This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

On the essential norms of Toeplitz operators
on abstract Hardy spaces built upon Banach function spaces

Oleksiy Karlovych, Eugene Shargorodsky Oleksiy Karlovych
Centro de Matemática e Aplicações, Departamento de Matemática, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Quinta da Torre, 2829–516 Caparica, Portugal
oyk@fct.unl.pt Eugene Shargorodsky
Department of Mathematics, King’s College London, Strand, London WC2R 2LS, United Kingdom
eugene.shargorodsky@kcl.ac.uk To the memory of Nikolai Vasilevski
Abstract.

Let XX be a Banach function space over the unit circle such that the Riesz projection PP is bounded on XX and let H[X]H[X] be the abstract Hardy space built upon XX. We show that the essential norm of the Toeplitz operator T(a):H[X]H[X]T(a):H[X]\to H[X] coincides with aL\|a\|_{L^{\infty}} for every aC+Ha\in C+H^{\infty} if and only if the essential norm of the backward shift operator T(𝐞1):H[X]H[X]T(\mathbf{e}_{-1}):H[X]\to H[X] is equal to one, where 𝐞1(z)=z1\mathbf{e}_{-1}(z)=z^{-1}. This result extends an observation by Böttcher, Krupnik, and Silbermann for the case of classical Hardy spaces.

Key words and phrases:
Banach function space, abstract Hardy space, Toeplitz operator, essential norm
2000 Mathematics Subject Classification:
47B35, 46E30

1. Introduction and the main result

For a Banach space 𝒳\mathcal{X}, let (𝒳)\mathcal{B}(\mathcal{X}) denote the Banach algebra of bounded linear operators on 𝒳\mathcal{X} and let 𝒦(𝒳)\mathcal{K}(\mathcal{X}) be the closed two-sided ideal of (𝒳)\mathcal{B}(\mathcal{X}) consisting of all compact linear operators on 𝒳\mathcal{X}. The norm of an operator A(𝒳)A\in\mathcal{B}(\mathcal{X}) is denoted by A(𝒳)\|A\|_{\mathcal{B}(\mathcal{X})}. The essential norm of A(𝒳)A\in\mathcal{B}(\mathcal{X}) is defined as follows:

A(𝒳),e:=inf{AK(𝒳):K𝒦(𝒳)}.\|A\|_{\mathcal{B}(\mathcal{X}),\mathrm{e}}:=\inf\{\|A-K\|_{\mathcal{B}(\mathcal{X})}\ :\ K\in\mathcal{K}(\mathcal{X})\}.

For a function fL1f\in L^{1} on the unit circle 𝕋:={z:|z|=1}\mathbb{T}:=\left\{z\in\mathbb{C}:\ |z|=1\right\} equipped with the Lebesgue measure mm normalised so that m(𝕋)=1m(\mathbb{T})=1, let

f^(n)=12πππf(eiθ)einθ𝑑θ,n\widehat{f}(n)=\frac{1}{2\pi}\int_{-\pi}^{\pi}f\left(e^{i\theta}\right)e^{-in\theta}\,d\theta,\quad n\in\mathbb{Z}

be the Fourier coefficients of ff. Let XX be a Banach function space on the unit circle 𝕋\mathbb{T}. We postpone the definition of this notion until Section 2.1. Here we only mention that the class of Banach function spaces is very reach, it includes all Lebesgue spaces LpL^{p}, 1p1\leq p\leq\infty, Orlicz spaces LΦL^{\Phi} (see, e.g., [1, Ch. 4, Section 8]), and Lorentz spaces Lp,qL^{p,q}, 1<p<1<p<\infty, 1q1\leq q\leq\infty (see, e.g., [1, Ch. 4, Section 4]). Moreover, all mentioned above spaces are rearrangement-invariant (see Section 2.2 for their definition).

Let

H[X]:={gX:g^(n)=0for alln<0}H[X]:=\left\{g\in X\ :\ \widehat{g}(n)=0\quad\mbox{for all}\quad n<0\right\}

denote the abstract Hardy space built upon the space XX. In the case X=LpX=L^{p}, where 1p1\leq p\leq\infty, we will use the standard notation Hp:=H[Lp]H^{p}:=H[L^{p}]. Consider the operators SS and PP, defined for a function fL1f\in L^{1} and an a.e. point t𝕋t\in\mathbb{T} by

(Sf)(t):=1πip.v.𝕋f(τ)τt𝑑τ,(Pf)(t):=12(f(t)+(Sf)(t)),(Sf)(t):=\frac{1}{\pi i}\,\mbox{p.v.}\int_{\mathbb{T}}\frac{f(\tau)}{\tau-t}\,d\tau,\quad(Pf)(t):=\frac{1}{2}(f(t)+(Sf)(t)),

respectively, where the integral is understood in the Cauchy principal value sense. The operator SS is called the Cauchy singular integral operator and the operator PP is called the Riesz projection. Assume that the Riesz projection is bounded on XX. For aLa\in L^{\infty}, the Toeplitz operator with symbol aa is defined by

T(a)f=P(af),fH[X].T(a)f=P(af),\quad f\in H[X].

It is clear that T(a)(H[X])T(a)\in\mathcal{B}(H[X]) and

T(a)(H[X]),eT(a)(H[X])P(X)aL.\|T(a)\|_{\mathcal{B}(H[X]),{\rm e}}\leq\|T(a)\|_{\mathcal{B}(H[X])}\leq\|P\|_{\mathcal{B}(X)}\|a\|_{L^{\infty}}. (1.1)

Let CC denote the Banach space of all complex-valued continuous functions on 𝕋\mathbb{T} with the supremum norm and let

C+H:={fL:f=g+h,gC,hH}.C+H^{\infty}:=\{f\in L^{\infty}\ :\ f=g+h,\ g\in C,\ h\in H^{\infty}\}.

In 1967, Sarason observed that C+HC+H^{\infty} is a closed subalgebra of LL^{\infty} (see, e.g., [4, Ch. IX, Theorem 2.2] for the proof of this fact).

Let 1<p<1<p<\infty and aLa\in L^{\infty}. It follows from [3, Theorem 2.30] and (1.1) that

aLT(a)(Hp),eT(a)(Hp)P(Lp)aL.\|a\|_{L^{\infty}}\leq\|T(a)\|_{\mathcal{B}(H^{p}),\mathrm{e}}\leq\|T(a)\|_{\mathcal{B}(H^{p})}\leq\|P\|_{\mathcal{B}(L^{p})}\|a\|_{L^{\infty}}.

I. Gohberg and N. Krupnik [5, Theorem 6] proved that P(Lp),e1/sin(π/p)\|P\|_{\mathcal{B}(L^{p}),\mathrm{e}}\geq 1/\sin(\pi/p) and conjectured that P(Lp)=1/sin(π/p)\|P\|_{\mathcal{B}(L^{p})}=1/\sin(\pi/p). This conjecture was confirmed by B. Hollenbeck and I. Verbitsky in [6]. Thus

aLT(a)(Hp),e1/sin(π/p)aL,aL.\|a\|_{L^{\infty}}\leq\|T(a)\|_{\mathcal{B}(H^{p}),\mathrm{e}}\leq 1/\sin(\pi/p)\|a\|_{L^{\infty}},\quad a\in L^{\infty}. (1.2)

A. Böttcher, N. Krupnik, and B. Silbermann [2, Section 7.6] asked whether the essential norm of Toeplitz operators T(a)T(a) with aCa\in C acting on the Hardy spaces HpH^{p} is independent of p(1,)p\in(1,\infty). The second author answered this question in the negative [12]. More precisely, it was shown that

T(a)(Hp),e=aLfor allaCif and only ifp=2.\|T(a)\|_{\mathcal{B}(H^{p}),\mathrm{e}}=\|a\|_{L^{\infty}}\quad\mbox{for all}\quad a\in C\quad\mbox{if and only if}\quad p=2. (1.3)

Nevertheless, the following estimates for T(a)(Hp),e\|T(a)\|_{\mathcal{B}(H^{p}),\mathrm{e}} were obtained for 1<p<1<p<\infty and aC+Ha\in C+H^{\infty}:

aLT(a)(Hp),emin{2|12/p|,1/sin(π/p)}aL\|a\|_{L^{\infty}}\leq\|T(a)\|_{\mathcal{B}(H^{p}),\mathrm{e}}\leq\min\left\{2^{|1-2/p|},1/\sin(\pi/p)\right\}\|a\|_{L^{\infty}}

(see (1.2) and [12, Theorem 4.1]).

We will use the following notation:

𝐞m(z):=zm,z,m.\mathbf{e}_{m}(z):=z^{m},\quad z\in\mathbb{C},\quad m\in\mathbb{Z}.

The following result extends (1.3) to the class of rearrangement-invariant Banach function spaces.

Theorem 1.1.

Let XX be a rearrangement-invariant Banach function space such that the Riesz projection PP is bounded on XX. Then the following statements are equivalent:

  1. (a)

    the equality

    T(a)(H[X]),e=aL\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}}=\|a\|_{L^{\infty}} (1.4)

    holds for every Toeplitz operator T(a):H[X]H[X]T(a):H[X]\to H[X] with aLa\in L^{\infty};

  2. (b)

    equality (1.4) holds for every Toeplitz operator T(a):H[X]H[X]T(a):H[X]\to H[X] with aC+Ha\in C+H^{\infty};

  3. (c)

    T(𝐞1)(H[X]),e=1\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X]),\mathrm{e}}=1;

  4. (d)

    T(𝐞1)(H[X])=1\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X])}=1;

  5. (e)

    P(X)=1\|P\|_{\mathcal{B}(X)}=1;

  6. (f)

    X=L2X=L^{2} and there exists C(0,)C\in(0,\infty) such that

    gX=CgL2for allgX.\|g\|_{X}=C\|g\|_{L^{2}}\quad\mbox{for all}\quad g\in X.

The implication (f) \Longrightarrow (a) follows from inequalities (1.2), which become equalities for p=2p=2. The implications (a) \Longrightarrow (b) \Longrightarrow (c) are trivial. The equivalences (d) \Longleftrightarrow (e) \Longleftrightarrow (f) were proved in [10, Theorems 1.1–1.2] for arbitrary (not necessarily rearrangement-invariant) Banach function spaces XX. The equality

T(𝐞1)(H[X])=T(𝐞1)(H[X]),e\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X])}=\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X]),\mathrm{e}}

was proved in [8, Theorem 1.2] for rearrangement-invariant Banach function spaces XX, which gives the equivalence (c) \Longleftrightarrow (d) and completes the proof of Theorem 1.1.

Böttcher, Krupnik, and Silbermann [2, p. 472] provided an argument allowing to show directly that (c) \Longrightarrow (b) in the case of classical Hardy spaces HpH^{p}, 1<p<1<p<\infty. The aim of this paper is to show that their reasoning can be extended to the case of arbitrary Banach function spaces (not necessarily rearrangement-invariant) on which the Riesz projection PP is bounded. Our main result is the following.

Theorem 1.2 (Main result).

Let XX be a Banach function space on which the Riesz projection is bounded. Then the following statements are equivalent:

  1. (i)

    the equality

    T(𝐞1)(H[X]),e=1\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X]),\mathrm{e}}=1

    holds for the backward shift operator T(𝐞1):H[X]H[X]T(\mathbf{e}_{-1}):H[X]\to H[X];

  2. (ii)

    the equality

    T(a)(H[X]),e=aL\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}}=\|a\|_{L^{\infty}}

    holds for every Toeplitz operator T(a):H[X]H[X]T(a):H[X]\to H[X] with aC+Ha\in C+H^{\infty}.

The paper is organised as follows. In Section 2, we recall the definition of the class of Banach function spaces and of its distinguished subclasss of rearrangement-invariant Banach function spaces. In Section 3, we prove that the Toeplitz operators T(𝐞nh)T(\mathbf{e}_{-n}h) with n+:={0,1,2,}n\in\mathbb{Z}_{+}:=\{0,1,2,\dots\} and hHh\in H^{\infty} are bounded on H[X]H[X]. Further, we show that T(𝐞1)T(𝐞nh)=T(𝐞n1h)T(\mathbf{e}_{-1})T(\mathbf{e}_{-n}h)=T(\mathbf{e}_{-n-1}h) for n+n\in\mathbb{Z}_{+} and hHh\in H^{\infty} on the space H[X]H[X]. Although our main results have been obtained under the assumption that PP is bounded on XX, we do not make this assumption in Section 3 as we believe that this more general case is of an independent interest. By using the results of Section 3, we prove Theorem 1.2 in Section 4.

2. Preliminaries

2.1. Banach function spaces

Let \mathcal{M} be the set of all measurable complex-valued functions on 𝕋\mathbb{T} equipped with the normalized Lebesgue measure mm and let +\mathcal{M}^{+} be the subset of functions in \mathcal{M} whose values lie in [0,][0,\infty]. Following [1, Ch. 1, Definition 1.1], a mapping ρ:+[0,]\rho:\mathcal{M}^{+}\to[0,\infty] is called a Banach function norm if, for all functions f,g,fn+f,g,f_{n}\in\mathcal{M}^{+} with nn\in\mathbb{N}, and for all constants a0a\geq 0, the following properties hold:

(A1)\displaystyle{\rm(A1)} ρ(f)=0f=0a.e.,ρ(af)=aρ(f),ρ(f+g)ρ(f)+ρ(g),\displaystyle\rho(f)=0\Leftrightarrow f=0\ \mbox{a.e.},\ \rho(af)=a\rho(f),\ \rho(f+g)\leq\rho(f)+\rho(g),
(A2)\displaystyle{\rm(A2)} 0gfa.e.ρ(g)ρ(f)(the lattice property),\displaystyle 0\leq g\leq f\ \mbox{a.e.}\ \Rightarrow\ \rho(g)\leq\rho(f)\quad\mbox{(the lattice property)},
(A3)\displaystyle{\rm(A3)} 0fnfa.e.ρ(fn)ρ(f)(the Fatou property),\displaystyle 0\leq f_{n}\uparrow f\ \mbox{a.e.}\ \Rightarrow\ \rho(f_{n})\uparrow\rho(f)\quad\mbox{(the Fatou property)},
(A4)\displaystyle{\rm(A4)} ρ(1)<,\displaystyle\rho(1)<\infty,
(A5)\displaystyle{\rm(A5)} 𝕋f(t)𝑑m(t)Cρ(f)\displaystyle\int_{\mathbb{T}}f(t)\,dm(t)\leq C\rho(f)

with a constant C(0,)C\in(0,\infty) that may depend on ρ\rho, but is independent of ff. When functions differing only on a set of measure zero are identified, the set XX of all functions ff\in\mathcal{M} for which ρ(|f|)<\rho(|f|)<\infty is called a Banach function space. For each fXf\in X, the norm of ff is defined by fX:=ρ(|f|)\|f\|_{X}:=\rho(|f|). The set XX equipped with the natural linear space operations and this norm becomes a Banach space (see [1, Ch. 1, Theorems 1.4 and 1.6]).

2.2. Rearrangement-invariant Banach function spaces

Let 0\mathcal{M}_{0} (resp. 0+\mathcal{M}_{0}^{+}) denote the set of all a.e. finite functions in \mathcal{M} (resp. in +\mathcal{M}^{+}). Following [1, Chap. 2, Definitions 1.1 and 1.2], the distribution function mfm_{f} of a function f0f\in\mathcal{M}_{0} is given by

mf(λ):=m{t𝕋:|f(t)|>λ},λ0.m_{f}(\lambda):=m\left\{t\in\mathbb{T}\ :\ |f(t)|>\lambda\right\},\quad\lambda\geq 0.

Two functions f,g0f,g\in\mathcal{M}_{0} are said to be equimeasurable if mf(λ)=mg(λ)m_{f}(\lambda)=m_{g}(\lambda) for all λ0\lambda\geq 0. A Banach function norm ρ:[0,]\rho:\mathcal{M}\to[0,\infty] is said to be rearrangement-invariant if ρ(f)=ρ(g)\rho(f)=\rho(g) for every pair of equimeasurable functions f,g0+f,g\in\mathcal{M}_{0}^{+}. In that case, the Banach function space XX generated by ρ\rho is said to be a rearrangement-invariant Banach function space (see [1, Ch. 2, Definition 4.1]).

3. Auxiliary results

3.1. Operator PnP_{n}

For nn\in\mathbb{N} and fL1f\in L^{1}, put

Pnf:=k=0n1f^(k)𝐞kH1.P_{n}f:=\sum_{k=0}^{n-1}\widehat{f}(k)\mathbf{e}_{k}\in H^{1}.
Lemma 3.1.

For every nn\in\mathbb{N}, the operator Pn:L1HP_{n}:L^{1}\to H^{\infty} is bounded and

Pn(L1,H)n.\|P_{n}\|_{\mathcal{B}(L^{1},H^{\infty})}\leq n.
Proof.

For every fL1f\in L^{1}, on has

PnfL\displaystyle\|P_{n}f\|_{L^{\infty}} =k=0n1f^(k)𝐞kLk=0n1|f^(k)|𝐞kL\displaystyle=\left\|\sum_{k=0}^{n-1}\widehat{f}(k)\mathbf{e}_{k}\right\|_{L^{\infty}}\leq\sum_{k=0}^{n-1}\left|\widehat{f}(k)\right|\|\mathbf{e}_{k}\|_{L^{\infty}}
=k=0n1|f^(k)|k=0n1fL1=nfL1.\displaystyle=\sum_{k=0}^{n-1}\left|\widehat{f}(k)\right|\leq\sum_{k=0}^{n-1}\|f\|_{L^{1}}=n\|f\|_{L^{1}}.

So, Pn(L1,H)P_{n}\in\mathcal{B}(L^{1},H^{\infty}) and Pn(L1,H)n\|P_{n}\|_{\mathcal{B}(L^{1},H^{\infty})}\leq n. ∎

Corollary 3.2.

Let XX be a Banach function space. For every nn\in\mathbb{N}, the operator Pn:XH[X]P_{n}:X\to H[X] is bounded.

Proof.

Axioms (A4) and (A5) imply the existence of a constant C>0C>0 such that

Pn(X,H[X])CPn(L1,H)Cn,\|P_{n}\|_{\mathcal{B}(X,H[X])}\leq C\|P_{n}\|_{\mathcal{B}(L^{1},H^{\infty})}\leq Cn,

which completes the proof. ∎

3.2. Boundendess of a special Toeplitz operator

We will need the following auxiliary lemma

Lemma 3.3 ([7, Lemma 3.1]).

Let fL1f\in L^{1}. Suppose there exists gH1g\in H^{1} such that f^(n)=g^(n)\widehat{f}(n)=\widehat{g}(n) for all n0n\geq 0. Then Pf=gPf=g.

As a consequence of the results of the previous subsection and Lemma 3.3, we will show that special Toeplitz operators with symbols of the form 𝐞nh\mathbf{e}_{-n}h, where nn\in\mathbb{N} and hHh\in H^{\infty}, are bounded on abstract Hardy spaces H[X]H[X] built upon Banach function spaces XX even without the assumption that the Riesz projection PP is bounded on XX.

Lemma 3.4.

Let XX be a Banach function space. If n+n\in\mathbb{Z}_{+} and hHh\in H^{\infty}, then the Toeplitz operator T(𝐞nh):H[X]H[X]T(\mathbf{e}_{-n}h):H[X]\to H[X] is bounded.

Proof.

Let fH[X]H1f\in H[X]\subset H^{1}. Since hHh\in H^{\infty}, it follows from [11, Section 3.3.1, properties (a), (g)] that hfH1hf\in H^{1}. In the case n=0n=0, Lemma 3.3 implies that

T(h)f=P(hf)=hfT(h)f=P(hf)=hf (3.1)

and

T(h)fH[X]=hfH[X]hLfH[X],\|T(h)f\|_{H[X]}=\|hf\|_{H[X]}\leq\|h\|_{L^{\infty}}\|f\|_{H[X]},

whence

T(h)(H[X])hL.\|T(h)\|_{\mathcal{B}(H[X])}\leq\|h\|_{L^{\infty}}. (3.2)

If nn\in\mathbb{N}, then

T(𝐞nh)f=P(𝐞nhf)=P(𝐞nPn(hf))+P(𝐞n(IPn)(hf)).\displaystyle T(\mathbf{e}_{-n}h)f=P(\mathbf{e}_{-n}hf)=P(\mathbf{e}_{-n}P_{n}(hf))+P(\mathbf{e}_{-n}(I-P_{n})(hf)).

It follows from the definition of PnP_{n} that

(𝐞nPn(hf))^(m)=0,m+;\displaystyle(\mathbf{e}_{-n}P_{n}(hf))\widehat{\hskip 5.69054pt}(m)=0,\quad m\in\mathbb{Z}_{+}; (3.3)
(𝐞n(IPn)(hf))^(m)=0,m+.\displaystyle(\mathbf{e}_{-n}(I-P_{n})(hf))\widehat{\hskip 5.69054pt}(m)=0,\quad m\in\mathbb{Z}\setminus\mathbb{Z}_{+}.

Hence

P(𝐞nPn(hf))=0,P(𝐞n(IPn)(hf))=𝐞n(IPn)(hf)P(\mathbf{e}_{-n}P_{n}(hf))=0,\quad P(\mathbf{e}_{-n}(I-P_{n})(hf))=\mathbf{e}_{-n}(I-P_{n})(hf)

(see Lemma 3.3). So,

T(𝐞nh)f=P(𝐞nhf)=𝐞n(IPn)(hf).T(\mathbf{e}_{-n}h)f=P(\mathbf{e}_{-n}hf)=\mathbf{e}_{-n}(I-P_{n})(hf). (3.4)

Hence, taking into account Corollary 3.2, we obtain

T(𝐞nh)fH[X]\displaystyle\|T(\mathbf{e}_{-n}h)f\|_{H[X]} =T(𝐞nh)fX=𝐞n(IPn)(hf)X\displaystyle=\|T(\mathbf{e}_{-n}h)f\|_{X}=\|\mathbf{e}_{-n}(I-P_{n})(hf)\|_{X}
=(IPn)(hf)X(1+Pn(X,H[X]))hfX\displaystyle=\|(I-P_{n})(hf)\|_{X}\leq\left(1+\|P_{n}\|_{\mathcal{B}(X,H[X])}\right)\|hf\|_{X}
=(1+Pn(X,H[X]))hLfH[X].\displaystyle=\left(1+\|P_{n}\|_{\mathcal{B}(X,H[X])}\right)\|h\|_{L^{\infty}}\|f\|_{H[X]}.

So, T(𝐞nh)(H[X])T(\mathbf{e}_{-n}h)\in\mathcal{B}(H[X]) and

T(𝐞nh)(H[X])(1+Pn(X,H[X]))hL,\|T(\mathbf{e}_{-n}h)\|_{\mathcal{B}(H[X])}\leq\left(1+\|P_{n}\|_{\mathcal{B}(X,H[X])}\right)\|h\|_{L^{\infty}},

which completes the proof. ∎

The above lemma can be complemented by the following (cf. [3, Proposition 2.14]).

Lemma 3.5.

Let XX be a Banach function space. If n+n\in\mathbb{Z}_{+} and hHh\in H^{\infty}, then

T(𝐞1)T(𝐞nh)=T(𝐞n1h)T(\mathbf{e}_{-1})T(\mathbf{e}_{-n}h)=T(\mathbf{e}_{-n-1}h)

on the space H[X]H[X].

Proof.

It follows from Lemma 3.4 that the Toeplitz operators T(𝐞1)T(\mathbf{e}_{-1}), T(𝐞nh)T(\mathbf{e}_{-n}h) and T(𝐞n1h)T(\mathbf{e}_{-n-1}h) are bounded on the space H[X]H[X]. Let fH[X]f\in H[X]. If n=0n=0, then it follows from (3.1) that

T(𝐞1)T(h)f=T(𝐞1)(hf)=P(𝐞1hf)=T(𝐞1h)f.T(\mathbf{e}_{-1})T(h)f=T(\mathbf{e}_{-1})(hf)=P(\mathbf{e}_{-1}hf)=T(\mathbf{e}_{-1}h)f.

If nn\in\mathbb{N}, then (3.4) and (3.3) imply that

T(𝐞1)T(𝐞nh)f=\displaystyle T(\mathbf{e}_{-1})T(\mathbf{e}_{-n}h)f= 𝐞1(IP1)(T(𝐞nh)f)\displaystyle\mathbf{e}_{-1}(I-P_{1})(T(\mathbf{e}_{-n}h)f)
=\displaystyle= 𝐞1[T(𝐞nh)f(T(𝐞nh)f)^(0)]\displaystyle\mathbf{e}_{-1}\left[T(\mathbf{e}_{-n}h)f-(T(\mathbf{e}_{-n}h)f)\widehat{\hskip 5.69054pt}(0)\right]
=\displaystyle= 𝐞1[𝐞n(IPn)(hf)(𝐞n(IPn)(hf))^(0)]\displaystyle\mathbf{e}_{-1}\left[\mathbf{e}_{-n}(I-P_{n})(hf)-(\mathbf{e}_{-n}(I-P_{n})(hf))\widehat{\hskip 5.69054pt}(0)\right]
=\displaystyle= 𝐞n1(IPn)(hf)𝐞1(𝐞nhf)^(0)\displaystyle\mathbf{e}_{-n-1}(I-P_{n})(hf)-\mathbf{e}_{-1}(\mathbf{e}_{-n}hf)\widehat{\hskip 5.69054pt}(0)
=\displaystyle= 𝐞n1(IPn+1)(hf)\displaystyle\mathbf{e}_{-n-1}(I-P_{n+1})(hf)
+𝐞n1(Pn+1Pn)(hf)𝐞1(𝐞nhf)^(0)\displaystyle+\mathbf{e}_{-n-1}(P_{n+1}-P_{n})(hf)-\mathbf{e}_{-1}(\mathbf{e}_{-n}hf)\widehat{\hskip 5.69054pt}(0)
=\displaystyle= T(𝐞n1h)f+𝐞n1hf^(n)𝐞n𝐞1hf^(n)\displaystyle T(\mathbf{e}_{-n-1}h)f+\mathbf{e}_{-n-1}\widehat{hf}(n)\mathbf{e}_{n}-\mathbf{e}_{-1}\widehat{hf}(n)
=\displaystyle= T(𝐞n1h)f,\displaystyle T(\mathbf{e}_{-n-1}h)f,

which completes the proof. ∎

4. Proof of the main result

4.1. Extending an observation by Böttcher, Krupnik, and Silbermann

We start with the following auxiliary result, containing the essence of the argument in [2, p. 472], in which we do not assume the boundedness of the Riesz projection on a Banach function space XX.

Lemma 4.1.

Let XX be a Banach function space. If T(𝐞1)(H[X]),e=1\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X]),\mathrm{e}}=1, then T(a)(H[X]),eaL\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}}\leq\|a\|_{L^{\infty}} for every Toeplitz operator T(a):H[X]H[X]T(a):H[X]\to H[X] with a{𝐞nh:n,hH}a\in\{\mathbf{e}_{-n}h:n\in\mathbb{N},h\in H^{\infty}\}.

Proof.

Let a=𝐞nha=\mathbf{e}_{-n}h for some nn\in\mathbb{N} and hHh\in H^{\infty}. By Lemma 3.5 and (3.2), we have

T(a)(H[X]),e\displaystyle\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}} =T(𝐞nh)(H[X]),e=(T(𝐞1))nT(h)(H[X]),e\displaystyle=\|T(\mathbf{e}_{-n}h)\|_{\mathcal{B}(H[X]),\mathrm{e}}=\|(T(\mathbf{e}_{-1}))^{n}T(h)\|_{\mathcal{B}(H[X]),\mathrm{e}}
T(𝐞1)(H[X]),enT(h)(H[X]),e=T(h)(H[X]),e\displaystyle\leq\|T(\mathbf{e}_{-1})\|_{\mathcal{B}(H[X]),\mathrm{e}}^{n}\|T(h)\|_{\mathcal{B}(H[X]),\mathrm{e}}=\|T(h)\|_{\mathcal{B}(H[X]),\mathrm{e}}
T(h)(H[X])hL=aL,\displaystyle\leq\|T(h)\|_{\mathcal{B}(H[X])}\leq\|h\|_{L^{\infty}}=\|a\|_{L^{\infty}},

which completes the proof. ∎

4.2. Proof of Theorem 1.2

It is clear that (ii) implies (i). Suppose (i) holds and aC+Ha\in C+H^{\infty}. Since the set G:={𝐞nh:n,hH}G:=\{\mathbf{e}_{-n}h:n\in\mathbb{N},h\in H^{\infty}\} is dense in C+HC+H^{\infty} (see, e.g., [4, Ch. IX, Theorem 2.2]), there is a sequence {am}\{a_{m}\} of elements of GG such that aamL0\|a-a_{m}\|_{L^{\infty}}\to 0 as mm\to\infty. By Lemma 4.1, T(am)(H[X]),eamL\|T(a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}}\leq\|a_{m}\|_{L^{\infty}} for all mm\in\mathbb{N}. On the other hand, it follows from [9, Theorem 5.2] that amLT(am)(H[X]),e\|a_{m}\|_{L^{\infty}}\leq\|T(a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}} for all mm\in\mathbb{N}. Thus T(am)(H[X]),e=amL\|T(a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}}=\|a_{m}\|_{L^{\infty}} for all mm\in\mathbb{N}. Since

|T(a)(H[X]),eT(am)(H[X]),e|T(aam)(H[X]),eP(X)aamL\left|\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}}-\|T(a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}}\right|\leq\|T(a-a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}}\leq\|P\|_{\mathcal{B}(X)}\|a-a_{m}\|_{L^{\infty}}

and |aLamL|aamL\left|\,\|a\|_{L^{\infty}}-\|a_{m}\|_{L^{\infty}}\right|\leq\|a-a_{m}\|_{L^{\infty}} for all mm\in\mathbb{N} and aamL0\|a-a_{m}\|_{L^{\infty}}\to 0 as mm\to\infty, we get

T(a)(H[X]),e=limnT(am)(H[X]),e=limmamL=aL,\|T(a)\|_{\mathcal{B}(H[X]),\mathrm{e}}=\lim_{n\to\infty}\|T(a_{m})\|_{\mathcal{B}(H[X]),\mathrm{e}}=\lim_{m\to\infty}\|a_{m}\|_{L^{\infty}}=\|a\|_{L^{\infty}},

which completes the proof of (ii). ∎

Acknowledgements

This work is funded by national funds through the FCT - Fundação para a Ciência e a Tecnologia, I.P., under the scope of the projects UIDB/00297/2020 (https://doi.org/10.54499/UIDB/00297/2020) and UIDP/ 00297/2020 (https://doi.org/10.54499/UIDP/00297/2020) (Center for Mathematics and Applications).

References

  • [1] C. Bennett and R. Sharpley. Interpolation of operators, volume 129 of Pure and Applied Mathematics. Academic Press, Inc., Boston, MA, 1988.
  • [2] A. Böttcher, N. Krupnik, and B. Silbermann. A general look at local principles with special emphasis on the norm computation aspect. Integral Equations Operator Theory, 11(4):455–479, 1988.
  • [3] A. Böttcher and B. Silbermann. Analysis of Toeplitz operators. Springer Monographs in Mathematics. Springer-Verlag, Berlin, second edition, 2006.
  • [4] J. B. Garnett. Bounded analytic functions, volume 236 of Graduate Texts in Mathematics. Springer, New York, first edition, 2007.
  • [5] I. C. Gohberg and N. J. Krupnik. The spectrum of singular integral operators in LpL_{p} spaces. Studia Math., 31:347–362, 1968.
  • [6] B. Hollenbeck and I. E. Verbitsky. Best constants for the Riesz projection. J. Funct. Anal., 175(2):370–392, 2000.
  • [7] A. Karlovich. Noncompactness of Toeplitz operators between abstract Hardy spaces. Adv. Oper. Theory, 6(2):Paper No. 29, 10, 2021.
  • [8] O. Karlovych and E. Shargorodsky. On the essential norms of singular integral operators with constant coefficients and of the backward shift. Proc. Amer. Math. Soc. Ser. B, 9:60–70, 2022.
  • [9] O. Karlovych and E. Shargorodsky. The Coburn lemma and the Hartman-Wintner-Simonenko theorem for Toeplitz operators on abstract Hardy spaces. Integral Equations Operator Theory, 95(1):Paper No. 6, 17, 2023.
  • [10] O. Karlovych and E. Shargorodsky. When are the norms of the Riesz projection and the backward shift operator equal to one? J. Funct. Anal., 285(12):Paper No. 110158, 29, 2023.
  • [11] N. Nikolski. Hardy spaces, volume 179 of Cambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge, french edition, 2019.
  • [12] E. Shargorodsky. On the essential norms of Toeplitz operators with continuous symbols. J. Funct. Anal., 280(2):108835, 11, 2021.