Universality of Some C ∗-Algebra Generated by a Unitary and a Self

Total Page:16

File Type:pdf, Size:1020Kb

Universality of Some C ∗-Algebra Generated by a Unitary and a Self Universality of some C∗-algebra generated by a unitary and a self-adjoint idempotent Dedicated to Nikolai Vasilevski on the occasioin of his 60th birthday Helena Mascarenhas and Bernd Silbermann1 Keywords: C∗-algebra, nitely generated, universality Abstract We prove that there is essentially only one C∗-algebra generated by a unitary element u and a self-adjoint idempotent p such that up = pup and up 6= pu . This result is related to a theorem of L. Coburn stating that there is essentially only one C∗-algebra generated by a non-unitary isometry. 1 Introduction A theorem of L. Coburn [?] tells that the C∗-algebra generated by a non- unitary isometry v (that is v∗v = e, vv∗ 6= e) is universal in the sense that any two C∗-algebras with this property are isometrically isomorphic. An instructive example for such a C∗-algebra is given as follows: Let H2 ⊂ L2(T), T := {z ∈ C : |z| = 1}, be the familiar Hardy space and P + : L2(T) → H2 the Riesz projection which is known to be selfadjoint and surjective. Given a ∈ L∞(T), dene the operator T (a): H2 → H2 , f 7→ P +af . The operator T (a) is clearly bounded, and is called a Toeplitz operator. Let be the functions ±1 . Then ∗ and χ±1 χ±1(t) = t , t ∈ T T (χ1) = T (χ−1) ∗ ∗ T (χ1)T (χ1) = I, but T (χ1)T (χ1) 6= I. Hence, T (χ1) is a non-unitary isometry. It is well-known that the smallest C∗-algebra T (C) ⊂ B(H2) 1Mathematics Subject Classication (2000) 46L05 1 generated by T (χ1) (B(X) stands for the Banach algebra of all bounded linear operators acting on the Banach space X) contains all Toeplitz operators T (a), with a ∈ C(T), where C(T) denotes the algebra of all continuous functions on T. Moreover, 2 T (C) = {T (a): a ∈ C(T)}+˙ K(H ) , where K(H2) denotes the ideal of all compact operators acting on H2 (see for instance [?], Corollary 4.15 and Theorem 4.24). Clearly, is a unitary operator on 2 ∗ , and + is a χ1I L (T), (χ1I) = χ−1I P selfadjoint projection. It is easy to see that + + + + + χ1P = P χ1P and χ1P 6= P χ1I. The smallest ∗-subalgebra 2 containing and + con- C SO(C) ⊂ B(L (T)) χ1I P tains all singular integral operators A = aP + + bP − with P − := I − P + and a, b ∈ C(T). Moreover, + + − − 2 SO(C) = {P aP + P bP : a, b ∈ C(T)}+˙ K(L (T)) , where K(L2(T)) is again the ideal of all compact operators acting on L2(T) (see [?], Corollary 4.76; notice that this corollary is applicable in our case, and that QC(U) can be identied with K(L2(T))). These considerations show that T (C) is a subalgebra of SO(C). The close relationship between Toeplitz and singular integral operators gives rise to the question whether SO(C) is a model for a universal C∗-algebra. We will show that this is indeed the case. Our proof is based on features known from the theory of singular integral operators with continuous coecients. This theory is developed in many text books, and we will use mainly [?] and [?] because the approach given there is mostly convenient for us. Let us also notice that the study of nitely generated algebras is an important task. More about this topic can be found, in particular, in [?] and [?]. 2 2 The main result and its proof Let A be any C∗-algebra generated by a unitary element u (that is uu∗ = u∗u = e) and by a selfadjoint idempotent p+ such that up+ = p+up+ und up+ 6= p+u . (1) Passing to adjoints and using u−1 = u∗ yields p+u−1 = p+u−1p+ , (2) and thus u−1p− = p−u−1p− , where p− := e − p+. We denote by P the class of all C∗-algebras of this type. ∗ P Theorem: Let A1 and A2 be arbitrary C -algebras belonging to . Then these algebras are isometrically isomorphic. This isomorphism can be chosen so that u1, p1 are taken into u2, p2, respectively. Proof: Due to the GNS-construction we may assume (without loss of gen- erality) that there is a Hilbert space H such that for A ∈ P the elements u and p+ are operators on H which generate A. Let q ∈ C(T) be an arbitrarily given quasipolynomial, k k X j and form X j q(t) = ajt , aj ∈ C q(u) = aju . j=−k j=−k We call q(u) ∈ A a quasipolynomial of u and let L0(u) stand for the (non- closed) algebra of all quasipolynomials of u, and let L(u) be the closure of L0(u) in B(H). Since u is a unitary element, we know that −1 sp(u) = sp(u ) = T . The general theory of commutative C∗-algebras entails that L(u) is isomet- rically isomorphic to , and the isomorphism takes into : C(T) u χ1 ∼ L(u) = C(T) . 3 Now let A ∈ P be generated by u and p+. Introduce the (dense) subalgebra Z of A: ( M N ) X Y + − Z := (avs(u)p + bvs(u)p ): avs(u), bvs(u) ∈ L0(u), M, N ∈ N . v=1 s=1 Take an element c ∈ Z, say M N X Y + − c := (avs(u)p + bvs(u)p ) v=1 s=1 and suppose that it is invertible in A. We would like to express its invertibility in terms which can be used for further analysis. For this aim we proceed as follows: Given let r stand for . r ∈ N H {(h1, . , hr): hj ∈ H, j = 1, . , r} This linear space becomes a Hilbert space by introducing the scalar product r X h(h1, . , hr), (g1, . , gr)i = hhj, gji , j=1 and B(Hr) can be identied with (B(H))r×r in a natural way. Then form a linear extension c˜ of c exactly as it is done in [?], Chapter VIII, 10. The properties of c˜ are (and only these are needed): 1. c˜ ∈ (B(H))r×r for some r ∈ N and the entries of c˜ are among the + − + − elements O, I, a11(u)p + b11(u)p , . , aMN (u)p + bMN (u)p . 2. c is invertible in B(H) (and therefore in A by the inverse closedness property of unital C∗-subalgebras) if and only if c˜ is invertible in B(Hr). Now observe that c˜ can be written by 1. as + + − − c˜ = A1(u)diag{p , . , p } + A2(u)diag{p , . , p } , | {z } | {z } r−times r−times where A1(u),A2(u) are r×r-matrices with entries from L0(u). Now we apply the theory of [?], Chapter VIII, 8 (the condition dim coker (u|imP +) < ∞, which is formulated there, can be droped down because we are only interested in invertibility): 4 a) the invertibility of c˜ implies the invertibility of A1(u) and A2(u), that is for all detA1(t) 6= 0, detA2(t) 6= 0 t ∈ T , where A1(t) and A2(t) are the matrices whose entries are the Gelfand transforms of the entries of A1(u) and A2(u), respectively. r×r b) Notice that the functions A1(t),A2(t) belong to the class W , where W is the Wiener algebra. Moreover, the invertibility of c˜ implies that the right canonical Wiener-Hopf factorization of −1 t 7→ A2 (t)A1(t)(∈ W r×r) exists with all partial indices equal to zero. Conversely, if this condition is fullled then c˜, and thus also c, is invertible. Notice that this fact is true independently of the choice of A. Now take two algebras P A1 and A2 from . The element c and the algebra Z are denoted now by c1, c2 and Z1,Z2, respectively. Further, we dene a map ψ : Z1 → Z2, c1 7→ c2 and we need to show that ψ is correctly dened. Using that c1 is invertible if and only if c2 is invertible (in Ai) then by the above argument, we get sp(c1) = sp(c2) . Because ∗ ∗ , we therefore obtain c1c1 ∈ Z1, c2c2 ∈ Z2 ∗ ∗ sp(c1c1) = sp(c2c2) , and using 2 ∗ ∗ 2, that kc1k = kc1c1k = kc2c2k = kc2k kc1k = kc2k . (Recall that for a selfadjoint element the norm and the spectral radius coincide.) This equality shows that ψ is correctly dened and that ψ represents an isometric isomorphism between Z1 and Z2. Moreover, + +. The continuous extension of to the whole of ψu1 = u2, ψp1 = p2 ψ A1 provides us with the wanted isomorphism and nishes the proof. Remark: This proof can easily be adapted to prove Coburns result which was originally proved by dierent methods. Acknowledgement. First author partially supported by CEAF, IST, Tech- nical Univ. of Lisbon, and Fundação para a Ciência e a Tecnologia through the program FCT/FEDER/POCTI/MAT/59972/2004. 5 References [1] L. Coburn: The C∗-algebra generated by an isometry. Bull. Amer. Soc. 73 (1967), 722-726. [2] I. Gohberg, I. Feldman: Convolution Equations and Projection Methods for Their Solution. Nauka, Moskva 1971 (Russian) [Engl. translation: Transl. of Math. Monographs 41, Amer. Math. Soc., Providence, R.I., 1974]. [3] S. Pröÿdorf, B. Silbermann: Numerical Analysis for Integral and Related Operator Equations. Akademie-Verlag Berlin 1991, and Birkhäuser Verlag, Basel-Boston-Stuttgart, 1991. [4] V. Ostrovskyi, Y. Samoilenko: Introduction to the theory of representations of nitely presented ∗-algebras. I. Representations by bounded operators, Harwood Academic Publishers, Amsterdam, 1999. [5] N. Vasilevskii: C∗-algebras generated by orthogonal projections and their applications. IEOT 31 (1998), no. 1, 113-132.
Recommended publications
  • Can One Identify Two Unital JB $^* $-Algebras by the Metric Spaces
    CAN ONE IDENTIFY TWO UNITAL JB∗-ALGEBRAS BY THE METRIC SPACES DETERMINED BY THEIR SETS OF UNITARIES? MAR´IA CUETO-AVELLANEDA, ANTONIO M. PERALTA Abstract. Let M and N be two unital JB∗-algebras and let U(M) and U(N) denote the sets of all unitaries in M and N, respectively. We prove that the following statements are equivalent: (a) M and N are isometrically isomorphic as (complex) Banach spaces; (b) M and N are isometrically isomorphic as real Banach spaces; (c) There exists a surjective isometry ∆ : U(M) →U(N). We actually establish a more general statement asserting that, under some mild extra conditions, for each surjective isometry ∆ : U(M) → U(N) we can find a surjective real linear isometry Ψ : M → N which coincides with ∆ on the subset eiMsa . If we assume that M and N are JBW∗-algebras, then every surjective isometry ∆ : U(M) → U(N) admits a (unique) extension to a surjective real linear isometry from M onto N. This is an extension of the Hatori–Moln´ar theorem to the setting of JB∗-algebras. 1. Introduction Every surjective isometry between two real normed spaces X and Y is an affine mapping by the Mazur–Ulam theorem. It seems then natural to ask whether the existence of a surjective isometry between two proper subsets of X and Y can be employed to identify metrically both spaces. By a result of P. Mankiewicz (see [34]) every surjective isometry between convex bodies in two arbitrary normed spaces can be uniquely extended to an affine function between the spaces.
    [Show full text]
  • The Index of Normal Fredholm Elements of C* -Algebras
    proceedings of the american mathematical society Volume 113, Number 1, September 1991 THE INDEX OF NORMAL FREDHOLM ELEMENTS OF C*-ALGEBRAS J. A. MINGO AND J. S. SPIELBERG (Communicated by Palle E. T. Jorgensen) Abstract. Examples are given of normal elements of C*-algebras that are invertible modulo an ideal and have nonzero index, in contrast to the case of Fredholm operators on Hubert space. It is shown that this phenomenon occurs only along the lines of these examples. Let T be a bounded operator on a Hubert space. If the range of T is closed and both T and T* have a finite dimensional kernel then T is Fredholm, and the index of T is dim(kerT) - dim(kerT*). If T is normal then kerT = ker T*, so a normal Fredholm operator has index 0. Let us consider a generalization of the notion of Fredholm operator intro- duced by Atiyah. Let X be a compact Hausdorff space and consider continuous functions T: X —>B(H), where B(H) is the set of bounded linear operators on a separable infinite dimensional Hubert space with the norm topology. The set of such functions forms a C*- algebra C(X) <g>B(H). A function T is Fredholm if T(x) is Fredholm for each x . Atiyah [1, Appendix] showed how such an element has an index which is an element of K°(X). Suppose that T is Fredholm and T(x) is normal for each x. Is the index of T necessarily 0? There is a generalization of this question that we would like to consider.
    [Show full text]
  • A Short Introduction to the Quantum Formalism[S]
    A short introduction to the quantum formalism[s] François David Institut de Physique Théorique CNRS, URA 2306, F-91191 Gif-sur-Yvette, France CEA, IPhT, F-91191 Gif-sur-Yvette, France [email protected] These notes are an elaboration on: (i) a short course that I gave at the IPhT-Saclay in May- June 2012; (ii) a previous letter [Dav11] on reversibility in quantum mechanics. They present an introductory, but hopefully coherent, view of the main formalizations of quantum mechanics, of their interrelations and of their common physical underpinnings: causality, reversibility and locality/separability. The approaches covered are mainly: (ii) the canonical formalism; (ii) the algebraic formalism; (iii) the quantum logic formulation. Other subjects: quantum information approaches, quantum correlations, contextuality and non-locality issues, quantum measurements, interpretations and alternate theories, quantum gravity, are only very briefly and superficially discussed. Most of the material is not new, but is presented in an original, homogeneous and hopefully not technical or abstract way. I try to define simply all the mathematical concepts used and to justify them physically. These notes should be accessible to young physicists (graduate level) with a good knowledge of the standard formalism of quantum mechanics, and some interest for theoretical physics (and mathematics). These notes do not cover the historical and philosophical aspects of quantum physics. arXiv:1211.5627v1 [math-ph] 24 Nov 2012 Preprint IPhT t12/042 ii CONTENTS Contents 1 Introduction 1-1 1.1 Motivation . 1-1 1.2 Organization . 1-2 1.3 What this course is not! . 1-3 1.4 Acknowledgements . 1-3 2 Reminders 2-1 2.1 Classical mechanics .
    [Show full text]
  • Geodesics of Projections in Von Neumann Algebras
    Geodesics of projections in von Neumann algebras Esteban Andruchow∗ November 5, 2020 Abstract Let be a von Neumann algebra and A the manifold of projections in . There is a A P A natural linear connection in A, which in the finite dimensional case coincides with the the Levi-Civita connection of theP Grassmann manifold of Cn. In this paper we show that two projections p, q can be joined by a geodesic, which has minimal length (with respect to the metric given by the usual norm of ), if and only if A p q⊥ p⊥ q, ∧ ∼ ∧ where stands for the Murray-von Neumann equivalence of projections. It is shown that the minimal∼ geodesic is unique if and only if p q⊥ = p⊥ q =0. If is a finite factor, any ∧ ∧ A pair of projections in the same connected component of A (i.e., with the same trace) can be joined by a minimal geodesic. P We explore certain relations with Jones’ index theory for subfactors. For instance, it is −1 shown that if are II1 factors with finite index [ : ] = t , then the geodesic N ⊂M M N 1/2 distance d(eN ,eM) between the induced projections eN and eM is d(eN ,eM) = arccos(t ). 2010 MSC: 58B20, 46L10, 53C22 Keywords: Projections, geodesics of projections, von Neumann algebras, index for subfac- tors. arXiv:2011.02013v1 [math.OA] 3 Nov 2020 1 Introduction ∗ If is a C -algebra, let A denote the set of (selfadjoint) projections in . A has a rich A P A P geometric structure, see for instante the papers [12] by H.
    [Show full text]
  • Noncommutative Geometry and Flavour Mixing
    Noncommutative geometry and flavour mixing prepared by Jose´ M. Gracia-Bond´ıa Department of Theoretical Physics Universidad de Zaragoza, 50009 Zaragoza, Spain October 22, 2013 1 Introduction: the universality problem “The origin of the quark and lepton masses is shrouded in mystery” [1]. Some thirty years ago, attempts to solve the enigma based on textures of the quark mass matrices, purposedly reflecting mass hierarchies and “nearest-neighbour” interactions, were very popular. Now, in the late eighties, Branco, Lavoura and Mota [2] showed that, within the SM, the zero pattern 0a b 01 @c 0 dA; (1) 0 e 0 a central ingredient of Fritzsch’s well-known Ansatz for the mass matrices, is devoid of any particular physical meaning. (The top quark is above on top.) Although perhaps this was not immediately clear at the time, paper [2] marked a water- shed in the theory of flavour mixing. In algebraic terms, it establishes that the linear subspace of matrices of the form (1) is universal for the group action of unitaries effecting chiral basis transformations, that respect the charged-current term of the Lagrangian. That is, any mass matrix can be transformed to that form without modifying the corresponding CKM matrix. To put matters in perspective, consider the unitary group acting by similarity on three-by- three matrices. The classical triangularization theorem by Schur ensures that the zero patterns 0a 0 01 0a b c1 @b c 0A; @0 d eA (2) d e f 0 0 f are universal in this sense. However, proof that the zero pattern 0a b 01 @0 c dA e 0 f 1 is universal was published [3] just three years ago! (Any off-diagonal n(n−1)=2 zero pattern with zeroes at some (i j) and no zeroes at the matching ( ji), is universal in this sense, for complex n × n matrices.) Fast-forwarding to the present time, notwithstanding steady experimental progress [4] and a huge amount of theoretical work by many authors, we cannot be sure of being any closer to solving the “Meroitic” problem [5] of divining the spectrum behind the known data.
    [Show full text]
  • Group of Isometries of the Hilbert Ball Equipped with the Caratheodory
    GROUP OF ISOMETRIES OF HILBERT BALL EQUIPPED WITH THE CARATHEODORY´ METRIC MUKUND MADHAV MISHRA AND RACHNA AGGARWAL Abstract. In this article, we study the geometry of an infinite dimensional Hyperbolic space. We will consider the group of isometries of the Hilbert ball equipped with the Carath´eodory metric and learn about some special subclasses of this group. We will also find some unitary equivalence condition and compute some cardinalities. 1. Introduction Groups of isometries of finite dimensional hyperbolic spaces have been studied by a number of mathematicians. To name a few Anderson [1], Chen and Greenberg [5], Parker [14]. Hyperbolic spaces can largely be classified into four classes. Real, complex, quaternionic hyperbolic spaces and octonionic hyperbolic plane. The respective groups of isometries are SO(n, 1), SU(n, 1), Sp(n, 1) and F4(−20). Real and complex cases are standard and have been discussed at various places. For example Anderson [1] and Parker [14]. Quaternionic spaces have been studied by Cao and Parker [4] and Kim and Parker [10]. For octonionic hyperbolic spaces, one may refer to Baez [2] and Markham and Parker [13]. The most basic model of the hyperbolic space happens to be the Poincar´e disc which is the unit disc in C equipped with the Poincar´e metric. One of the crucial properties of this metric is that the holomorphic self maps on the unit ball satisfy Schwarz-Pick lemma. Now in an attempt to generalize this lemma to higher dimensions, Carath´eodory and Kobayashi metrics were discovered which formed one of the ways to discuss hyperbolic structure on domains in Cn.
    [Show full text]
  • Approximation by Unitary and Essentially Unitary Operators
    Acta Sci. Math., 39 (1977), 141—151 Approximation by unitary and essentially unitary operators DONALD D. ROGERS In troduction. In [9] P. R. HALMOS formulated the problem of normal spectral approximation in the algebra of bounded linear operators on a Hilbert space. One special case of this problem is the problem of unitary approximation; this case has been studied in [3], [7, Problem 119], and [13]. The main purpose of this paper is to continue this study of unitary approximation and some related problems. In Section 1 we determine the distance (in the operator norm) from an arbitrary operator on a separable infinite-dimensional Hilbert space to the set of unitary operators in terms of familiar operator parameters. We also study the problem of the existence of unitary approximants. Several conditions are given that are sufficient for the existence of a unitary approximant, and it is shown that some operators fail to have a unitary approximant. This existence problem is solved completely for weighted shifts and compact operators. Section 2 studies the problem of approximation by two sets of essentially unitary operators. It is shown that both the set of compact perturbations of unitary operators and the set of essentially unitary operators are proximinal; this latter fact is shown to be equivalent to the proximinality of the unitary elements in the Calkin algebra. Notation. Throughout this paper H will denote a fixed separable infinite- dimensional complex Hilbert space and B(H) the algebra of all bounded linear operators on H. For an arbitrary operator T, we write j|T"[| = sup {||Tf\\ :f in //and ||/|| = 1} and m{T)=mi {|| 7/11:/ in H and ||/|| = 1}.
    [Show full text]
  • Fundamental Theorems in Mathematics
    SOME FUNDAMENTAL THEOREMS IN MATHEMATICS OLIVER KNILL Abstract. An expository hitchhikers guide to some theorems in mathematics. Criteria for the current list of 243 theorems are whether the result can be formulated elegantly, whether it is beautiful or useful and whether it could serve as a guide [6] without leading to panic. The order is not a ranking but ordered along a time-line when things were writ- ten down. Since [556] stated “a mathematical theorem only becomes beautiful if presented as a crown jewel within a context" we try sometimes to give some context. Of course, any such list of theorems is a matter of personal preferences, taste and limitations. The num- ber of theorems is arbitrary, the initial obvious goal was 42 but that number got eventually surpassed as it is hard to stop, once started. As a compensation, there are 42 “tweetable" theorems with included proofs. More comments on the choice of the theorems is included in an epilogue. For literature on general mathematics, see [193, 189, 29, 235, 254, 619, 412, 138], for history [217, 625, 376, 73, 46, 208, 379, 365, 690, 113, 618, 79, 259, 341], for popular, beautiful or elegant things [12, 529, 201, 182, 17, 672, 673, 44, 204, 190, 245, 446, 616, 303, 201, 2, 127, 146, 128, 502, 261, 172]. For comprehensive overviews in large parts of math- ematics, [74, 165, 166, 51, 593] or predictions on developments [47]. For reflections about mathematics in general [145, 455, 45, 306, 439, 99, 561]. Encyclopedic source examples are [188, 705, 670, 102, 192, 152, 221, 191, 111, 635].
    [Show full text]
  • Approximation with Normal Operators with Finite Spectrum, and an Elementary Proof of a Brown–Douglas–Fillmore Theorem
    Pacific Journal of Mathematics APPROXIMATION WITH NORMAL OPERATORS WITH FINITE SPECTRUM, AND AN ELEMENTARY PROOF OF A BROWN–DOUGLAS–FILLMORE THEOREM Peter Friis and Mikael Rørdam Volume 199 No. 2 June 2001 PACIFIC JOURNAL OF MATHEMATICS Vol. 199, No. 2, 2001 APPROXIMATION WITH NORMAL OPERATORS WITH FINITE SPECTRUM, AND AN ELEMENTARY PROOF OF A BROWN–DOUGLAS–FILLMORE THEOREM Peter Friis and Mikael Rørdam We give a short proof of the theorem of Brown, Douglas and Fillmore that an essentially normal operator on a Hilbert space is of the form “normal plus compact” if and only if it has trivial index function. The proof is basically a modification of our short proof of Lin’s theorem on almost commuting self- adjoint matrices that takes into account the index. Using similar methods we obtain new results, generalizing results of Lin, on approximating normal operators by ones with finite spectrum. 1. Introduction. Let H be an infinite-dimensional separable Hilbert space, let K denote the compact operators on H, and consider the short-exact sequence 0 −−−→ K −−−→ B(H) −−−→π Q(H) −−−→ 0 where Q(H) is the Calkin algebra B(H)/K. An operator T ∈ B(H) is essentially normal if T ∗T − TT ∗ ∈ K, or equivalently, if π(T ) is normal. An operator T ∈ B(H) is Fredholm if π(T ) is invertible in Q(H), and it’s Fredholm index is denoted by index(T ). The essential spectrum spess(T ) is the spectrum of π(T ). The index function of T is the map C \ spess(T ) → Z; λ 7→ index(T − λ·1).
    [Show full text]
  • Von Neumann Algebras
    1 VON NEUMANN ALGEBRAS ADRIAN IOANA These are lecture notes from a topics graduate class taught at UCSD in Winter 2019. 1. Review of functional analysis In this section we state the results that we will need from functional analysis. All of these are stated and proved in [Fo99, Chapters 4-7]. Convention. All vector spaces considered below are over C. 1.1. Normed vector spaces. Definition 1.1. A normed vector space is a vector space X over C together with a map k · k : X ! [0; 1) which is a norm, i.e., it satisfies that • kx + yk ≤ kxk + kyk, for al x; y 2 X, • kαxk = jαj kxk, for all x 2 X and α 2 C, and • kxk = 0 , x = 0, for all x 2 X. Definition 1.2. Let X be a normed vector space. (1) A map ' : X ! C is called a linear functional if it satisfies '(αx + βy) = α'(x) + β'(y), for all α; β 2 C and x; y 2 X. A linear functional ' : X ! C is called bounded if ∗ k'k := supkxk≤1 j'(x)j < 1. The dual of X, denoted X , is the normed vector space of all bounded linear functionals ' : X ! C. (2) A map T : X ! X is called linear if it satisfies T (αx+βy) = αT (x)+βT (y), for all α; β 2 C and x; y 2 X. A linear map T : X ! X is called bounded if kT k := supkxk≤1 kT (x)k < 1. A linear bounded map T is usually called a linear bounded operator, or simply a bounded operator.
    [Show full text]
  • Matrix Intersection Problems for Conditioning 1
    MATRIX INTERSECTION PROBLEMS FOR CONDITIONING MARKO HUHTANEN∗ AND OTTO SEISKARI † Abstract. Conditioning of a nonsingular matrix subspace is addressed in terms of its best conditioned elements. The problem is computationally challenging. Associating with the task an intersection problem with unitary matrices leads to a more accessible approach. A resulting matrix nearness problem can be viewed to generalize the so-called L¨owdin problem in quantum chemistry. For critical points in the Frobenius norm, a differential equation on the manifold of unitary matrices is derived. Another resulting matrix nearness problem allows locating points of optimality more directly, once formulated as a problem in computational algebraic geometry. Key words. conditioning, matrix intersection problem, matrix nearness problem, L¨owdin’s problem, generalized eigenvalue problem AMS subject classifications. 15A12, 65F35 1. Introduction. This paper is concerned with the problem of conditioning of a nonsingular matrix subspace of Cn×n over C (or R). Matrix subspaces typically appear in large scale numericalV linear algebra problems where assuming additional structure is quite unrealistic. Nonsingularity means that there exists invertible ele- ments in . The conditioning of is then defined in terms of its best conditioned elements.V In the applications thatV we have in mind, typically dim n2. For exam- ple, in the generalized eigenvalue problem dim = 2 only. In thisV paper ≪ the task of assessing conditioning is formulated as a matrixV intersection problem for and the set of unitary matrices.1 Since this can be done in many ways, the interpretationV is amenable to computations through matrix nearness problems and versatile enough in view of addressing operator theoretic problems more generally.
    [Show full text]
  • Variable-Coefficient Toeplitz Matrices with Symbols Beyond the Wiener
    Operator Theory: Advances and Applications, Vol. 199, 191–202 c 2009 Birkh¨auser Verlag Basel/Switzerland Variable-coefficient Toeplitz Matrices with Symbols beyond the Wiener Algebra Albrecht B¨ottcher and Sergei Grudsky In Memory of Georg Heinig Abstract. Sequences of so-called variable-coefficient Toeplitz matrices arise in many problems, including the discretization of ordinary differential equations with variable coefficients. Such sequences are known to be bounded if the generating function satisfies a condition of the Wiener type, which is far away from the minimal requirement in the case of constant coefficients. The purpose of this paper is to uncover some phenomena beyond the Wiener condition. We provide counterexamples on the one hand and prove easy-to-check sufficient conditions for boundedness on the other. Mathematics Subject Classification (2000). Primary 47B35; Secondary 15A60, 65F35. Keywords. Toeplitz matrix, variable coefficients, matrix norm. 1. Introduction Let a be a complex-valued continuous function on [0, 1] × [0, 1] × T,whereT is the complex unit circle, a :[0, 1] × [0, 1] × T → C, (x, y, t) → a(x, y, t). For n ∈ Z,we −n puta ˆn(x, y)= T a(x, y, t) t |dt|/(2π) and so have the Fourier series ∞ n a(x, y, t)= aˆn(x, y)t , (1) n=−∞ 2 where equality holds at least in the L sense. Let AN (a)bethematrix N j k AN (a)= aˆj−k , . (2) N N j,k=0 Occasionally we allow us to write a(x, y, t)andAN (a(x, y, t)) for a and AN (a). This work was partially supported by CONACYT project U46936-F, Mexico.
    [Show full text]