THREE NON-LINEAR PROBLEMS ON NORMED SPACES
A dissertation submitted to Kent State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy
by
Francisco J. Garc´ıa
February, 2007 Dissertation written by
Francisco J. Garc´ıa
B.S., University of C´adiz, Spain, 2000
M.S., University of C´adiz, Spain, 2004
Ph.D., University of C´adiz, Spain, 2005
M.A., Kent State University, 2006
Ph.D., Kent State University, 2007
Approved by
Richard M. Aron, Chair, Doctoral Dissertation Committee
Andrew Tonge, Members, Doctoral Dissertation Committee
Per H. Enflo,
Johnnie W. Baker, (Outside Person)
Paul S. Wang, (Graduate Representative)
Accepted by
Andrew Tonge, Chair, Department of Mathematical Sciences
Jerry Feezel, Dean, College of Arts and Sciences
ii TABLE OF CONTENTS
ACKNOWLEDGEMENTS ...... v
INTRODUCTION ...... 1
1 THE LINEABILITY PROBLEM FOR FUNCTIONALS ...... 7
1.1 Preliminaries ...... 7
1.2 Lineability of NA (X)...... 16
1.3 Lineability of X∗ \ NA (X)...... 21
1.4 Density of X∗ \ NA (X)...... 24
2 THE MINIMUM-NORM PROBLEM FOR TRANSLATIONS ..... 31
2.1 Preliminaries ...... 31
2.2 Minimum-norm elements and norm-attaining functionals ...... 42
2.3 Non-complete normed spaces having only norm-attaining functionals . . . . 45
2.4 Partial solutions ...... 48
3 THE BANACH-MAZUR CONJECTURE FOR ROTATIONS ...... 51
3.1 Preliminaries ...... 51
iii 3.2 Geometrical conditions ...... 58
3.3 Topological conditions ...... 60
3.4 Intermediate solutions ...... 63
BIBLIOGRAPHY ...... 68
iv ACKNOWLEDGEMENTS
There are several people to whom I should be giving thanks for my stay at Kent State
University: Richard Aron, Joe Diestel, Andrew Tonge, Artem Zvavitch, Per Enflo, and
Juan Seoane. I would like to use this opportunity to thank all of them very much: Thanks to Richard for accepting me as his student and for all the invitations to his house; thanks to
Joe and Artem for the recommendation letters that they wrote for me; thanks to Andrew for all the forms that he signed for me; thanks to Per for his excellent mathematical support; and, finally, thanks very much Juan for making my stay at Kent as good as possible.
Also, I want to give thanks to the Department of Mathematical Sciences at Kent State
University for the privilege of having been supported by the Graduate Program. Specially,
I wish to thank Virginia Wright (Secretary of the Math Department,) Misty Tackett (Sec- retary of the Math Graduate Students,) and Michelle Cordier (Secretary Assistant,) for attending me so well every time I needed their help.
Other people I would like to thank are my friends in the Math Department like Mienie,
Ramiro, Tom, Hongcheng, Jeff, Terry, Brian, Antonia, Daniele, and Alejandro. Thanks to all of you guys, overall for the rides from and to Cleveland Hopkins International Airport.
Finally I mention all my friends and my family from my town. Brothers and sisters, thanks for always being there for me. Please, never change!
Francisco J. Garc´ıa
Febraury 2007
v GARC´IA, FRANCISCO J., Ph.D., February, 2007 PURE MATHEMATICS
THREE NON-LINEAR PROBLEMS ON NORMED SPACES (71 pp.)
Director of Dissertation: Richard M. Aron
In this dissertation, we will study the following three non-linear problems:
1. The lineability problem for functionals.
2. The minimum-norm problem for translations.
3. The Banach-Mazur conjecture for rotations.
As far as we know, all of them are currently open, and we believe that any approach to their solutions will constitute a work of great interest to the mathematical community. In this dissertation, we obtain progresses that lead to partial solutions of these problems. INTRODUCTION
The first chapter of this dissertation is on the structure of the set of functionals on a Ba- nach space that attain their norm. The reason for this interest comes from an open problem concerning the lineability of the set NA (X) of norm-attaining functionals on a Banach space
X. Specifically, it is unknown if NA (X) always contains an infinite dimensional, or even a 2-dimensional subspace. The paper [9] is an excellent reference about this problem. As we will see, from this problem arises another one corresponding to the complementary set,
X∗ \ NA (X) , of non-norm-attaining functionals on X. In concrete terms, for non-reflexive spaces X, is X∗ \ NA (X) always lineable, or dense? Our best results from this chapter are the following:
1. Every Banach space admitting an infinite dimensional separable quotient can be equiv-
alently renormed to make the set of norm-attaining functionals lineable.
2. There exists a non-reflexive dual Banach space that cannot be equivalently dually
renormed to make the set of non-norm-attaining functionals even 2-lineable.
3. Every Banach space can be equivalently renormed to make the set of non-norm-
attaining functionals non-dense.
All the results presented in this chapter, unless explicitly stated, can be found in [2].
The second chapter consists of the study of norm-attaining functionals on non-complete spaces. In 1964, James proved that a Banach space is reflexive if and only if every func- tional is norm-attaining (see [28].) Besides, James showed in 1971 that the completeness hypothesis cannot be skipped, since he found a non-complete normed space on which every
1 2
functional is norm-attaining (see [29].) Afterwards, Blatter proved in 1976 that a necessary and sufficient condition for a normed space to be reflexive is that every closed convex set has a minimum-norm element (see [17].) This was the first time when minimum-norm elements appeared in the literature. However, they appeared again in 2005, when Aizpuru and the author generalized Blatter’s result by proving that a normed space is reflexive if and only if every bounded closed convex set with non-empty interior has a minimum-norm element
(see [4].) Here, the bounded closed convex sets with non-empty interior became important, and with this we have the following conjecture: A necessary and sufficient condition for a normed space to have only norm-attaining functionals is that every bounded closed convex set with non-empty interior can be translated to have a non-zero minimum-norm element.
Another interesting problem arising from this is to characterize the reflexive spaces that contain a proper dense subspace on which every functional is norm-attaining. Our best results from this chapter are the following:
1. A necessary and sufficient condition for a normed space to have only norm-attaining
functionals is that every closed convex subset with non-empty interior and non-empty
boundary can be translated to have a non-zero minimum-norm element.
2. If a non-complete norm space is so that every bounded closed convex subset of it with
non-empty interior can be translated to have a non-zero minimum-norm element, then
its completion cannot be rotund.
3. Every infinite dimensional reflexive Banach space can be equivalently renormed to be
non-rotund and to not possess dense proper subspaces on which every functional is
norm-attaining.
All the results presented in this chapter, unless explicitly stated, can be found in [24]. 3
The third and last chapter is about a famous old problem. It has been always well known that every Hilbert space is transitive. On the other hand, it seems likely that
Banach already knew some examples of transitive Banach spaces that were not Hilbert.
Apparently, Mazur, who was working with separable spaces at that time, asked Banach for the existence of transitive and separable Banach spaces different from `2. Banach came up
with no answer and this is how the Banach-Mazur conjecture was born: every transitive
and separable Banach space is Hilbert. It is believed that Mazur conjectured a positive
answer. An excellent reference is [12], and another good one is [18]. Basically, there are two
geometrical properties clearly differentiated: Rotundity and smoothness. Therefore, it is
natural to wonder whether a transitive and separable Banach space is rotund and smooth.
In 1932 (see [33]) Mazur proved that in every separable Banach space the set of smooth
points of the unit ball is a Gδ dense subset of the unit sphere. As a consequence, every
transitive and separable Banach space is smooth. Now, the question remains of whether a
transitive and separable Banach space is rotund. Our best results from this chapter are the
following:
1. If the unit ball of a smooth and separable Banach space is free of rotund points, then
the set of non-norm-attaining functionals on it contains a Gδ dense subset.
2. If a transitive and separable Banach space is so that the set of non-norm-attaining
functionals on it is not dense, then the space is rotund, the set of rotund points of the
unit ball of its dual is dense in the unit sphere of its dual, and the set of norm-attaining
functionals on it is open.
3. If the unit ball of a transitive and separable Banach space either has normal structure
or is dentable, then the space is rotund.
All the results presented in this chapter, unless explicitly stated, can be found in [5]. 4
As the reader may suppose, in much of this dissertation we will make use of various classical notions, such as smoothness, rotundity, etc., from the geometry of Banach spaces.
Now, we will briefly describe the notation we have followed through the whole dissertation.
Assume that X denotes a topological vector space. Then:
1. If X is metrizable then BX (x, δ), UX (x, δ), and SX (x, δ) will denote the closed ball
of center x and radius δ, the open ball of center x and radius δ, and the sphere of
center x and radius δ, respectively.
2. If X is normable then BX , UX , and SX will denote the closed unit ball, the open unit
ball, and the unit sphere, respectively.
3. X∗ will denote the topological dual of X.
4. If M is any subset of X, then:
(a) span (M) will denote the vector subspace generated by M and span (M) will
denote the closed vector subspace generated by M.
(b) co (M) will denote the convex hull of M and co (M) will denote the closed convex
hull of M.
(c) int (M), bd (M), and cl (M) will denote the interior of M, the boundary of M,
and the closure of M.
To finish the introduction, we will review some basic concepts from the geometry of
Banach spaces. Again, assume that X denotes a topological vector space and consider a
convex subset H of X. Then:
1. The slice of H determined by f ∈ X∗ \{0} and δ > 0 is defined by slc (H, f, δ) =
{h ∈ H : f (h) ≥ sup (Ref (H)) − δ} . 5
2. A subset C of H is said to be a face of H if it is convex and verifies that, for every
x, y ∈ H and every α ∈ (0, 1) with αx + (1 − α) y ∈ C, then x, y ∈ C.
3. A subset C of H is said to be an exposed face of H if there exists f ∈ X∗ such that
C = {x ∈ H : f (x) = sup (Ref (H))} .
4. A subset C of H is said to be a strongly exposed face of H if there exists f ∈ X∗ such
that C = {x ∈ H : f (x) = sup (Ref (H))} and for every open set U of H containing
C there is δ > 0 such that slc (H, f, δ) ⊆ U.
5. C is said to be a maximal face of H if it is a non-trivial face maximal among the
non-trivial faces of H.
6. A subset C of H is said to be a smooth face of H if it is a non-trivial face and has
non-empty interior relative to the boundary of H.
7. A point c ∈ H is said to be an extreme point of H if {c} is a face of H. We will let
ext (H) denote the set of extreme points of H.
8. A point c ∈ H is said to be an exposed point of H if {c} is an exposed face of H. We
will let exp (H) denote the set of exposed points of H.
9. A point c ∈ H is said to be a strongly exposed point of H if {c} is a strongly exposed
face of H. We will let exps (H) denote the set of strongly exposed points of H.
10. A point c ∈ H is said to be a rotund point of H if {c} is a maximal face of H. We
will let rot (H) denote the set of rotund points of H.
Now, assume that X is a normed space. Then:
1. A point x ∈ SX is said to be a smooth point of BX if every sequence (fn) ⊂ SX∗ n∈N 6
∗ such that (fn (x)) converges to 1 is ω -convergent. The set of the smooth points n∈N
of BX is denoted by smo (BX ).
2. A point x ∈ SX is said to be a strongly smooth point of BX if every sequence (fn) ⊂ n∈N
SX∗ such that (fn (x)) converges to 1 is convergent. The set of the strongly smooth n∈N
points of BX is denoted by smos (BX ).
Finally, if X is a smooth normed space, then the dual map of X is defined as the function
∗ ∗ JX : X −→ X such that, for every x ∈ X, JX (x) is the unique element in X verifying
2 that kJX k = kxk and JX (x)(x) = kxk .
More background information and notation can be obtained in [19], [20], and [34]. We
also refer the reader to [3], [10], [11], and [26] for a wider perspective of these concepts.
Francisco J. Garc´ıa
February 2007 CHAPTER 1
THE LINEABILITY PROBLEM FOR FUNCTIONALS
1.1 Preliminaries
We begin by recalling the following relatively new concepts related to the “size” of subsets of Banach spaces.
Definition 1.1.1 (Gurariy, 1991) A subset M of a Banach space is said to be
1. n-lineable if M ∪ {0} contains an n-dimensional vector subspace;
2. lineable if M ∪ {0} contains an infinite dimensional vector subspace;
3. dense-lineable if M ∪ {0} contains an infinite dimensional dense vector subspace;
4. spaceable if M ∪ {0} contains an infinite dimensional closed vector subspace.
In order to have a better perspective of these new concepts, we refer the reader to the papers in [7], [8], and [25], where it is proved that several pathological properties occur more often than one might expect in the sense described in the definitions above.
The main questions that we want to attack in this chapter are the following:
Problem 1.1.2 (Aron/Gurariy, 2004) Is the set of norm-attaining functionals on an infinite dimensional Banach space always lineable?
7 8
The very first results relative to this question and trying to answer it are presented now.
All of them appeared in [9].
Remark 1.1.3 (Bandyopadhyay/Godefroy, 2005) Let X be a Banach space. Then,
X ⊆ NA (X∗) and hence NA (X∗) is spaceable if X is infinite dimensional. Observe that X
is reflexive if and only if X = NA (X∗).
Theorem 1.1.4 (Bandyopadhyay/Godefroy, 2005) Let X be a Banach space such that
∗ BX∗ is ω -sequentially compact. The following conditions are equivalent:
1. There exists an equivalent norm on X that makes NA (X) spaceable.
2. There exists an infinite dimensional quotient of X which is isomorphic to a dual space.
Theorem 1.1.5 (Bandyopadhyay/Godefroy, 2005) Let X be an Asplund Banach space with the Dunford-Pettis property. Then, the closed vector subspaces of NA (X) are finite di- mensional. In particular, X cannot be equivalently renormed to make NA (X) spaceable.
Theorem 1.1.6 (Bandyopadhyay/Godefroy, 2005) Let X be a Banach space with the
Radon-Nikodym property or an almost locally uniformly rotund norm. Then, we have that span (NA (X)) = X∗. In particular, if X is not reflexive but has the Radon-Nikodym prop- erty then X cannot be equivalently renormed to make NA (X) be a vector space.
Now we will present several results on concrete spaces giving partially solutions to the previous question.
We begin with spaces of continuous functions, but first we need the following two lemmas involving compact Hausdorff spaces and subspaces of `1. 9
Lemma 1.1.7 Let K be an infinite compact Hausdorff topological space. Then, there exists a sequence (lj) ⊆ K such that li 6= lj if i 6= j, and verifying one of the following j∈N conditions:
1. The sequence (lj) converges to some point ∞ ∈ K. j∈N
2. The set {lj : j ∈ N} does not contain isolated points.
Proof Firstly, note that if K is scattered then we deduce (see [32]) that K is sequentially compact. Therefore, we can assume that K is not scattered. It is known (see [32]) that
K = P ∪ D where P is closed and perfect (the perfect kernel of K) and D is open and scattered (the scattered kernel of K.) (Note that this decomposition holds not only for compact Hausdorff spaces, but for any topological space.) Now, there exists a sequence
(lj) ⊆ P such that li 6= lj for i 6= j and the set {lj : j ∈ } does not contain isolated j∈N N points.
Lemma 1.1.8 There exists an infinite dimensional vector subspace M of `1 such that every
(αi) ∈ M \{0} verifies one of the following conditions: ∈N
1. The set {j ∈ N : Re (αj) > 0} is non-empty and finite and the set {j ∈ N : Re (αj) < 0} is infinite.
2. The set {j ∈ N : Re (αj) > 0} is infinite and the set {j ∈ N : Re (αj) < 0} is non- empty and finite. 10
Proof Let us consider the following elements of `1: y = 1, −1, 1 , 1 , 1 , 1 , 1 , 1 ,... 1 23 24 25 26 27 28 1 1 1 1 y2 = 0, 0, 1, −1, 5 , 6 , 7 , 8 ,... 3 3 3 3 1 1 y3 = 0, 0, 0, 0, 1, −1, 47 , 48 ,... . . . y = 0,..., 0, 1, −1, 1 , 1 ,... k (k+1)2k+1 (k+1)2k+2 . .
We will take M := span {yk : k ∈ N}. Indeed, assume that 0 6= y ∈ M and take λ1, . . . , λk ∈
K not all zero so that y = λ1y1 + ··· + λkyk. Then, y has the form ! λ1 λ1 λ1 λ2 λk (1.1) λ1, −λ1, + λ2, − λ2,..., + + ··· + ,... . 23 24 22k+1 32k+1 (k + 1)2k+1
Therefore, from equation (1.1) we can see that M is the desired vector space.
In the above proof we note the following remark.
Remark 1.1.9 If y ∈ span {yk : k ∈ N}\ span {yk : k ∈ N}, then y has the form
λ λ λ λ λ λ λ , −λ , 1 + λ , 1 − λ , 1 + 2 + λ , 1 + 2 − λ ,... , 1 1 23 2 24 2 25 35 3 26 36 3
which shows that the closure of M does not serve our purposes.
Theorem 1.1.10 Let K be an infinite compact Hausdorff topological space. Then, both
NA (C (K)) and C (K)∗ \ NA (C (K)) are lineable. If, in addition, K possesses a non-trivial convergent sequence, then NA (C (K)) is spaceable.
Proof In the first place, let us consider the vector space V of all regular Borel measures µ, with bounded variation on K, such that there exists a finite set L ⊂ K with |µ| (K \ L) = 0. 11
(Observe that V is no other than the vector space generated by {δt : t ∈ K}.) Also notice P that if µ ∈ V and kµk = 1, then l∈L |µ ({l})| = 1. Now, define the continuous function f : L −→ [−1, 1] as |µ({l})| if µ ({l}) 6= 0, f (l) = µ({l}) 0 if µ ({l}) = 0, for all l ∈ L. Since K is normal and L is closed, we deduce by Urysohn’s Lemma that there
˜ ˜ exists a continuous extension f : K −→ [−1, 1] of f. To finish, we have that f = 1 and Z X X fdµ˜ = f (l) µ ({l}) = |µ ({l})| = 1. K l∈L l∈L
In the second place, by using Lemma 1.1.7, we will firstly assume the existence in K
of a non-trivial convergent sequence. So, let L = (lj) ⊆ K be a convergent sequence to j∈N S∞ some point ∞ ∈ K and verifying that li 6= lj if i 6= j. We can set N = k=1 Nk where
the Nk’s are pairwise disjoint containing infinitely many even numbers and infinitely many S∞ odd numbers. Thus, we can write L = k=1 Lk where l ∈ Lk if and only if l = lj for some
j ∈ Nk. Now choose, for each k ∈ N, any regular Borel measure µk with bounded variation j on K, such that |µk| (K \ Lk) = 0 and the sign of each µk ({lj}) is (−1) , for j ∈ Nk. Let
P∞ n 0 6= µ ∈ W := span {µk : k ∈ } and consider a sequence ( λ µk) converging to µ, N k=1 k n∈N n where (λ ) ∈ c00 for every n ∈ . Since k k∈N N ∞ ! X n λk µk ({l}) −→ µ ({l}) as n → ∞ k=1 for every l ∈ L, we deduce that |µ| (K \ L) = 0. Now, since µ 6= 0, there exists l0 ∈ L such
P∞ n that µ ({l0}) 6= 0. There also exists k0 ∈ N such that l0 ∈ Lk0 and so ( k=1 λk µk)({l0}) = λn µ ({l }) for every n ∈ , which means that k0 k0 0 N
n µ ({l0}) λk0 −→ λk0 := as n → ∞. µk0 ({l0})
Then, if l ∈ Lk0 we have that µ ({l}) = λk0 µk0 ({l}). Therefore, the signs of µ ({l}) are P alternating for l ∈ Lk0 . Finally, assume that kµk = 1. Then l∈L |µ ({l})| = 1. Now if 12
R P f ∈ C (K), kfk = 1, and K fdµ = 1, we have that l∈L f (l) µ ({l}) = 1, which means |µ({l})| that f (l) = µ({l}) for all l ∈ L such that µ ({l}) 6= 0. However, this is impossible because
(f (lj)) converges to f (∞). Therefore, each non-zero measure µ ∈ W is, in fact, in j∈N C (K)∗ \ NA (C (K)).
In the third and last place, we will assume the existence in K of a non-trivial sequence with no isolated points. So, let (lj) ⊆ K be a sequence free of isolated points and verifying j∈N ∗ that li 6= lj if i 6= j. Let us consider in C (K) the infinite dimensional vector subspace nP∞ o W := αjδl :(αj) ∈ M , where M is the infinite dimensional vector subspace j=1 j j∈N constructed in Lemma 1.1.8. Observe that W is linearly isometric to M. Let µ ∈ W with
P∞ kµk = 1, so that µ is of the form αjδl with (αj) ∈ M. Since kµk = 1, we have that j=1 j j∈N P∞ j=1 |αj| = 1. We can assume, without loss of generality, that A = {j ∈ N : Re (αj) < 0} is non-empty and finite and B = {j ∈ N : Re (αj) > 0} is infinite. Let f ∈ C (K, R) with R P∞ P∞ kfk = K fdµ = 1. Then, j=1 αjf (lj) = 1 = j=1 Re (αj) f (lj), and from here we
deduce that, if j ∈ A then f (lj) < 0, and if j ∈ B then f (lj) > 0. However, by taking into
account that lk is a cluster point of {lj : j ∈ N} for every k ∈ A, we have that there must
be an infinite amount of natural numbers k ∈ N with f (lk) < 0, which is a contradiction. Therefore, each non-zero measure µ ∈ W is, in fact, in C (K)∗ \ NA (C (K)).
In view of Theorem 1.1.5 we have the following remark.
Remark 1.1.11 Given any compact Hausdorff topological space K we have the equivalence of the following assertions (see [21] and [27]):
1. The space C (K) has the hereditary Dunford-Pettis property.
2. The space C (K) is Asplund. 13
3. The space K is scattered.
Therefore, if K is scattered then for every infinite dimensional closed subspace X of C (K) we have that NA (X) is not spaceable.
We now turn to spaces of integrable functions.
∗ ∗ Lemma 1.1.12 Let (Ω, Σ, µ) be a σ-finite measure space. Then, an element x ∈ L1 (µ) (=
L∞ (µ)) attains its norm if and only if there is a measurable set A with µ (A) > 0 satisfying that |x∗ (t)| = kx∗k for all t ∈ A.
Proof First, if some subset A of positive measure satisfies the above condition, then the function θ :Ω −→ K given by |x∗(t)| ∗ if t ∈ A, θ (t) = x (t) 0 otherwise,
is measurable and clearly satisfies that kθk∞ ≤ 1. Since µ is σ-finite, there is a measurable
χB subset B ⊂ A such that 0 < µ (B) < ∞. Thus, the element g := θ µ(B) is a function in the unit ball of L1 (µ) and satisfies the inequality
Z Z Z ∗ ∗ ∗ ∗ ∗ (1.2) kx k∞ = x gdµ ≤ |x g| dµ ≤ kx k∞ |g| dµ = kx k∞ . Ω Ω Ω
∗ Conversely, assume that x 6= 0 attains its norm at g ∈ SL1 (µ). Then we have that
∗ ∗ ∗ both x and g verify equation (1.2). Hence we obtain that |x (t) g (t)| = kx k∞ |g (t)| almost everywhere. If we set C = {t ∈ Ω: g (t) 6= 0}, then C is measurable and µ (C) > 0.
∗ ∗ Therefore, there is a subset Z ⊂ C such that µ (Z) = 0 and |x (t)| = kx k∞ for all t ∈ A := C \ Z. 14
Theorem 1.1.13 Assume that (Ω, Σ, µ) is a σ-finite measure space such that L1 (µ) is
∗ infinite dimensional. Then, both NA (L1 (µ)) and L1 (µ) \ NA (L1 (µ)) are spaceable. If, in
∗ ∗ addition, (Ω, Σ, µ) is atomless, then L1 (µ) \ NA (L1 (µ)) is dense in L1 (µ) .
Proof Let us first show the spaceability of NA (L1 (µ)). Since L1 (µ) is infinite dimensional, there exists a countable family (An) of pairwise disjoint measurable sets such that, n∈N
for every n ∈ N, µ (An) > 0. Now, consider the infinite dimensional closed subspace P∞ M := αnχA :(αn) ∈ c0 . Observe that M is a closed subspace of L∞ (µ) linearly n=1 n n∈N
isometric to c0. Now, if (αn) ∈ c0 then there exists n ∈ such that |αn| = (αn) = n∈N N n∈N P∞ P∞ k n=1 αnχAn k, and if t ∈ An then | n=1 αnχAn (t)| = |αn|.
∗ Let us second show the spaceability of L1 (µ) \ NA (L1 (µ)). Since L1 (µ) is infinite
dimensional, there exists a countable family (An) of pairwise disjoint measurable sets n∈N S∞ k k such that, for each n ∈ , µ (An) > 0 and An = Bn, where Bn is a sequence of N k=1 k∈N pairwise disjoint measurable sets of positive measure. Next, consider the vector subspace of L∞ (µ) defined as
( ∞ ∞ ! ) X X M := x (n) α (k) χ k : x ∈ ` , Bn ∞ n=1 k=1 where α := (α (k)) is a fixed strictly increasing sequence of real positive numbers such k∈N ∗ P∞ P∞ that kαk = 1. If x = x (n) α (k) χ k is an element of M, then ∞ n=1 k=1 Bn
∗ ∗ kx k∞ = sup {kx χAn k∞ : n ∈ N}
n ∗ o = sup x χBk : n, k ∈ N n ∞
= sup {|x (n) α (k)| : n, k ∈ N}
= sup {|x (n)| : n ∈ N}
= kxk∞ . 15
∗ ∗ S For 0 6= x ∈ M and t ∈ Ω, we clearly have that x (t) = 0 for all t∈ / An. If t ∈ An n∈N k ∗ for some n ∈ N, then there is k ∈ N such that x ∈ Bn and so we have that |x (t)| =
∗ ∗ ∗ |x (n) α (k)| < kxk∞ = kx k∞. Therefore for every t ∈ Ω, we have that |x (t)| < kx k∞,
∗ and thus x cannot attain its norm on L1 (µ).
∗ ∗ We now turn to density of L1 (µ) \ NA (L1 (µ)). We take x ∈ L∞ (µ) and r > 0. We will show that the ball centered at x∗ of radius r contains a functional that does not attain its norm. We can clearly assume that x∗ attains its norm and x∗ 6= 0, and hence there is
∗ ∗ a measurable set A ⊂ Ω with µ (A) > 0 such that |x (t)| = kx k∞ for all t ∈ A. Since µ S is atomless, we can write A = An, where (An) is a sequence of pairwise disjoint n∈N n∈N measurable sets such that µ (An) > 0 for each n ∈ N. Next, we choose a strictly increasing convergent sequence (rn) of real numbers such that rn ≥ 1 for all n ∈ and whose limit n∈N N r ∗ l satisfies l ≤ 1 + ∗ . We will denote by y the element of L∞ (µ) given by kx k∞
∞ ∗ X ∗ ∗ y = rnχAn x + (1 − χA) x , n=1 where the above convergence is pointwise. Then
∗ ∗ ∗ ky k∞ = max {ky χAk∞ , k(1 − χA) y k∞}
∗ ∗ ≤ max {sup {rn : n ∈ N} kx k∞ , kx k∞}
∗ = l kx k∞ .
∗ ∗ ∗ ∗ ∗ In fact, rn kx k∞ = krnχAn x k∞ ≤ ky k∞ for all n ∈ N, so l kx k∞ ≤ ky k∞, and we
∗ ∗ have that ky k = l kx k . Since the sequence (rn) is strictly increasing and since the ∞ n∈N essential supremum is not attained at any measurable set with positive measure, y∗ does 16
not attain its norm at L1 (µ). Also,
∞ X ky∗ − x∗k = (r − 1) χ x∗ ∞ n An n=1 ∞ ∗ ≤ sup {rn − 1 : n ∈ N} kx k∞
∗ ≤ kx k∞ (l − 1)
≤ r.
∗ ∗ ∗ As a consequence, we have checked that y ∈ x + rBL1(µ) and y does not attain its norm.
Remark 1.1.14 If the measure space (Ω, Σ, µ) is σ-finite and has an atom A of finite
∗ ∗ 1 ∗ measure, then for all x ∈ L∞ (µ) such that kx − χAk∞ < 2 , we have that x is norm- ∗ attaining. As a consequence, L1 (µ) \ NA (L1 (µ)) is not dense.
Note that these previous results motivate the following question.
Question 1.1.15 Is the set of non-norm-attaining functionals on a non-reflexive Banach space always lineable?
1.2 Lineability of NA (X)
We will begin by providing positive results, in other words, sufficient conditions for the set NA (X) to be lineable (sometimes under renormings.)
In the first place, we have the following result.
Proposition 1.2.1 Let X be an infinite dimensional Banach space. Then, for every n ∈ N, X can be equivalently renormed so that NA (X) is n-lineable. 17
Proof Observe that X is isomorphic to a Banach space of the form Y = H ⊕2 M where
∗ ∗ ∗ H a Hilbert space of dimension n and M is a Banach space. Next, Y = H ⊕2 M and
H∗ ⊆ NA (Y ).
Theorem 1.2.2 Let X be a separable Banach space with a Markushevich basis (en) . n∈N Then:
1. If (en) is monotone, then NA (X) is lineable. n∈N
2. If (en) is a monotone and shrinking Schauder basis, then NA (X) is dense-lineable. n∈N
Proof
1. We will show that every vector in
( n ) X ∗ M = αkek : αk ∈ K for 1 ≤ k ≤ n and n ∈ N k=1
is norm-attaining, where (e∗ ) is the sequence of biorthogonal functionals associated n n∈N
to (en) . Notice that, since the basis is monotone, n∈N n n ! X X α e∗ = α e∗ | , k k k k span{e1,...,en} k=1 k=1
and hence every element of M is norm-attaining.
2. Note that, if (en) is a monotone and shrinking Schauder basis, then the sequence n∈N of biorthogonal functionals (e∗ ) in X∗ is a Schauder basis for the dual X∗, and n n∈N hence M is dense in X∗. 18
Observe that the previous theorem says that any example of an infinite dimensional separable Banach space such that the set of norm-attaining functionals on it is not lineable is an example of a separable space admitting no monotone Markushevich basis. On the other hand, following the line proposed by Theorem 1.2.2, we can try to see what happens for Banach spaces admitting an infinite dimensional separable quotient.
Theorem 1.2.3 Let X be a Banach space. Assume that M is a closed subspace of X.
Then:
1. If M is 1-complemented and NA (M) is lineable (spaceable) then NA (X) is lineable
(spaceable.)
2. If M is proximinal and NA (X/M) is lineable (spaceable) then NA (X) is lineable
(spaceable.)
Proof
1. Let p : X −→ X be a linear projection of norm 1 so that p (X) = M. If Y ⊆
NA (M) is an infinite dimensional vector space then Z := {f ◦ p : f ∈ Y } is an infinite
dimensional vector space contained in NA (X). Indeed, since kpk = 1 and every f ∈ Y
is norm-attaining, we have that kfk = kf ◦ pk and f ◦ p is norm-attaining for every
f ∈ Y (at the same point at which f is.) Finally, if Y is closed then Z is closed too.
Indeed, if (fn ◦ p) is a sequence converging to g, then (fn) is Cauchy (because n∈N n∈N
kfnk = kfn ◦ pk,) therefore it converges to some f ∈ Y , and hence g = f ◦ p.
2. Let Y ⊆ NA (X/M) be an infinite dimensional vector space. Then we have that Z :=
{f ◦ p : f ∈ Y } is an infinite dimensional vector space contained in NA (X), where p
denotes the quotient map from X onto X/M. Indeed, since M is proximinal and every 19
f ∈ Y is norm-attaining, we have that kfk = kf ◦ pk and f ◦ p is norm-attaining for
every f ∈ Y . Finally, if Y is closed then Z is closed too. Indeed, if (fn ◦ p) is a n∈N
sequence converging to g, then (fn) is Cauchy (because kfnk = kfn ◦ pk,) therefore n∈N it converges to some f ∈ Y , and hence g = f ◦ p.
Corollary 1.2.4 Let X be a Banach space. If X admits an infinite dimensional separable quotient, then X can be equivalently renormed so that NA (X) is lineable.
Proof According to [35], if X admits an infinite dimensional separable quotient, then X admits an infinite dimensional quotient X/M with a Schauder basis. Now, according to [22],
X/M can be endowed with an equivalent norm |·| so that X/M has a monotone Schauder
basis. Next, according to [9, Lemma 2.4], there exists an equivalent norm k|·|k on X which
coincides with the original norm on M, whose quotient norm on X/M is a positive multiple
of |·|, and which makes M proximinal. Finally, X with the norm k|·|k has a proximinal
subspace M so that NA (X/M) is lineable (by Theorem 1.2.2,) and so, by Theorem 1.2.3,
NA (X) is lineable.
Now, we will try to approach the converse of the previous corollary. To do this, the following will be helpful (see [9].)
Lemma 1.2.5 (Bandyopadhyay/Godefroy, 2005) Let X be a Banach space. Then:
1. If Y ⊆ NA (X) is a closed separable subspace not containing `1 then X has a quotient
isomorphic to Y ∗. 20
∗ 2. If BX∗ is sequentially ω -compact then every closed separable subspace Y ⊆ NA (X)
does not contain `1.
∗ Theorem 1.2.6 Let X be a Banach space such that BX∗ is sequentially ω -compact. If
NA (X) contains an infinite dimensional closed subspace with an unconditional basic se- quence then X admits an infinite dimensional separable quotient.
Proof Let Y ⊆ NA (X) be a closed separable subspace with an unconditional basic se- quence. Then, we have three possibilities for Y :
1. c0 ⊆ Y . In this case, by Lemma 1.2.5, we have that X has a quotient isomorphic to
`1.
2. `1 ⊆ Y . This is impossible if we take into account Lemma 1.2.5.
3. There exists an infinite dimensional reflexive Banach space R contained in Y . Then,
take any closed separable subspace W of R. Then, W is reflexive and it does not
∗ contain `1. Again by Lemma 1.2.5 we have that X has a quotient isomorphic to W
which is separable.
At this moment, we will concentrate on providing negative results, in other words, on the non-lineability of NA (X). As we will see, the hypothesis of smoothness will be crucial for the development of this section, and everything here will be based upon the following fact.
Remark 1.2.7 Let X be a smooth Banach space. If x∗ 6= y∗ ∈ X∗ \{0} are so that
∗ ∗ kx∗k = ky∗k = x +y , then x∗ + y∗ cannot be norm-attaining. As a consequence, every 2 21
element of NA (X) ∩ SX∗ is an extreme point of BX∗ .
A direct consequence of this remark is the next sufficient condition.
Theorem 1.2.8 Let X be a smooth Banach space. If X∗ does not contain rotund subspaces then NA (X) is not even 2-lineable.
Remark 1.2.9 Observe that the Banach space c0 almost verifies the conditions of the pre-
vious theorem, because according to the Mazur Theorem (see [33]) we have that smo (Bc0 )
is a Gδ dense subset of Sc0 . Besides, c0 is also closed to verify the thesis of Theorem 1.2.8
since rot (B`1 ) = ∅.
1.3 Lineability of X∗ \ NA (X)
In this section we will show that the answer to Question 1.1.15 is negative. Nevertheless,
we will begin by presenting some positive results, in other words, by showing conditions for
X∗ \ NA (X) to be spaceable.
∗ ∗ Remark 1.3.1 Let X be a smooth Banach space. If x ∈ NA (X) ∩ SX∗ then x is not only
∗ an extreme point of BX∗ but a (ω -strongly) exposed point.
A direct consequence of this remark is the next “hint.”
Proposition 1.3.2 Let X be a smooth Banach space. If Y is a vector subspace of X∗ so
∗ that Y ∩ exp (BX∗ ) = ∅, then Y ⊆ X \ NA (X) ∪ {0}.
Lemma 1.3.3 Let X be a Banach space with a Schauder basis (en) . Then, the set n∈N
X \ {±en : n ∈ N} is spaceable. 22
Proof It can be checked that
span {e1 + e2, e3 + e4, e5 + e6,... } ⊆ X \ {±en : n ∈ N} .
Lemma 1.3.4 Let K be an infinite Hausdorff compact topological space. Let X be a rotund Banach space. Then, C (K,X) \ ext BC(K,X) is spaceable.
Proof Let us fix an arbitrary s ∈ K. Then, consider the continuous linear operator
δs : C (K,X) −→ X
f 7−→ f (s) .
Since δs is a non-zero operator and C (K,X) is infinite dimensional, we deduce that ker (δs) is an infinite dimensional closed subspace of C (K,X). Now, according to [15],