National and Kapodistrian University of Athens
Διαχωρισμένες ακολουθίες σε χώρους με νόρμα
Abstract
dc:descriptionThe problems that we are concerned in this Phd. Thesis have their roots in a well known consequence of the classical Riesz’s lemma: Let X be an infinite dimensional normed space, then there exists a normalized sequence (x_n) (‖x_n ‖=1) in X such that the distance between its elements is greater or equal to 1 (‖x_n-x_m ‖≥1,n≠m). There are two closely related problems in the geometry of Banach spaces: which is the maximal cardinality of an equilateral set in a Banach space X (a subset of X is called equilateral if the distance between its elements is constant) and which is the maximal cardinality of a set of normed-1 vectors that are separated from a given number (typically greater than 1). In the first chapter we concern ourselves with various extensions, in finite dimensional normed spaces, of a result of C. A. Kottman [19] which improves Riesz’s lemma: Let X be an infinite dimensional Banach space, then there exists an infinite subset Δ of S_X such that ‖x-y‖>1, for every x,y∈Δ with x≠y. Moreover the elements of Δ can be chosen as linear combinations, with coordinates 0,±1, of an Auerbach system in X. So we have the following results.Theorem (Theorem 1.1.8). Let n∈N,then there exists no subset A of C_n={0,±1}^n such that: (a) e_k∈A,for 1≤k≤n (b) A is a symmetric set (that is if x∈A then-x∈A) (c) for every subset B of A with |B|=n+1 there exist x≠y∈B such that x-y∈A. Thus if n∈N and A ⊆ C_n such that satisfies properties (a) and (b), then there exists a subset B of A with |B|=n+1, which is difference free with respect to A. Utilizing the above combinatorial result we can prove an analogous of Kottman’s theorem for finite dimensional Banach spaces. Theorem (Theorem 1.1.12). Let X be a finite dimensional Banach with dimX=n. Then for each Auerbach basis { (x_k,x_k^* ) ∶ 1≤k≤n} of X we can determine n+1 li-near combinations z_1,…,z_(n+1) of x_1,…,x_n with coordinates 0,±1 such that ‖z_k ‖=1, for 1≤k≤n+1 and ‖z_k-z_l ‖>1,for 1≤k<l≤n+1. By our methods we can also prove Kottman’s result for infinite dimensional Banach spaces (Proposition 1.1.14). In the second section of chapter 1 are presented results in the spirit of Theorems 1.1.8 and 1.1.12. These results concern sum-free sets and complex Banach spaces of finite dimension. Indicatively we mention the following results. Theorem (Theorem 1.2.6).Let X be a finite dimensional Banach space over real nu- mbers with dimX=n.Then for each Auerbach basis { (x_k,x_k^* ) ∶ 1≤k≤n} of X we can determine n linear combinations z_1,…,z_n of x_1,…,x_n with coordinates 0,±1 such that ‖z_k ‖=1, for 1≤k≤n and ‖z_k+z_l ‖>1,for 1≤k<l≤n. We also have the following complex analog of Theorem 1.1.12.Theorem (Theorem 1.2.13). Let X be a finite dimensinal complex Banach space with dimX=n. Then for each Auerbach basis { (x_k,x_k^* ) ∶ 1≤k≤n} we can determine 2n+2 linear combinations z_1,…,z_(2n+2) of x_1,…,x_n with coordinates 0,±1 and±i such that ‖z_k ‖=1, for 1≤k≤2n+2 and ‖z_k-z_l ‖>1,for 1≤k<l≤2n+2. In the second chapter we study the problem of not existence of infinite equilateral sets in infinite dimensional Banach spaces. The first to construct an example (1987) of an infinite dimensional Banach space which does not contain an infinite equilateral set was Terenzi [29]. Terenzi’s example is the classical space l_1 with a suitable equivalent norm. Further Terenzi in 1989 constructed an infinite dimensional subspace of l_1, which considered with the usual norm of l_1 does not contain an infinite equilateral set [30]. In the present thesis we construct, using Terenzi’s method, an example of an infinite dimensional Banach space not admitting an infinite equilateral set and is not isomorphic to a subspace of l_1. More specifically we have the following results. Theorem (Theorem 2.2.2). Let (X_n ) be a sequence offinite dimensional Banach spaces, each having an 1-unconditional basis. Then the space Z=(∑_(n=1)^∞▒⊕X_n )_1 has an equivalent dual norm ‖∙‖^' such that the space ( Z,‖∙‖^') does not admit an infinite equilateral set. Terenzi’s result [29] follows readily by Theorm (2.2.2), just by putting X_n=R,n≥1. Theorem (Corollary 2.2.3). There exists an equivalent norm on l_1 such that the Ba-nach space ( l_1,‖∙‖ ) does not admit an infinite equilateral set. If we choose p>2 and put X_n=l_p^n,n≥1, then by known arguments (concerning type and cotype of Banach spaces) and Theorem 2.2.2 we have the following. Theorem (Corollary 2.2.5). There exists a separable Banach space not isomorphic to a subspace of l_1 and not containing an equilateral set. Further, according to Remark (2.2.4), we have an example of a w^*-closed and strictly convex subspace of l_1, which does not contain an infinite equilateral set. Thus we also get the second example of Terenzi [30]. In chapter 3 we investigate an extension of the classical concept of antipodal sets, which concerns real vector spaces of finite dimension [26], in normed spaces of any ( finite or infinite) dimension. This extension was presented in the paper of Mercourakis-Vassiliadis [24], where was also introduced the stronger concept of bounded and separated antipodal set, in order to investigate the isomorphic structure of equilateral sets in Banach spaces. Let X be a Banach space, a subset S of X is called bounded separated and antipodal if S⊆B_X and there exists d>0 such that for any x≠y∈S there is f∈B_(X^* ) such that f(x)-f(y)≥d and f(y)≤f(z)≤f(x), for every z∈S. It is clear that such a set is a d-separated subset of B_X. This last remark combined with the classical Theorem of Elton and Odell [10], which states that in every Banach space there exists an infinite (1+ε)-separated subset Δ of S_X (ε>0), leads us to the following question: Does there exist an infinite subset S of B_X bounded, separated and antipodal with constant d>1? In the present chapter we give an affirmative answer to the above question and we investigate some related topics. To this end, we define the following parameter for an infinite dimensional Banach space, which was firstly introduced in [24] : K_a (X)=sup{ d>0 ∶ there exists a bounded and separated subset of Xc_1=c_2=1 and d}. This parameter is closely related with the parameterK(X)=sup{ d∶∃ S⊆B_X infinite and d-separated}, which is known as Kottman’s constant. Let us note that by the existence of an infinite Auerbach system in any Banach space X ( [7], [17] ), we can easily conclude that 1≤K_a (X)≤K(X)≤2. We now present the basic results of this chapter. The first one is the analog of a result of Kottman (Theorem 3 of [19]). Theorem (Theorem 3.5). Let X be a Banach space. If X contains an isimorphic copy of c_0 or l_1, then K_a (X)=2. If X contains an isomorphic copy of some l_p,1<p<∞,then K_a (X)≥2^(1⁄p). The next result gives an affirmative answer to the question posed earlier, strengthening the Theorem of Elton and Odell. Theorem (Theorem 3.19). Let X be an infinite dimensional Banach space.Then K_a (X)>1. While we do not know if the equality K_a (X)=K(X) holds for any Banach space X, we can prove that it indeed holds in the case of reflexive Banach spaces.Theorem (Theorem 3.20). Let X be a reflexive ifinite dimensional Banach space, then K_a (X)=K(X). Finally we prove one more analog of a Kottman’s result (Theorem 7 of [19]).Theorem (Theorem 3.22). For every infinite dimensional Banach space X there exists b∈[0,2] such that the set (K_a (X))≔{ K_a (Y) ∶Y isomorphic toX} is either equal to the interval (b,2] or to the interval [b,2].
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Glakousakis, Eftychios
- Γλακουσάκης, Ευτύχιος
Subjects
dc:subject × 14Rights
dc:rights- Statement dc:rights
-
- BY_NC
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/43670