6. — С. 772-796. — Бібліогр.: 37 назв. — англ.
1027-3190
https://nasplib.isofts.kiev.ua/handle/123456789/164414
517.9
We prove the existence of integral (stable, unstable, and center) manifolds for the solutions to a semilinear integral equation in the case where the evolution family (U(t, s)) t≥s has an exponential trichotomy on a half line or on the whole line, and the nonlinear forcing term f satisfies the φ-Lipschitz conditions, i.e., where φ(t) belongs to some classes of admissible function spaces. Our main method is based on the Lyapunov–Perron methods, rescaling procedures, and the techniques of using the admissibility of function spaces.
Доведено iснування iнтегральних (стiйких, нестiйких, центральних) многовидiв для розв’язкiв напiвлiнiйного iнтегрального рiвняння у випадку, коли сiм’я еволюцiй (U(t,s))tleqs має експоненцiальну трихотомiю на пiвосi або на всiй осi, а нелiнiйний збурюючий член f задовольняє φ-лiпшицевi умови, тобто належить до деяких класiв допустимих просторiв функцiй. Наш основний метод базується на методах Ляпунова – Перрона, процедурах перемасштабування та технiцi застосування допустимостi просторiв функцiй.
On leave from Hanoi University of Science and Technology as a research fellow of the Alexander von Humboldt Foundation at Technical University of Darmstadt. The support by the Alexander von Humboldt Foundation is gratefully acknowledged. The author thanks Prof. Matthias Hieber for his strong support and inspiration. This work is financially supported by the Vietnamese National Foundation for Science and Technology Development (NAFOSTED) under Project 101.01-2011.25.
en
Інститут математики НАН України
Український математичний журнал
Статті
Integral manifolds for semilinear evolution equations and admissibility of function spaces
Інтегральнi многовиди для напiвлiнiйних еволюцiйних рiвнянь та допустимiсть просторiв функцiй
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Integral manifolds for semilinear evolution equations and admissibility of function spaces
spellingShingle Integral manifolds for semilinear evolution equations and admissibility of function spaces
Vu Thi Ngoc Ha
Nguyen Thieu Huy
Ha Phi
Статті
title_short Integral manifolds for semilinear evolution equations and admissibility of function spaces
title_full Integral manifolds for semilinear evolution equations and admissibility of function spaces
title_fullStr Integral manifolds for semilinear evolution equations and admissibility of function spaces
title_full_unstemmed Integral manifolds for semilinear evolution equations and admissibility of function spaces
title_sort integral manifolds for semilinear evolution equations and admissibility of function spaces
author Vu Thi Ngoc Ha
Nguyen Thieu Huy
Ha Phi
author_facet Vu Thi Ngoc Ha
Nguyen Thieu Huy
Ha Phi
topic Статті
topic_facet Статті
publishDate 2012
language English
container_title Український математичний журнал
publisher Інститут математики НАН України
format Article
title_alt Інтегральнi многовиди для напiвлiнiйних еволюцiйних рiвнянь та допустимiсть просторiв функцiй
description We prove the existence of integral (stable, unstable, and center) manifolds for the solutions to a semilinear integral equation in the case where the evolution family (U(t, s)) t≥s has an exponential trichotomy on a half line or on the whole line, and the nonlinear forcing term f satisfies the φ-Lipschitz conditions, i.e., where φ(t) belongs to some classes of admissible function spaces. Our main method is based on the Lyapunov–Perron methods, rescaling procedures, and the techniques of using the admissibility of function spaces. Доведено iснування iнтегральних (стiйких, нестiйких, центральних) многовидiв для розв’язкiв напiвлiнiйного iнтегрального рiвняння у випадку, коли сiм’я еволюцiй (U(t,s))tleqs має експоненцiальну трихотомiю на пiвосi або на всiй осi, а нелiнiйний збурюючий член f задовольняє φ-лiпшицевi умови, тобто належить до деяких класiв допустимих просторiв функцiй. Наш основний метод базується на методах Ляпунова – Перрона, процедурах перемасштабування та технiцi застосування допустимостi просторiв функцiй.
issn 1027-3190
url https://nasplib.isofts.kiev.ua/handle/123456789/164414
citation_txt Integral manifolds for semilinear evolution equations and admissibility of function spaces / Vu Thi Ngoc Ha, Nguyen Thieu Huy, Ha Phi // Український математичний журнал. — 2012. — Т. 64, № 6. — С. 772-796. — Бібліогр.: 37 назв. — англ.
work_keys_str_mv AT vuthingocha integralmanifoldsforsemilinearevolutionequationsandadmissibilityoffunctionspaces
AT nguyenthieuhuy integralmanifoldsforsemilinearevolutionequationsandadmissibilityoffunctionspaces
AT haphi integralmanifoldsforsemilinearevolutionequationsandadmissibilityoffunctionspaces
AT vuthingocha íntegralʹnimnogovididlânapivliniinihevolûciinihrivnânʹtadopustimistʹprostorivfunkcii
AT nguyenthieuhuy íntegralʹnimnogovididlânapivliniinihevolûciinihrivnânʹtadopustimistʹprostorivfunkcii
AT haphi íntegralʹnimnogovididlânapivliniinihevolûciinihrivnânʹtadopustimistʹprostorivfunkcii
first_indexed 2025-11-27T04:38:59Z
last_indexed 2025-11-27T04:38:59Z
_version_ 1850800516776329216
fulltext UDC 517.9 Nguyen Thieu Huy∗ (Hanoi Univ. Sci. and Technology, Vietnam; Techn. Univ. Darmstadt, Germany), Vu Thi Ngoc Ha (Hanoi Univ. Sci. and Technology, Vietnam), Ha Phi (Hanoi Univ. Education, Vietnam) INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY OF FUNCTION SPACES∗∗ IНТЕГРАЛЬНI МНОГОВИДИ ДЛЯ НАПIВЛIНIЙНИХ ЕВОЛЮЦIЙНИХ РIВНЯНЬ ТА ДОПУСТИМIСТЬ ПРОСТОРIВ ФУНКЦIЙ We prove the existence of integral (stable, unstable, center) manifolds for the solutions to the semilinear integral equation u(t) = U(t, s)u(s) + ∫ t s U(t, ξ)f(ξ, u(ξ))dξ in the case where the evolution family (U(t, s))t≥s has an exponential trichotomy on a half-line or on the whole line, and the nonlinear forcing term f satisfies the ϕ-Lipschitz conditions, i.e., ‖f(t, x)− f(t, y)‖ 6 ϕ(t)‖x− y‖, where ϕ(t) belongs to some classes of admissible function spaces. Our main method invokes the Lyapunov – Perron methods, rescaling procedures, and the techniques of using the admissibility of function spaces. Доведено iснування iнтегральних (стiйких, нестiйких, центральних) многовидiв для розв’язкiв напiвлiнiйного iн- тегрального рiвняння u(t) = U(t, s)u(s) + ∫ t s U(t, ξ)f(ξ, u(ξ))dξ у випадку, коли сiм’я еволюцiй (U(t, s))t≥s має експоненцiальну трихотомiю на пiвосi або на всiй осi, а нелiнiйний збурюючий член f задовольняє ϕ-лiпшицевi умови, тобто ‖f(t, x) − f(t, y)‖ 6 ϕ(t)‖x − y‖, де ϕ(t) належить до деяких класiв допустимих просторiв функ- цiй. Наш основний метод базується на методах Ляпунова – Перрона, процедурах перемасштабування та технiцi застосування допустимостi просторiв функцiй. 1. Introduction and preliminaries. Consider the semilinear evolution equation of the form dx dt = A(t)x(t) + f(t, x(t)), t ∈ J, (1.1) where J is a subinterval of the real line R; each A(t) is a (possibly unbounded) linear operator acting in a Banach space X , x(t) ∈ X, and f(·, ·) : J × X → X is a nonlinear operator. When the linear part (i.e., the equation dx/dt = A(t)x(t)) of the above equation has an exponential dichotomy (or trichotomy), one shall try to find conditions imposed on the nonlinear forcing term f such that the equation (1.1) has an integral manifold (e.g., a stable, unstable, or center manifold). Such early results can be traced back to Hadamard [10], Perron [29, 30], Bogoliubov and Mitropolsky [4, 5] for the case of matrix coefficients A(t), to Daleckii and Krein [8] for the case of bounded coefficients acting on Banach spaces, and to Henry [12] for the case of unbounded coefficients. At this point, we would like to quote the sentence by Anosov [1]: ,,Every five years or so, if not more often, some one ,,discovers” the theorem of Hadamard and Perron, proving it by Hadamard’s method of proof or by Perron’s”. ∗ On leave from Hanoi University of Science and Technology as a research fellow of the Alexander von Humboldt Foundation at Technical University of Darmstadt. The support by the Alexander von Humboldt Foundation is gratefully acknowledged. The author thanks Prof. Matthias Hieber for his strong support and inspiration. ∗∗ This work is financially supported by the Vietnamese National Foundation for Science and Technology Development (NAFOSTED) under Project 101.01-2011.25. c© NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI, 2012 772 ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 773 The Hadamard’s method is generalized to the so-called graph transform method which has been used, e.g., in the works [2, 13, 22] to prove the existence of invariant manifolds. This method is more far-reaching and related to complicated choices of the transforms between graphs represent- ing the involved manifolds. Meanwhile, the Perron’s method is now extended to the well-known Lyapunov – Perron method aimed at the construction of the so-called Lyapunov – Perron equations (or operators) involving the differential equations under consideration to show the existence of the integral manifolds. It seems to be more natural to use the Lyapunov – Perron method to handle with the flows or semiflows which are generated by semilinear evolution equations since in this case it is relatively simple to construct such Lyapunov – Perron equations or operators. We refer the reader to [3, 7, 8, 11, 12, 15, 16, 35] and reference therein for more information on the matter. To our best knowledge, the most popular conditions for the existence of invariant manifolds are the exponential dichotomy (or trichotomy) of the linear part dx dt = A(t)x and the uniform Lipschitz continuity of the nonlinear part f(t, x) with sufficiently small Lipschitz constants (i.e., ‖f(t, x) − − f(t, y)‖ 6 q‖x− y‖ for q small enough). The purpose of this paper is establishing the existence of stable, unstable, and center-stable manifolds when the linear part of equation (1.1) has an exponential trichotomy on the half-line or on the whole line under more general conditions on the nonlinear term f(t, x), that is the non-uniform Lipschitz continuity of f, i.e., ‖f(t, x) − f(t, y)‖ 6 ϕ(t)‖x − y‖ for ϕ being a real and positive function which belongs to admissible function spaces defined in Definition 2.3 below. Under some conditions on ϕ, we will prove the existence of center manifolds for the equation (1.1) provided that the linear part dx dt = A(t)x has an exponential trichotomy. Our method is to transform to the case of exponential dichotomy by some rescaling procedures, and then applying our techniques and results in [15] where we have used the Lyapunov – Perron method and the characterization (obtained in [14]) of the exponential dichotomy of evolution equations in admissible spaces of functions defined on the half-line R+ to construct the structures of solutions of the equation (1.1) in a mild form, which belong to some certain classes of admissible spaces on which we could implement some well-known procedures in functional analysis such as: constructing of contraction mapping; using of Implicite Function Theorem, etc. The use of admissible spaces has helped us to construct the invariant manifolds for equation (1.1) in the case of dichotomic linear parts without using the smallness of Lipschitz constants of nonlinear forcing terms in classical sense. Instead, the “smallness” is understood as the sufficient smallness of supt≥0 ∫ t+1 t ϕ(τ)dτ (see the conditions in Theorem 4.7 in [15]). Consequently, we have obtained the existence of invariant-stable manifolds for the case of dichotomic linear parts under very general conditions on the nonlinear term f(t, x) (see [15]). Using these results and rescaling procedures we shall prove, in the present paper, the existence of center manifolds for the mild solutions of the equation (1.1) in the case of trichotomic linear parts under the same conditions on the nonlinear term f(t, x) as in [15]. Moreover, using the same method we can also obtain the existence of unstable and center-unstable manifolds in the case of dichotomic and trichotomic linear parts (respectively) for the evolution equations defined on the whole line. Our main results are contained in Theorems 3.1, 4.1, 4.2, and Corollaries 4.1, 4.2, 4.3. We also illustrate our results in the Examples 5.2, 5.3. We now recall some notions. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 774 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI Definition 1.1. Let J be one of the following intervals: R+ or R. A family of operators {U(t, s)}t≥s, t,s∈J acting on a Banach space X is a (strongly continuous, exponential bounded) evolution family on J if (i) U(t, t) = Id and U(t, r)U(r, s) = U(t, s) for all t ≥ r ≥ s and t, s, r ∈ J, (ii) the map (t, s) 7→ U(t, s)x is continuous on J for every x ∈ X, (iii) ‖U(t, s)x‖ 6 Keω(t−s)‖x‖ for all t ≥ s, t, s, r ∈ J, and x ∈ X, for some constants K, ω. The notion of an evolution family arises naturally from the theory of evolution equation which are well-posed. Meanwhile, if the abstract Cauchy problem du(t) dt = A(t)u(t), t ≥ s, t, s ∈ J, u(s) = xs ∈ X, (1.2) is well-posed, there exists an evolution family (U(t, s))t≥s, t,s∈J such that the solution of the prob- lem (1.2) is given by u(t) = U(t, s)u(s). For more details on the notion and some problems focus on properties and applications of evo- lution family we refer the reader to Pazy [28], Henry [12], and Nagel and Nickel [9]. For a given evolution family, we have the following concept of an exponential trichotomy of evolution families on J as follows. Definition 1.2. Let J be one of the following intervals: R+ or R. A given evolution family (U(t, s))t≥s, t,s∈J on J is said to have an exponential trichotomy on J if there are three families of projections (Pj(t))t∈J, j = 1, 2, 3, positive constants N, α, β with α < β such that the following conditions are satisfied: (i) supt∈J ‖Pj(t)‖ <∞, j = 1, 2, 3, (ii) P1(t) + P2(t) + P3(t) = Id for all t ∈ J, and Pj(t)Pi(t) = 0 for all j 6= i, (iii) Pj(t)U(t, s) = U(t, s)Pj(s), for all t ≥ s ≥ 0, j = 1, 2, 3, (iv) U(t, s)|ImPj(s) are homeomorphisms from ImPj(s) onto ImPj(t) for all t ≥ s, t, s ∈ J, and j = 2, 3, respectively; also we denote the inverse of U(t, s)|ImP2(s) by U(s, t)| (here s 6 t), (v) the following estimates hold: ‖U(t, s)P1(s)x‖ 6 Ne−β(t−s)‖P1(s)x‖, ‖U(s, t)|P2(t)x‖ 6 Ne−β(t−s)‖P2(t)x‖, ‖U(t, s)P3(s)x‖ 6 Neα(t−s)‖P3(s)x‖, for all t ≥ s, t, s ∈ J, x ∈ X. The evolution family is said to have an exponential dichotomy on J if it has an exponential trichotomy for which the family of projections P3(t) is trivial, i.e., P3(t) = 0 for all t ∈ J. In this case, we remark that the property (i) is a consequence of other properties (see [21], Lemma 4.2), and we also denote P (t) := P1(t) called dichotomy projections. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 775 2. Function spaces and admissibility. We recall some notions of function spaces and admissibility. We refer the readers to Massera and Schäffer [20] (Chapter 2) for wide classes of function spaces that play a fundamental role throughout the study of differential equations in the case of bounded coefficients A(t) (see also Räbiger and Schnaubelt [31] (§ 1) for some classes of admissible Banach function spaces of functions defined on the whole line R). Denote by B the Borel algebra and by λ the Lebesgue measure on R+. As already known, the set of real-valued Borel-measurable functions on R+ (modulo λ-nullfunctions) that are integrable on every compact subinterval J ⊂ R+ becomes, with the topology of convergence in the mean on every such J, a locally convex topological vector space, which we denote by L1,loc(R+). A set of seminorms defining the topology of L1,loc(R+) is given by pn(f) := ∫ Jn |f(t)|dt, n ∈ N, where {Jn}n∈N = {[n, n+1]}n∈N is a countable set of abutting compact intervals whose union is R+. With this set of seminorms one can see (see [20], Chapter 2, § 20) that L1,loc(R+) is a Fréchet space. Let V be a normed space (with norm ‖ · ‖V ) and W be a locally convex Hausdorff topological vector space. Then, we say that V is stronger than W if V ⊆W and the indentity map from V into W is continuous. The latter condition is equivalent to the fact that for each continuous seminorm π of W there exists a number βπ > 0 such that π(x) 6 βπ‖x‖V for all x ∈ V. We write V ↪→ W to indicate that V is stronger than W. If, in particular, W is also a normed space (with norm ‖ · ‖W ) then the relation V ↪→ W is equivalent to the fact that V ⊆ W and there is a number α > 0 such that ‖x‖W 6 α‖x‖V for all x ∈ V (see [20], Chapter 2 for detailed discussions on this matter). We can now define Banach function spaces as follows. Definition 2.1. A vector space E of real-valued Borel-measurable functions on R+ (modulo λ-nullfunctions) is called a Banach function space (over (R+,B, λ) if (1) E is Banach lattice with respect to a norm ‖ · ‖E , i.e., (E, ‖ · ‖E) is a Banach space, and if ϕ ∈ E and ψ is a real-valued Borel-measurable function such that |ψ(·)| 6 |ϕ(·)| λ-a.e., then ψ ∈ E and ‖ψ‖E 6 ‖ϕ‖E , (2) the characteristic functions χA belong to E for all A ∈ B of finite measure, and supt≥0 ‖χ[t,t+1]‖E <∞ and inft≥0 ‖χ[t,t+1]‖E > 0, (3) E ↪→ L1,loc(R+). For a Banach function space E we remark that the condition (3) in the above definition means that for each compact interval J ⊂ R+ there exists a number βJ ≥ 0 such that ∫ J |f(t)|dt 6 βJ‖f‖E for all f ∈ E. We state the following trivial lemma which will be frequently used in our strategy. Lemma 2.1. Let E be a Banach function space. Let ϕ and ψ be real-valued, measurable functions on R+ such that they coincide with each other outside a compact interval and they are essentially bounded (in particular, continuous) on this compact interval. Then ϕ ∈ E if and only if ψ ∈ E. We then define Banach spaces of vector-valued functions corresponding to Banach function spaces as follows. Definition 2.2. Let E be a Banach function space and X be a Banach space endowed with the norm ‖ · ‖. We set ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 776 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI E := E(R+, X) := { f : R+ → X : f is strongly measurable and ‖f(·)‖ ∈ E } (modullo λ-nullfunctions) endowed with the norm ‖f‖E := ‖‖f(·)‖‖E . One can easily see that E is a Banach space. We call it the Banach space corresponding to the Banach function space E. We now introduce the notion of admissibility in the following definition. Definition 2.3. The Banach function space E is called admissible if it satisfies (i) there is a constant M ≥ 1 such that for every compact interval [a, b] ∈ R+ we have b∫ a |ϕ(t)|dt 6 M(b− a) ‖χ[a,b]‖E ‖ϕ‖E for all ϕ ∈ E, (2.1) (ii) for ϕ ∈ E the function Λ1ϕ defined by Λ1ϕ(t) := ∫ t+1 t ϕ(τ)dτ belongs to E, (iii) E is T+ τ -invariant and T−τ -invariant, where T+ τ and T−τ are defined, for τ ∈ R+, by T+ τ ϕ(t) := ϕ(t− τ) for t ≥ τ ≥ 0, 0 for 0 6 t 6 τ, T−τ ϕ(t) := ϕ(t+ τ) for t ≥ 0. (2.2) Moreover, there are constants N1, N2 such that ‖T+ τ ‖ 6 N1, ‖T−τ ‖ 6 N2 for all τ ∈ R+. Example 2.1. Besides the spaces Lp(R+), 1 6 p 6∞, and the space M(R+) := f ∈ L1,loc(R+) : sup t≥0 t+1∫ t |f(τ)|dτ <∞  endowed with the norm ‖f‖M := supt≥0 ∫ t+1 t |f(τ)|dτ, many other function spaces occuring in in- terpolation theory, e.g. the Lorentz spaces Lp,q, 1 < p <∞, 1 6 q <∞ (see [6, p. 284], Theorem 3, [36]) and, more general, the class of rearrangement invariant function spaces over (R+,B, λ) (see [17]) are admissible. Remark 2.1. If E is an admissible Banach function space then E ↪→ M(R+). Indeed, put β := inft≥0 ‖χ[t,t+1]‖E > 0 (by Definition 2.1 (2)). Then, from Definition 2.3 (i) we derive t+1∫ t |ϕ(τ)|dτ 6 M β ‖ϕ‖E for all t ≥ 0 and ϕ ∈ E. (2.3) Therefore, if ϕ ∈ E then ϕ ∈M(R+) and ‖ϕ‖M 6 M β ‖ϕ‖E . We thus obtain E ↪→M(R+). We now collect some properties of admissible Banach function spaces in the following proposi- tion (see [14], Proposition 2.6 and originally in [20]). ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 777 Proposition 2.1. Let E be an admissible Banach function space. Then the following assertions hold. (a) Let ϕ ∈ L1,loc(R+) such that ϕ ≥ 0 and Λ1ϕ ∈ E, where, Λ1 is defined as in Defini- tion 2.3(ii). For σ > 0 we define functions Λ′σϕ and Λ′′σϕ by Λ′σϕ(t) := t∫ 0 e−σ(t−s)ϕ(s)ds, Λ′′σϕ(t) := ∞∫ t e−σ(s−t)ϕ(s)ds. Then Λ′σϕ and Λ′′σϕ belong to E. In particular, if supt≥0 ∫ t+1 t ϕ(τ)dτ < ∞ (this will be satisfied if ϕ ∈ E (see Remark 2.1)) then Λ′σϕ and Λ′′σϕ are bounded. Moreover, denoted by ‖ · ‖∞ for ess sup-norm, we have ‖Λ′σϕ‖∞ 6 N1 1− e−σ ‖Λ1T + 1 ϕ‖∞ and ‖Λ′′σϕ‖∞ 6 N2 1− e−σ ‖Λ1ϕ‖∞ (2.4) for operator T+ 1 and constants N1, N2 defined as in Definition 2.3. (b) E contains exponentially decaying functions ψ(t) = e−αt for t ≥ 0 and any fixed constant α > 0. (c) E does not contain exponentially growing functions f(t) := ebt for t ≥ 0 and any fixed constant b > 0. Remark 2.2. If we replace the half-line R+ by any infinite (or half-infinite) interval I (e.g., I = R−,R, or any (−∞, t0] for fixed t0 ∈ R, etc.), then we have the similar notions of admissible spaces on the interval I with slight changes as follow: (1) In Definition 2.3, the translations semigroups T+ τ and T−τ for τ ∈ R+ should be replaced by T+ τ and T−τ defined for τ ∈ I as T+ τ ϕ(t) := ϕ(t− τ) for t and t− τ belonging to I, 0 for t ∈ I but t− τ /∈ I, T−τ ϕ(t) := ϕ(t+ τ) for t ∈ I. (2.5) (2) In Proposition 2.1 (a), the functions Λ′σ and Λ′′σ should be replaced by Λ′σϕ(t) := t0∫ t e−σ|t−s|ϕ(s)ds, here t0 =∞ if I = R, and t0 = 0 if I = R−, Λ′′σϕ(t) := t∫ −∞ e−σ|s−t|ϕ(s)ds. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 778 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI (3) In Proposition 2.1 (b) and (c) the functions ψ(t) = e−αt (t ≥ 0, and fixed α > 0) should be replaced by ψ(t) = e−α|t|, t ∈ I and fixed α > 0; and the functions f(t) := ebt for t ≥ 0 and any fixed constant b > 0 should be replaced by f(t) := eb|t|, t ∈ I and fixed b > 0. These notions will be used in Section 5. We denote the admissible function space of the functions defined on I by EI. If I = R+, we denote simply E := ER+ . For a function ϕ defined on the whole line we denote the restriction of ϕ on I by ϕ|I. It is obvious that, if the function ϕ ∈ ER, then ϕ|I ∈ EI. In the case of infinite-dimensional phase spaces, instead of equation (1.1), for an evolution family (U(t, s))t≥s, t,s∈J where J = R+or R, we consider the integral equation u(t) = U(t, s)u(s) + t∫ s U(t, ξ)f(ξ, u(ξ))dξ for a.e. t ≥ s, t, s ∈ J. (2.6) We note that, if the evolution family (U(t, s))t≥s, t,s∈J arises from the well-posed Cauchy prob- lem (1.2) then the function u, which satisfies (2.6) for some given function f, is called a mild solution of the inhomogeneous problem du(t) dt = A(t)u(t) + f(t, u(t)), t ≥ s, t, s ∈ J, u(s) = xs ∈ X. We refer the reader to Pazy [28] for more detailed treatment on the relations between classical and mild solutions of evolution equations (see also [9, 18, 35]). To obtain the existence of an integral manifold for equation (2.6), beside the exponential di- chotomy (or trichotomy) of the evolution family, we also need the properties of (local) ϕ-Lipschitz of the nonlinear term f in the following definitions in which we suppose as above that J is one of the infinite intervals R+ or R. Also, we let EJ be an admissible Banach function space on J. When J = R+, we simply write E instead of ER+ . Definition 2.4 (Local ϕ-Lipschitz functions). Let ϕ be a positive function belonging to EJ, and Bρ be the ball with radius ρ in X, i.e., Bρ := {x ∈ X : ‖x‖ 6 ρ}. A function f : J×Bρ → X is said to be local ϕ-Lipschitz of the class (M,ϕ, ρ) for some positive constants M, ρ if f satisfies (i) ‖f(t, x)‖ 6Mϕ(t) for a.e. t ∈ J and all x ∈ Bρ, (ii) ‖f(t, x1)− f(t, x2)‖ 6 ϕ(t)‖x1 − x2‖ for a.e. t ∈ J and all x1, x2 ∈ Bρ. Remark 2.3. If f(t, 0) = 0 then, the condition (ii) in the above definition already implies that f belongs to class (ρ, ϕ, ρ). We next recall the definition of ϕ-Lipschitz functions. Definition 2.5 (ϕ-Lipschitz functions). Let ϕ be a positive function belongs to EJ. A function f : J× X→ X is said to be ϕ-Lipschitz if f satisfies (i) f(t, 0) = 0 for a.e. t ∈ J, (ii) ‖f(t, x1)− f(t, x2)‖ 6 ϕ(t)‖x1 − x2‖ for a.e. t ∈ J and all x1, x2 ∈ X. 3. Exponential trichotomy and center-stable manifolds on R+. In this section, we will gen- eralize Theorem 4.7 in [15] to the case that the evolution family (U(t, s))t≥s≥0 has an exponential trichotomy on R+ and the nonlinear forcing term f is ϕ-Lipschitz. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 779 In this case, the interval J = R+. For an evolution family (U(t, s))t≥s≥0 we rewrite the integral equation (2.6) for J = R+ as u(t) = U(t, s)u(s) + t∫ s U(t, ξ)f(ξ, u(ξ))dξ for a.e. t ≥ s, t, s ∈ R+. (3.1) Precisely, we will prove that there exists a center-stable manifold for the solutions of equation (3.1). Theorem 3.1. Let the evolution family (U(t, s))t≥s≥0 have an exponential trichotomy with the corresponding constants N, α, β (α < β), and projections (Pj(t))t≥0, j = 1, 2, 3, given in Definition 1.2. Suppose that f : R+ × X → X be ϕ-Lipschitz, where ϕ is the positive function belonging to E such that k < min { 1 N + 1 , 1− eα−β 1− e−β } , here k is defined by k := (1 +H)N 1− e−β ( N1‖Λ1T + 1 ϕ‖∞ +N2‖Λ1ϕ‖∞ ) . (3.2) Then there exists a center-stable manifold C = { (t,Ct) | t ∈ R+ and Ct ⊂ X } for the solutions of equation (3.1), with the family (Ct)t≥0 being the graphs of the family of Lipschitz continuous mappings (gt)t≥0 (i.e., Ct := graph(gt) = { x + gtx | x ∈ Im(P1(t) + P3(t)) } for each t ≥ 0) where gt : Im (P1(t) + P3(t)) → ImP2(t) has the Lipschitz constant Nk 1− k independent of t, such that the following properties hold: (i) to each x0 ∈ Ct0 there corresponds one and only one solution u(t) of equation (3.1) on [t0,∞) and it satisfies u(t0) = x0 and ess supt≥t0 ‖e −γtu(t)‖ <∞, where γ := α+ β 2 , (ii) Ct is homeomorphism to X1(t)⊕X3(t) for all t ≥ 0, where X1(t) = P1(t)X, X3(t) = P3(t)X, (iii) C is invariant under the equation (3.1) in the sense that, if u(t) is the solution of equa- tion (3.1) satisfying u(t0) = x0 ∈ Ct0 and ess supt≥t0 ‖e −γtu(t)‖ < ∞, then u(s) ∈ Cs for all s ≥ t0, (iv) every two solutions u1(t), u2(t) on the center-stable manifold C satisfy the condition that there exist positive constants µ and Cµ independent of t0 ≥ 0 such that ‖x(t)− y(t)‖ 6 Cµe (γ−µ)(t−t0) ∥∥(P1(t0) + P3(t0))x(t0)− (P1(t0) + P3(t0))y(t0) ∥∥ for all t ≥ t0. Proof. Set P (t) := P1(t) + P3(t) and Q(t) := P2(t) = I − P (t). We consider the following rescaling evolution family: Ũ(t, s)x := e−γ(t−s)U(t, s)x for all t ≥ s ≥ 0, x ∈ X, where γ := α+ β 2 . It is easy to check that (Ũ(t, s))t≥s≥0 is an evolution family on X. We now claim that (Ũ(t, s))t≥s≥0 has an exponential dichotomy with the projection P (t) and Q(t) on the half-line. Infact, it suffices to verify the estimates in Definition 1.2. By the definition of exponential trichotomy we have ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 780 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI ‖Ũ(s, t)Q(t)x‖ 6 Ne−(β−γ)(t−s)‖Q(t)x‖ = Ne− (β−α) 2 (t−s)‖Q(t)x‖ for all t ≥ s ≥ 0, x ∈ X. On the other hand, ‖Ũ(t, s)P (s)x‖ = e−γ(t−s)‖U(t, s)[P1(s) + P3(s)]x‖ 6 6 Ne−(γ+α)(t−s)‖P1(s)x‖+ e−(γ−α)(t−s)‖P3(s)x‖ 6 6 Ne− (β−α) 2 (t−s)(‖P1(s)x‖+ ‖P3(s)x‖) = = Ne− (β−α) 2 (t−s)(‖P1(s)(P1(s) + P3(s))x‖+ ‖P3(s)(P1(s) + P3(s))x‖ ) 6 6 NHe− (β−α) 2 (t−s)(‖(P1(s) + P3(s))x‖+ ‖(P1(s) + P3(s))x‖ ) = = 2NHe− (β−α) 2 (t−s)‖P (s)x‖ for all t ≥ s ≥ 0, x ∈ X (here we use the fact that H := supt≥0{‖P1(t)‖, ‖P2(t)‖, ‖P3(t)‖} <∞). We finally obtain the following estimate: ‖Ũ(t, s)P (s)x‖ 6 2NHe− (β−α) 2 (t−s)‖P (s)x‖ for all t ≥ s ≥ 0, x ∈ X. Therefore, (Ũ(t, s))t≥s≥0 has an exponential dichotomy with the projections (P (t))t≥0 and the di- chotomy constants N ′ := max{N, 2NH}, β′ = β − α 2 > 0. Put x̃(t) := e−γtx(t), and define the mapping F as follows: F : R+ × X→ X, F (t, x) = e−γtf(t, eγtx) for all t ≥ 0, x ∈ X. We can easily verify that the operator F is also ϕ-Lipschitz. Thus, we can rewrite the equation (3.1) in the new form x̃(t) = Ũ(t, s)x̃(s) + t∫ s Ũ(t, ξ)F (ξ, x̃(ξ))dξ for a.e. t ≥ s ≥ 0. (3.3) Hence, by [15] (Theorem 4.7), we obtain that, if k = (1 +H)N(N1‖Λ1T + 1 ϕ‖∞ +N2‖Λ1ϕ‖∞) 1− e−β < 1 1 +N , then there exists an invariant stable manifold C for the solutions of equation (3.3). Return to equa- tion (3.1) by using the relation x(t) := eγtx̃(t) we can easily verify the properties of C which are stated in (i), (ii), (iii), and (iv). Thus, C is an invariant center-stable manifold for the solutions of equation (3.1). Theorem 3.1 is proved. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 781 Remark 3.1. In case the evolution family has an exponential trichotomy and the nonlinear term f satisfies the local ϕ-Lipschitz of the class (M,ϕ, ρ) with f(t, 0) = 0 and the positive function ϕ ∈ E satisfying k < min { ρ 2M , 1 N + 1 , 1− eα−β 1− e−β } (here k is defined as in (3.2)), then by the similar ways as above and using the results in [15] (Theorem 3.8) we can obtain the existence of a local center-stable manifold for the solutions of equation (3.1), that is a set C ⊂ R+ ×X such that there exist positive constants ρ, ρ0, ρ1 and a family of Lipschitz continuous mappings gt : Bρ0 ∩ Im(P1(t) + P3(t))→ Bρ1 ∩ ImP2(t), t ∈ R+, with Lipschitz constants independent of t satisfying: (i) C = {(t, x+gt(x)) ∈ R+× ( Im(P1(t)+P3(t))⊕ ImP2(t) ) | t ∈ R+, x ∈ Bρ0 ∩ Im(P1(t)+ + P3(t))}, and we denote by Ct := {x+ gt(x) | (t, x+ gt(x)) ∈ C}, (ii) Ct is homeomorphic to Bρ0 ∩ Im(P1(t) +P3(t)) = {x ∈ Im(P1(t) +P3(t)) | ‖x‖ 6 ρ0} for all t ≥ 0, (iii) to each x0 ∈ Ct0 there corresponds one and only one solution u(t) of equation (3.1) on [t0,∞) and it satisfies u(t0) = x0 and ess supt≥t0 ‖e −γtu(t)‖ <∞, where γ := α+ β 2 , (iv) every two solutions u1(t), u2(t) on the local center-stable manifold C satisfy the condition that there exist positive constants µ and Cµ independent of t0 ≥ 0 such that ‖x(t)− y(t)‖ 6 Cµe (γ−µ)(t−t0) ∥∥(P1(t0) + P3(t0))x(t0)− (P1(t0) + P3(t0))y(t0) ∥∥ (3.4) for all t ≥ t0. 4. Unstable manifolds for equations defined on the whole line. We now consider the case that the evolution family (U(t, s))t≥s and the nonlinear forcing term f are defined on the whole line (i.e., the case J = R). That is to say, we will consider the integral equation x(t) = U(t, s)x(s) + t∫ s U(t, ξ)f(ξ, x(ξ))dξ for a.e. t ≥ s, t, s ∈ R. (4.1) As in Section 1, the solutions of the equation (4.1) is called the mild solutions of the equation dx dt = A(t)x+ f(t, x), t ∈ R, x ∈ X, (4.2) where A(t), t ∈ R (in general case), are unbounded operators in X, which are coefficients of a well-posed Cauchy problem du(t) dt = A(t)u(t), t ≥ s, u(s) = xs ∈ X, whose solutions are given by x(t) = U(t, s)x(s) as mentioned in Section 1. In this case, the exis- tences of (local- or invariant-) stable manifolds on R are defined and proved by the same way as in the case of equations defined on a half-line R+ (see [15], Theorem 4.7). Therefore, we will pay our attention to the case of the unstable manifolds which are defined below. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 782 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI 4.1. Local-unstable manifolds on R. We shall prove the existence of the local-unstable mani- fold under the conditions that the evolution family (U(t, s))t≥s has an exponential dichotomy and the nonlinear term f is local ϕ-Lipschitz of the class (M,ϕ, ρ) for a relevant positive function ϕ ∈ ER. We now give the description of a local-unstable manifold for the solutions of the integral equa- tion (4.1) in the following definition in which we remind that by Br we denote the ball in X with radius r centered at 0, i.e., Br = {x ∈ X | ‖x‖ 6 r}. Definition 4.1. A set U ⊂ R × X is said to be a local-unstable manifold for the solutions of equation (4.1) if for every t ∈ R the phase spaces X splits into a direct sum X = X1(t)⊕ X2(t) such that inf t∈R+ Sn(X1(t),X2(t)) := inf t∈R+ inf { ‖x1 + x2‖ : xi ∈ Xi(t), ‖xi‖ = 1, i = 1, 2 } > 0, and if there exist positive constants ρ, ρ0, ρ1 and a family of Lipschitz continuous mappings ht : Bρ0 ∩ X2(t)→ Bρ1 ∩ X1(t), t ∈ R, with the Lipschitz constants independent of t such that (i) U = {(t, x + ht(x)) ∈ R × (X2(t) ⊕ X1(t)) | x ∈ Bρ0 ∩ X2(t)}, and we denote by Ut := := {x+ ht(x) | (t, x+ ht(x)) ∈ U}, (ii) Ut is homeomorphic to Bρ0 ∩ X2(t) for all t ∈ R, (iii) to each x0 ∈ Ut0 there corresponds one and only one solution x(t) of equation (4.1) satisfying the conditions x(t0) = x0 and ess supt6t0 ‖x(t)‖ 6 ρ. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projection P (t), t ∈ R, and the dichotomy constants N, β > 0. Then, we can define the Green’s function as follows: G(t, τ) := P (t)U(t, τ) for t ≥ τ, −U(t, τ)|[I − P (τ)] for t < τ. (4.3) Thus, we have ‖G(t, τ)‖ 6 (1 +H)Ne−β|t−τ | for all t 6= τ, where H = sup t∈R ‖P (t)‖ <∞. We now prove the existence of a local-unstable manifold. To do that, we first construct the form of the solutions of the equation (4.1) which are bounded on the half-line (−∞, t0]. We denote by ‖ · ‖∞ the sup-norm on the half-line (−∞, t0]. Lemma 4.1. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that ϕ is the positive function which belongs to ER. Let f : R×Bρ → X be local ϕ-Lipschitz of the class (M,ϕ, ρ) for some positive constants M, ρ. Let x(t) be a solution of (4.1) such that ess supt6t0 ‖x(t)‖ 6 ρ for some fixed t0. Then, for t 6 t0, we have that x(t) can be rewritten in the form x(t) = U(t, t0)|v + t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0, (4.4) and some v ∈ X2(t0) = (I − P (t0))X, where G(t, τ) is the Green’s function defined above. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 783 Proof. Let y(t) := t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0. (4.5) Then the function y(·) is bounded. Indeed, by the estimates of the Green’s function G and the function f we have ‖y(·)‖∞ 6 t0∫ −∞ (1 +H)Ne−β|t−τ |‖f(τ, x(τ))‖dτ 6 6 (1 +H)NM  t∫ −∞ e−β(t−τ)‖ϕ(τ)‖dτ + t0∫ t eβ(t−τ)‖ϕ(τ)‖dτ  6 6 (1 +H)NM [ N1‖Λ1ϕ‖∞ +N2‖Λ1T + 1 ϕ‖∞ 1− e−β ] <∞. Next, by computing directly we verify that y(·) satisfies the integral equation y(t0) = U(t0, t)y(t) + t0∫ t U(t0, τ)f(τ, x(τ))dτ for all t 6 t0. (4.6) Indeed, subtituting y from (4.5) to the right-hand side of (4.6) we obtain U(t0, t)y(t) + t0∫ t U(t0, τ)f(τ, x(τ))dτ = = U(t0, t) t0∫ −∞ G(t, τ)f(τ, x(τ))dτ + t0∫ t U(t0, τ)f(τ, x(τ))dτ = = U(t0, t) t∫ −∞ U(t, τ)P (τ)f(τ, x(τ))dτ− −U(t0, t) t0∫ t U(t, τ)|(I − P (τ))f(τ, x(τ))dτ + t0∫ t U(t0, τ)f(τ, x(τ))dτ = = t∫ −∞ U(t0, τ)P (τ)f(τ, x(τ))dτ− ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 784 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI − t0∫ t U(t0, t)U(t, τ)|(I − P (τ))f(τ, x(τ))dτ + t0∫ t U(t0, τ)f(τ, x(τ))dτ = = t0∫ −∞ U(t0, τ)P (τ)f(τ, x(τ))dτ = t0∫ −∞ G(t0, τ)f(τ, x(τ)) = y(t0), here we use the fact U(t0, t)U(t, τ)|(I − P (τ)) = U(t0, τ)(I − P (τ)) that for all t 6 τ 6 t0. Thus, we have y(t0) = U(t0, t)y(t) + t0∫ t U(t0, τ)f(τ, x(τ))dτ. On the other hand, x(t0) = U(t0, t)x(t) + t0∫ t U(t0, τ)f(τ, x(τ))dτ. Then x(t0)− y(t0) = U(t0, t)[x(t)− y(t)]. We need to prove that x(t0)− y(t0) ∈ (I − P (t0))X. Applying the operator P (t0) to the expression x(t0)− y(t0) = U(t0, t)[x(t)− y(t)], we have ‖P (t0)[x(t0)− y(t0)]‖ = ‖U(t0, t)P (t)[x(t)− y(t)]‖ 6 Ne−β(t0−t)‖P (t)‖.‖x(t)− y(t)‖. Since supt∈R ‖P (t)‖ < ∞ and ‖x(t) − y(t)‖ 6 ‖x(·)‖∞ + ‖y(·)‖∞ < ∞, letting t → −∞ we obtain that ‖P (t0)[x(t0)− y(t0)]‖ = 0. It means that, v := x(t0)− y(t0) ∈ (I − P (t0))X = X2(t0) finishing the proof. Remark 4.1. By computing directly, we can see that the converse of Lemma 4.1 is also true. It means, all solutions of equation (4.4) satisfied the equation (4.1) for t 6 t0. Lemma 4.2. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that ϕ is the positive function which belongs to E. Put k := (1 +H)N 1− e−β [ N1‖Λ1ϕ‖∞ +N2‖Λ1T + 1 ϕ‖∞ ] . (4.7) Let f : R× Bρ → X be local ϕ-Lipschitz of the class (M,ϕ, ρ) such that k < min { 1, ρ 2M } . Then there corresponds to each v ∈ Bρ/2N ∩ X2(t0) one and only one solution x(t) of the equation (4.1) on (−∞, t0] satisfying the conditions that (I − P (t0))x(t0) = v and ess supt6t0 ‖x(t)‖ 6 ρ. Proof. We consider in the space L∞((−∞, t0], X) the ball Bρ := { x(·) ∈ L∞((−∞, t0], X) : ‖x(·)‖∞ := ess sup t6t0 ‖x(t)‖ 6 ρ } . For v ∈ Bρ/2N ∩X2(t0) we will prove the transformation T defined by ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 785 (Tx)(t) = U(t, t0)|v + t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0, acts from Bρ into Bρ and is a contraction. In fact, for x(·) ∈ Bρ we have that ‖f(t, x(t))‖ 6Mϕ(t). Therefore, putting y(t) = U(t, t0)|v + t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0, we obtain that ‖y(t)‖ 6 Ne−β(t0−t)‖v‖ + (1 + H)NM ∫ t0 −∞ e−β|t−τ |ϕ(τ)dτ. It follows from the admissibility of L∞ that, y(·) ∈ L∞ and ‖y(·)‖∞ 6 N‖v‖+ (1 +H)NM 1− e−β ( N1‖Λ1T + 1 ϕ‖∞ +N2‖Λ1ϕ‖∞ ) . Using now the fact that ‖v‖ 6 ρ 2N and (1 +H)N 1− e−β (N1‖Λ1T + 1 ϕ‖∞ +N2‖Λ1ϕ‖∞) < ρ 2M , we have that ‖y(·)‖∞ 6 ρ. Therefore, the transformation T acts from Bρ to Bρ. It follows from the estimates of G and U(t, s) that ‖T (x)− T (y)‖∞ 6 6 (1 +H)N‖x(·)− y(·)‖∞ 1− e−β [ N1‖Λ1ϕ‖∞ +N2‖Λ1T + 1 ϕ‖∞ ] = k‖x(·)− y(·)‖∞. Since k < 1, we obtain that T is a contraction. By the Banach contraction mapping theorem, the lemma follows. From Lemmas 4.1, 4.2 and using the same arguments as in [15] (Theorem 3.8) we obtain the existence of an unstable manifold in the following theorem. Theorem 4.1. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Then, for any ρ > 0 and M > 0, we have that, if f is local ϕ-Lipschitz of the class (M,ϕ, ρ) with the positive function ϕ ∈ ER such that k < min { ρ 2M , 1 N + 1 } , here k is defined as in 4.7, there exists a local unstable manifold for the solutions of equation (4.1). Moreover, for any two solution x1(·) and x2(·) belonging to this manifold we have∥∥x1(t)− x2(t)∥∥ 6 Ceµ(t−t0) ∥∥(I − P (t0))x1(t0)− (I − P (t0))x2(t0) ∥∥ for all t 6 t0, (4.8) where C, µ be the positive constants independent of t0, x1(·) and x2(·). ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 786 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI Proof. The proof of this theorem can be done by the same way as in [15] (Theorem 3.8) replacing R+ by R and using the structures of bounded solutions as in Lemmas 4.1, 4.2. We just note that the family of Lipschitz mappings (ht)t∈R determining the local-unstable manifold is define by ht : Bρ/2N ∩X2(t)→ Bρ/2 ∩X1(), ht(y) = t∫ −∞ G(t, s)f(s, x(s))ds for y ∈ Bρ/2N ∩ X2(t), where x(·) is the unique solution in L∞((−∞, t], X) of equation (4.1) on (−∞, t] satisfying (I − P (t))x(t) = y (note that the existence and uniqueness of x(·) is obtained in Lemma 4.2). Furthermore, the Lipschitz constant of ht is kN 1− k < 1 which is the same as that of gt determining the local-stable manifold (see [15], Theorem 3.8). Theorem 4.1 is proved. From the existence of the local-stable and local-unstable manifolds of equation (4.1) defined on the whole line we have the following important corollary which describes the geometric picture of solutions to equation (4.1). Corollary 4.1. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Then, for any ρ > 0 and M > 0, we have that, if f is local ϕ-Lipschitz of the class (M,ϕ, ρ) with the positive function ϕ ∈ ER such that k < min { ρ 2M , 1 N + 1 , ρ 2MN } , here k is defined as in (4.7), then there exist a local-stable manifold S and a local-unstable manifold U for the solutions of equation (4.1) having the following properties: (a) for each t0 the intersection St0 ∩Ut0 contains the unique element zt0 , (b) the solution u0(t) of equation (4.1) with initial condition u0(t0) = zt0 is bounded on the whole line R, (c) the solutions u(t) of equation (4.1) satisfying u(t0) ∈ St0 exponentially approach u0(t) as t→∞, (d) the solutions u(t) of equation (4.1) satisfying u0(t) ∈ Ut0 exponentially approach u0(t) as t→ −∞. Proof. (a) The condition that x ∈ St0 ∩Ut0 is equivalent to the fact that there are w ∈ Bρ0 ∩ ∩X1(t0) and y ∈ Bρ0 ∩X2(t0) such that x = w+ gt0w = ht0y + y where gt0 and ht0 are members of the families of Lipschitz continuous mappings (gt)t∈R determining S and (ht)t∈R determining U, respectively. Then w − ht0y = y − gt0w ∈ X1(t0) ∩X2(t0) = {0}. This follows that w = ht0y and y = gt0w. Therefore, w = ht0(gt0w) = (ht0 ◦ gt0)w. We now estimate gt0w for w ∈ Bρ0 ∩X1(t0) by using the formula (see [15], equation (18)) gt0(w) = ∞∫ t0 G(t0, s)f(s, x(s))ds, (4.9) where w ∈ Bρ/2N∩X1(t0) and x(·) is the unique solution in Bρ of equation (4.1) on [t0,∞) satisfying P (t0)x(t0) = w (note that the existence and uniqueness of x(·) is obtained in [15] (Theorem 3.7)). ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 787 By (4.9) we have that ‖gt0(w)‖ 6 ∞∫ 0 ‖G(t0, s)‖‖f(s, x(s))‖ds 6 (1 +H)NM ∞∫ 0 e−|t0−s|ϕ(s)ds 6 6 (1 +H)NM 1− e−β (N1‖Λ1T + 1 ϕ‖∞ +N2‖Λ1ϕ‖∞) = kM < ρ 2N ( since k < ρ 2MN ) . Therefore, we obtain that gt0 : Bρ/2N ∩X1(t0)→ Bρ/2N ∩X2(t0). Similarly, we have ht0 : Bρ/2N ∩ ∩X2(t0)→ Bρ/2N ∩X1(t0). This follows that ht0 ◦ gt0 : Bρ/2N ∩X1(t0)→ Bρ/2N ∩X1(t0). Since the mappings gt0 and ht0 are both Lipschitz continuous with the same Lipschitz constant kN 1− k < 1 (see the proof of [15] (Theorem 3.8)), we obtain that ht0 ◦gt0 is a contraction. Thus, there exists a unique w0 such that w0 = (ht0 ◦ gt0)w0. Putting zt0 = w0 + gt0w0 we obtain that zt0 is the unique element of the intersection St0 ∩Ut0 . The property (b) follows from the definitions of the local-stable and local-unstable manifolds, respectively. The properties (c) and (d) are consequences of the inequalities in [15] (Theorem 3.7, ineq. (13)) and (4.8), respectively. 4.2. Invariant unstable manifolds on R. In this subsection we consider the existence of the invariant unstable manifold under the conditions that the evolution family has an exponential di- chotomy, and the nonlinear term f is ϕ-Lipschitz continuous. We now give the definition of an invariant unstable manifold for the solutions of the integral equation (4.1). Definition 4.2. A set S ⊂ R× X is said to be an invariant unstable manifold for the solutions of equation (4.1) if for every t ∈ R the phase spaces X splits into a direct sum X = X1(t) ⊕ X2(t) such that inf t∈R+ Sn(X1(t),X2(t)) := inf t∈R+ inf { ‖x1 + x2‖ : xi ∈ Xi(t), ‖xi‖ = 1, i = 1, 2 } > 0, and if there exists a family of Lipschitz continuous mappings gt : X2(t)→ X1(t), t ∈ R, with the Lipschitz constants independent of t such that (i) S = {(t, x+ gt(x)) ∈ R× (X2(t)⊕X1(t)) | x ∈ X2(t)}, and we denote by St := {x+ gt(x) | (t, x+ gt(x)) ∈ S}, (ii) St is homeomorphic to X2(t) for all t ∈ R, (iii) to each x0 ∈ St0 there corresponds one and only one solution x(t) of equation (4.1) satisfying the conditions x(t0) = x0 and ess supt6t0 ‖x(t)‖ <∞, (iv) S is invariant under the equation (4.1) in the sense that, if x(·) is a solution of equation (4.1) satisfying x(t0) ∈ St0 and ess supt6t0 ‖x(t)‖ <∞, then x(t) ∈ St for all t 6 t0. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 788 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI As in the previous subsection, we can construct the form of the solutions of equation (4.1) which are bounded on the half-line (−∞, t0] in the following lemma whose proof can be done by the same way as in Lemma 4.1. Lemma 4.3. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that ϕ is the positive function which belongs to ER. Let f : R× X→ X be ϕ-Lipschitz. Let x(t) be a solution of (4.1) such that ess supt6t0 ‖x(t)‖ <∞ for some fixed t0. Then, for t 6 t0, we have that x(t) can be rewritten in the form x(t) = U(t, t0)|v + t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0, (4.10) and some v ∈ X2(t0) = (I − P (t0))X, where G(t, τ) is the Green’s function defined above. Remark 4.2. By computing directly, we can see that the converse of Lemma 4.3 is also true. It means, all solutions of equation (4.10) satisfied the equation (4.1) for t 6 t0. Similarly to Lemma 4.2 we have the following lemma which describes the existence and unique- ness of certain bounded solutions. Lemma 4.4. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that ϕ is the positive function which belongs to E. Let f : R × X → X be ϕ-Lipschitz satisfying k < < 1, where k is defined as in (4.7). Then there corresponds to each v ∈ X2(t0) one and only one solution x(t) of the equation (4.1) on (−∞, t0] satisfying the condition (I − P (t0))x(t0) = v and ess supt6t0 ‖x(t)‖ <∞. Proof. For each t0 ∈ R, v ∈ X2(t0) we consider the operator T : L∞((−∞, t0],X)→ L∞((−∞, t0],X), x 7→ (Tx)(t) = U(t, t0)|v + t0∫ −∞ G(t, τ)f(τ, x(τ))dτ for all t 6 t0. It follows from the estimates of G and U(t, s) that ‖T (x)− T (y)‖∞ 6 6 (1 +H)N‖x(·)− y(·)‖∞ 1− e−β [ N1‖Λ1ϕ‖∞ +N2‖Λ1T + 1 ϕ‖∞ ] = k‖x(·)− y(·)‖∞. Since k < 1, we obtain that T is a contraction. By the Banach contraction mapping theorem, the lemma follows. From Lemmas 4.3, 4.4 and using the same arguments as in [15] (Theorem 4.7) we obtain the existence of an invariant unstable manifold in the following theorem. Theorem 4.2. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that f : R × X → X be ϕ-Lipschitz, where ϕ is the positive function which belongs to ER such that ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 789 k < 1, here k defined as in (4.7). Then there exists an invariant unstable manifold for the solutions of equation (4.1). Moreover, for any two solution x1(·) and x2(·) belonging to this unstable manifold we have ‖x1(t)− x2(t)‖ 6 Ceµ(t−t0)‖(I − P (t0))x1(t0)− (I − P (t0))x2(t0)‖ for all t 6 t0, where C, µ be the positive constants independent of t0, x1(·) and x2(·). Proof. The proof of this theorem can be done by the same way as in [15] (Theorem 4.7) replacing R+ by R and using the structures of bounded solutions as in Lemmas 4.3, 4.4. We just note that the family of Lipschitz mappings (gt)t∈R determining the unstable manifold is define by gt : X2(t)→ X1(t), gt(y) = t∫ −∞ G(t, s)f(s, x(s))ds for y ∈ X2(t), where x(·) is the unique solution in L∞((−∞, t], X) of equation (4.1) on (−∞, t] sat- isfying (I−P (t))x(t) = y (note that the existence and uniqueness of x(·) is obtained in Lemma 4.4). Theorem 4.2 is proved. Using now the similar arguments as in Corollary 4.1, we easily obtain the following corollary which describes the relations of solutions of equation (4.1) with initial values lying on the invariant stable or unstable manifolds and the solution lying on the intersection of the two manifolds. Corollary 4.2. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, β > 0. Suppose that f is ϕ-Lipschitz with the positive function ϕ ∈ ER such that k < 1 N + 1 , here k defined as in (4.7). Then there exist an invariant stable manifold S and an invariant unstable manifold U for the solutions of equation (4.1) having the following properties: (a) for each t0 the intersection St0 ∩Ut0 contains the unique element zt0 , (b) the solution u0(t) of equation (4.1) with initial condition u0(t0) = zt0 is bounded on the whole line R, (c) the solutions u(t) of equation (4.1) satisfying u(t0) ∈ St0 exponentially approach u0(t) as t→∞, (d) the solutions u(t) of equation (4.1) satisfying u(t0) ∈ Ut0 exponentially approach u0(t) as t→ −∞. 4.3. Invariant center-unstable manifolds on R. Using Theorem 4.2 and rescaling procedures similar to Theorem 3.1 to transform the trichotomy case to the dichotomy case, we can easily obtain the exsitence of an invariant center-unstable manifolds in the following theorem. Theorem 4.3. Let the evolution family (U(t, s))t≥s have an exponential trichotomy with the corresponding constants K, α, β (α < β), and projections (Pj(t))t∈R, j = 1, 2, 3, given in Def- inition 1.2. Suppose that f : R+ × X → X be ϕ-Lipschitz, where ϕ is the positive function which belongs to ER such that k < min { 1 N + 1 , 1− eα−β 1− e−β } , here k is defined by (4.7). Then there exists ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 790 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI a center-unstable manifold Cu = {(t,Cu t ) | t ∈ R+ and Cu t ⊂ X} for the solutions of equa- tion (3.1), with the family (Cu t )t∈R being the graphs of the family of Lipschitz continuous mappings (ht)t∈R ( i.e., Cu t := graph(ht) = { x + htx | x ∈ Im(P2(t) + P3(t)) } for each t ∈ R ) where ht : Im(P2(t) + P3(t))→ ImP1(t) has the Lipschitz constant Nk 1− k independent of t, such that the following properties hold: (i) to each x0 ∈ Cu t0 there corresponds one and only one solution u(t) of equation (3.1) on (−∞, t0] satisfying u(t0) = x0 and ess supt6t0 ‖e γtu(t)‖ <∞, where γ := α+ β 2 , (ii) Cu t is homeomorphism to X2(t)⊕ X3(t) for all t ∈ R, where X2(t) = ImP2(t) and X3(t) = = ImP3(t), (iii) Cu is invariant under the equation (3.1) in the sense that, if u(t) is the solution of equa- tion (3.1) satisfying u(t0) = x0 ∈ Cu t0 and ess supt6t0 ‖e γtu(t)‖ < ∞, then u(s) ∈ Cu s for all s 6 t0, (iv) every two solutions u1(t), u2(t) on the center-unstable manifold Cu satisfy the condition that there exist positive constants µ and Cµ independent of t0 ≥ 0 such that ‖x(t)− y(t)‖ 6 Cµe (µ−γ)(t−t0) ∥∥(P1(t0) + P3(t0))x(t0)− (P1(t0) + P3(t0))y(t0) ∥∥ (4.11) for all t 6 t0. Note that the existence of an invariant center-stable manifold on R is defined and proved by the same ways as in the case of half-line R+ (see Theorem 3.1). From the existence of the invariant center-stable and center-unstable manifolds of equation (4.1) defined on the whole line we have the following important corollary describing the behavior of solutions to equation (4.1). Corollary 4.3. Let the evolution family (U(t, s))t≥s have an exponential dichotomy with the corresponding projections P (t), t ∈ R, and the dichotomy constants N, α, β > 0. Suppose that f is ϕ-Lipschitz with the positive function ϕ ∈ ER such that k < min { 1 N + 1 , 1− eα−β 1− e−β , √ 2− 1 N + √ 2− 1 } , here k defined as in (4.7). Then there exist an invariant center-stable manifold C and an invariant center-unstable manifold Cu for the solutions of equation (4.1) having the following properties: (a) for each t0 ∈ R the intersection Ct0 ∩Cu t0 is homeomorphism to X3(t0) = P3(t0)X, (b) the solution u0(t) of equation (4.1) with initial condition u0(t0) ∈ Ct0 ∩ Cu t0 satisfies that ess supt∈R ‖e−γ|t|u(t)‖ <∞, where γ := α+ β 2 , (c) for the solution u(t) of equation (4.1) satisfying u(t0) ∈ Ct0 we have that e−γtu(t) exponen- tially approaches e−γtu0(t) as t→∞, (d) for the solution u(t) of equation (4.1) satisfying u(t0) ∈ Cu t0 we have that eγtu(t) exponen- tially approaches eγtu0(t) as t→ −∞. Proof. (a) Let us first prove that for each z ∈ X3(t) there exists a unique w ∈ X1(t) ⊕ X3(t) such that w = ht(z + gt(w)) + z, where gt and ht are the members of the Lipschitz mapping families (gt)t∈R and (ht)t∈R determining the invariant center-stable and center-unstable manifolds, ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 791 respectively. Indeed, the mapping L : X1(t)⊕ X3(t)→ X1(t)⊕ X3(t), y 7→ ht(z + gt(y)) + z satisfies that ‖Ly1 − Ly2‖ = ‖ht(z + gt(y1))− ht(z + gt(y2))‖ 6 Nk 1− k ‖gt(y1)− gt(y2)‖ 6 6 ( Nk 1− k )2 ‖y1 − y2‖. Since Nk 1− k < 1 we obtain that L is a contraction. Let w be its unique fixed point. Then w is the unique element in X1(t)⊕ X3(t) such that w = ht(z + gt(w)) + z. Define now the mapping D : X3(t) → Ct ∩ Cu t by D(z) = w + gt(w), where w is the unique element in X1(t)⊕ X3(t) such that w = ht(z + gt(w)) + z. Then we have w+ gt(w) = z + gt(w) + + ht(z + gt(w)) ∈ Ct ∩Cu t . The uniqueness of w yields that D is a well-defined mapping. We next prove the sujectiveness of D. For x ∈ Ct ∩Cu t we have that there is u ∈ X1(t)⊕ X3(t) and v ∈ X2(t) ⊕ X3(t) such that x = u + gt(u) = v + ht(v). Then we have u − ht(v) = v − − gt(u) ∈ ( X1(t) ⊕ X3(t) ) ∩ ( X2(t) ⊕ X3(t) ) = X3(t). Therefore, there is a z ∈ X3(t) such that u− ht(v) = v − gt(u) = z. This follows that u− ht(z + gt(u)) = z. As shown above, this relation means that Dz = u+ gt(u) = x. Therefore, D is surjevtive. We now prove that D is a Lipschitz mapping. In fact, by the definition of D we have D(z1) = = w1 + gt(w1) and D(z2) = w2 + gt(w2) for w1 and w2 being the unique solutions in X1(t)⊕X3(t) of equations w1 = ht(z1 + gt(w1)) + z1 and w2 = ht(z2 + gt(w2)) + z2, respectively. Then putting l = Nk 1− k (the Lipschitz constant of gt and ht), we have (1− l)‖w1 − w2‖ 6 ‖D(z1)−D(z2)‖ = = ‖z1 + ht(z1 + gt(w1)) + gt(w1)− ( z2 + ht(z2 + gt(w2)) + gt(w2) ) ‖ 6 6 ‖z1 − z2‖+ l‖z1 − z2‖+ l‖gt(w1)− gt(w2)‖+ ‖gt(w2)− gt(w2)‖ 6 6 (1 + l)‖z1 − z2‖+ l(l + 1)‖w2 − w2‖. Therefore, we obtain that ‖D(z1)−D(z2)‖ 6 (1 + l)‖z1 − z2‖+ l(l + 1) 1− l ‖D(z1)−D(z2)‖. Thus, ‖D(z1)−D(z2)‖ 6 1− l2 2− (1 + l)2 ‖z1 − z2‖, here we note that 2 − (1 + l)2 > 0 since k < √ 2− 1 N + √ 2− 1 . Hence, we obtain that D is a Lipschitz mapping with Lischitz constant 1− l2 2− (1 + l)2 . This follows that D is continuous and injective. As ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 792 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI shown above D is already surjective, therefore, D is bijetive. The inverse D−1 of D is defined as D−1 : Ct ∩ Cu t → X3(t) with D−1(w + gt(w)) = z if z = w − ht(z + gt(w)) (note that, by the contraction-mapping arguments we can easily show that for each w ∈ X1(t) ⊕ X3(t) there exists a unique z ∈ X3(t) such that z = w − ht(z + gt(w))). We then prove that D−1 is also a Lipschitz mapping. Indeed, for x1 = u+ gt(u) and x2 = v + gt(v) belonging to Ct ∩Cu t we have that ‖D−1x1 −D−1x2‖ = ‖z1 − z2‖ 6 6 ‖w1 − ht(z1 + gt(w1))− ( w2 − ht(z2 + gt(w2)) ) ‖ 6 6 ‖w1 − w2‖+ l‖z1 − z2‖+ l2|‖w1 − w2‖ = = (1 + l2)‖w1 − w2)‖+ l‖D−1x1 −D−1x2‖ 6 6 1 + l2 1− l ‖w1 + gt(w1)− w2 − gt(w2)‖+ l‖D−1x1 −D−1x2‖ = = 1 + l2 1− l ‖x1 − x2‖+ l‖D−1x1 −D−1x2‖. Therefore, we obtain that ∥∥D−1x1 −D−1x2 ∥∥ 6 1 + l2 (1− l)2 ‖x1 − x2‖. Hence, D−1 is also Lipschitz mapping. This follows that D is a homeomorphism, and we obtain that Ct ∩Cu t is homeomorphism to X3(t) for all t ∈ R. The property (b) follows from the definitions of the invariant center-stable and center-ustable manifolds, respectively. The properties (c) and (d) are consequences of the inequalities (3.4) and (4.11), respectively. Corollary 4.3 is proved. 5. Examples. In this section, we give some concrete examples of reaction-diffusion equations to illustrate our abstract results. The reaction-diffusion processes are modeled by the following equation: dx(t) dt = A(t)x(t) + f(t, x), where x(t) is the density of material, the partial differential operators A(t) represent the diffusion, and f represents the source of material which, in many contexts, depends on time in diversified manners (see [23] (Chapter 11), [24, 37]). Therefore, sometimes one may not hope to have the uniformly Lipschitz continuity of f. Our theoretical results hence give a chance to consider the above reaction-diffusion equation in general cases. Let us start by the following equation. Example 5.1. Consider the reaction-diffusion equation of the form dx(t) dt = Ax(t) + f(t, x), ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 793 where A is a sectorial operator satisfying that the spectrum σ(A) of A is decomposed in three disjoint sets that are {λ ∈ σ(A) | Reλ < 0}, {λ ∈ σ(A) | Reλ > 0}, and {λ ∈ σ(A) | Reλ = 0} such that σ(A)∩ iR is of finitely many points. Then, A is a generator of an analytic semigroup (T (t))t≥0. We define the evolution family U(t, s) := T (t − s) for all t ≥ s ≥ 0. We now claim that it has an exponential trichotomy with an appropriate choice of projections. By the spectral mapping theorem for analytic semigroups we have that, for fixed t0, the spectrum of the operator T (t0) splits into three disjoint sets σ1, σ2, σ3, where σ1 ⊂ {|z| < 1}, σ2 ⊂ {|z| > 1}, σ3 ⊂ {|z| = 1} with σ3 consisting of finitely many points. Next, we choose P1 = P1(t0), P2 = P2(t0), P3 = P3(t0) be the Riesz projections corresponding to the spectral sets σ1, σ2, σ3, respectively. Clearly, P1, P2 and P3 commute with T (t) for all t ≥ 0. Obviously, P1 + P2 + P3 = I and PiPj = 0 for i 6= j, and there are positive constants M, δ such that ‖T (t)P1‖ 6 Me−δt for all t ≥ 0. Furthermore, let Q := P2 + P3 = I − P1 and consider the strongly continuous semigoup (TQ(t))t≥0 on the space ImQ, where TQ(t) := T (t)Q. Since σ2 ∪ σ3 = σ(TQ(t0)), (TQ(t))t≥0 can be extended to a group (TQ(t))t∈R in ImQ. As well-known in the semigroup theory, there are positive constants K, α, γ such that α can be chosen as small as required (we may let α < γ), and the following estimates hold: ‖TQ(−t)P2‖ 6 Ke−γt for all t ≥ 0, ‖TQ(t)P3‖ 6 Keα|t| for all t ∈ R. Summing up the above discussions, we conclude that the evolution family (U(t, s))t≥s has an expo- nential trichotomy with projections Pj , j = 1, 2, 3, and positive constants N, α, β, where β := min{δ, γ}, N := max{K,M}. Thus, if f is ϕ-Lipschitz for some positive function ϕ satisfying that supt∈R ∫ t+1 t ϕ(τ)dτ is small enough, then the integral equation x(t) = U(t, s)x(s) + t∫ s U(t, ξ)f(ξ, x(ξ))dξ for all t ≥ s, has a center manifold. Example 5.2. For fixed n ∈ N∗, consider the equation wt(x, t) = wxx(x, t) + n2w(x, t) + ϕ(t) sin(w(x, t)), 0 6 x 6 π, t ∈ R, w(0, t) = w(π, t) = 0, t ∈ R, (5.1) where the step function ϕ(t) is defined as in formula (5.2). We define X := L2[0, π], and let A : X→ X be defined by A(y) = y′′ + n2y, with D(A) = { y ∈ X : y and y′′ are absolutely continuous, y′′ ∈ X, y(0) = y(π) = 0 } . ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 794 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI The equation (5.1) can now be rewritten as du dt = Au+ f(t, u) for u(t) = w(·, t), where f : R× X→ X, with f(t, u) = ϕ(t) sin(u) for ϕ being defined for a constant c > 1 by ϕ(t) = |k| if t ∈ [ 2k + 1 2 − 1 2|k|+c , 2k + 1 2 + 1 2|k|+c ] for k = 0,±1,±2, . . . , 0 otherwise. (5.2) Here, we note that ϕ can take any arbitrarily large value but we still have that sup t∈R t+1∫ t |ϕ(τ)|dτ 6 2 sup k∈Z 2k+1 2 + 1 2|k|+c∫ 2k+1 2 − 1 2|k|+c |k|dt = 2 sup k∈Z |k| 2|k|+c−2 6 1 2c−1 . Therefore, ϕ ∈M(R) which is an admissible space. It can be seen that (see [9]) that A is the generator of an analytic semigroup (T (t))t≥0. Since σ(A) = {−1+n2,−4+n2, . . . , 0,−(1+n)2+n2, . . .}, applying the spectral mapping theorem for analytic semigroups we get σ(T (t)) = etσ(A) = = {et(n2−1), et(n 2−4), . . . , et((n−1) 2−n2)} ∪ {1} ∪ {e−t((1+n)2−n2), e−t((2+n) 2−n2), . . .}. One can see easily that the nonlinear forcing term f is ϕ-Lipschitz. Using Example 5.1 we obtain that, if supt∈R ∫ t+1 t ϕ(τ)dτ, which is less than 1 2c−1 , is sufficient small (or c is sufficiently large), then there exists a center manifold for mild solutions of equation (5.1). Example 5.3. For fixed n ∈ N∗, consider the equation wt(x, t) = a(t)[wxx(x, t) + n2w(x, t)] + ϕ(t) sin(w(x, t)), 0 6 x 6 π, t ∈ R, w(0, t) = w(π, t) = 0, t ∈ R, (5.3) where ϕ is defined as in (5.2); the function a(·) ∈ L1,loc(R) and satisfies the condition γ1 ≥ a(t) ≥ ≥ γ0 > 0 for fixed γ0, γ1 and a.e. t ∈ R. We put X := L2[0, π], and let A : X→ X be defined by A(y) = y′′ + n2y, with D(A) = { y ∈ X : y and y′′ are absolutely continuous, y′′ ∈ X, y(0) = y(π) = 0 } . Putting A(t) := a(t)A, the equation (5.3) can now be rewritten as du dt = A(t)u+ f(t, u) for u(t) = w(·, t), where f : R× X→ X, with f(t, u) = ϕ(t) sin(u). Thus, as the above examples, A is a sectorial operator and generates an analytic semigroup (T (t))t≥0, and σ(A) satisfies the conditions as in Examples 5.1 and 5.2. Therefore, A(t) “generates” the evolution family (U(t, s))t≥s which is defined by the formula ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 INTEGRAL MANIFOLDS FOR SEMILINEAR EVOLUTION EQUATIONS AND ADMISSIBILITY . . . 795 U(t, s) = T  t∫ s a(τ)dτ . Using the above arguments as in Examples 5.1 and 5.2 we have that the analytic semigroup (T (t))t≥0 has an exponential trichotomy with the projections Pk, k = 1, 2, 3, and the trichotomy constants N, α, β where α is as small as required. Also, the following estimates hold: (i) ‖T (t)|P1X‖ 6 Ne−βt, (ii) ‖T2(−t)‖ = ‖(T (t)|P2X)−1‖ 6 Ne−βt, (iii) ‖T (t)|P3X‖ 6 Neαt for all t ≥ 0. From this, it is straightforward to check that the evolution family (U(t, s))t≥s has an exponential trichotomy with the trichotomy projection Pk, k = 1, 2, 3, and the trichotomy constants N, β, α by the following estimates: ‖U(t, s)|P1X‖ = ∥∥∥∥∥∥T  t∫ s a(τ)dτ ∣∣∣∣∣∣ P1X ∥∥∥∥∥∥ 6 Ne−β(t−s), ‖U(s, t)|‖ = ‖(U(t, s)|P2X)−1‖ = ∥∥∥∥∥∥T − t∫ s a(τ)dτ ∣∣∣∣∣∣ P2X ∥∥∥∥∥∥ 6 Ne−β(t−s), ‖U(t, s)|P3X‖ = ∥∥∥∥∥∥T  t∫ s a(τ)dτ ∣∣∣∣∣∣ P3X ∥∥∥∥∥∥ 6 Neα(t−s) for all t ≥ s ≥ 0. Therefore, we obtain that, if supt∈R ∫ t+1 t ϕ(τ)dτ = 1 2c−1 is sufficient small, then there exists a center manifold for mild solutions of equation (5.3). 1. Anosov D. Geodesic flows on closed Riemann manifolds with negative curvature // Proc. Steklov Inst. Math. – 1967. – 90. 2. Aulbach B., Minh N. V. Nonlinear semigroups and the existence and stability of semilinear nonautonomous evolution equations // Abstract Appl. Anal. – 1996. – 1. – P. 351 – 380. 3. Bates P., Jones C. Invariant manifolds for semilinear partial differential equations // Dynam. Rep. – 1989. – 2. – P. 1 – 38. 4. Bogoliubov N., Mitropolsky Yu. The method of integral manifolds in nonlinear mechanics // Contrib. Different. Equat. – 1963. – 2. – P. 123 – 196. 5. Bogoliubov N., Mitropolsky Yu. Asymptotic methods in the theory of non-linear oscillations // Monogr. Adv. Math. and Phys. – New York: Gordon and Breach Sci. Publ., 1961. 6. Calderon A. P. Spaces between L1 and L∞ and the theorem of Marcinkiewicz // Stud. Math. – 1996. – 26. – P. 273 – 299. 7. Carr J. Applications of centre manifold theory // Appl. Math. Sci. – New York; Berlin: Springer, 1981 – 35. 8. Daleckii Ju. L., Krein M. G. Stability of solutions of differential equations in Banach spaces. – Provindence RI: Amer. Math. Soc., 1974. 9. Engel K. J., Nagel R. One-parameter semigroups for linear evolution equations // Grad. Text Math. – 2000. – 194. 10. Hadamard J. Sur l’interation et les solutions asymptotiques des equations differentielles // Bull. Soc. Math. France. – 1991. – 29. – P. 224 – 228. ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6 796 NGUYEN THIEU HUY, VU THI NGOC HA, HA PHI 11. Hale J., Magalhães L. T., Oliva W. M. Dynamics in infinite dimensions // Appl. Math. Sci. – 2002. – 47. 12. Henry D. Geometric theory of semilinear parabolic equations // Lect. Notes Math. – 1981. – 840. 13. Hirsch N., Pugh C., Shub M. Invariant manifolds // Lect. Notes Math. – 1977. – 183. 14. Nguyen Thieu Huy. Exponential dichotomy of evolution equations and admissibility of function spaces on a half-line // J. Funct. Anal. – 2006. – 235. – P. 330 – 354. 15. Nguyen Thieu Huy. Stable manifolds for semilinear evolution equations and admissibility of function spaces on a half-line // J. Math. Anal. and Appl. – 2009. – 354. – P. 372 – 386. 16. Nguyen Thieu Huy. Invariant manifolds of admissible classes for semilinear evolution equations // J. Different. Equat. – 2009. – 246. – P. 1820 – 1844. 17. Lindenstrauss J., Tzafriri L. Classical Banach spaces II // Function spaces. – Berlin: Springer, 1979. 18. Lunardi A. Analytic semigroups and optimal regularity in parabolic problems. – Basel: Birkhäuser, 1995. 19. Martin. Nonlinear operators and differential equations in Banach spaces. – New York: Wiley Intersci., 1976. 20. Massera J. J., Schäffer J. J. Linear differential equations and function spaces. – New York: Acad. Press, 1966. 21. Minh N. V., Räbiger F., Schnaubelt R. Exponential stability, exponential expansiveness and exponential dichotomy of evolution equations on the half line // Integr. Equat. and Oper. Theory. – 1998. – 32. – P. 332 – 353. 22. Minh N. V., Wu J. Invariant manifolds of partial functional differential equations // J. Different. Equat. – 2004. – 198. – P. 381 – 421. 23. Murray J. D. Mathematical biology I: An introduction. – Berlin: Springer, 2002. 24. Murray J. D. Mathematical biology II: Spatial models and biomedical applications. – Berlin: Springer, 2003. 25. Nagel R., Nickel G. Well-posedness for non-autonomous abtract Cauchy problems // Prog. Nonlinear Different. Equat. and Appl. – 2002. – 50. – P. 279 – 293. 26. Nickel G. On evolution semigroups and wellposedness of non-autonomous Cauchy problems: PhD Thesis. – Tübingen, 1996. 27. Nitecki. An introduction to the orbit structure of diffeomorphisms. – Cambridge, MA: MIT Press, 1971. 28. Pazy A. Semigroup of linear operators and application to partial differential equations. – Berlin: Springer, 1983. 29. Perron O. Über Stabilität und asymptotisches Verhalten der Integrale von Differentialgleichungssystemen // Math. Z. – 1929. – 29, №1. – S. 129 – 160. 30. Perron O. Die Stabilitätsfrage bei Differentialgleichungen // Math. Z. – 1930. – 32. – S. 703 – 728. 31. Räbiger F., Schaubelt R. The spectral mapping theorem for evolution semigroups on spaces of vector-valued functions // Semigroup Forum. – 1996. – 48. – P. 225 – 239. 32. Schnaubelt R. Exponential bounds and hyperbolicity of evolution families: PhD Thesis. – Tübingen, 1996. 33. Schnaubelt R. Exponential dichotomy of non-autonomous evolution equations // Habilitationsschrift. – Tübingen, 1999. 34. Schnaubelt R. Asymptotically autonomous parabolic evolution equations // J. Evol. Equat. – 2001. – 1. – P. 19 – 37. 35. Sell G. R., You Y. Dynamics of evolutionary equations // Appl. Math. Sci. – 2002. – 143. 36. Triebel H. Interpolation theory, function spaces, differential operators. – Amsterdam etc.: North-Holland, 1978. 37. Yagi A. Abstract parabolic evolution equations and their applications. – Springer, 2009. Received 02.06.11, after revision — 20.04.12 ISSN 1027-3190. Укр. мат. журн., 2012, т. 64, № 6