Abstract
Let E be a closed subset of the unit circle \(\mathbb {T}\), and let \(\alpha \in (0,1)\). Nikolski’s result states that if the Hausdorff dimension of E is strictly greater than \(\alpha \), then for any operator T on a separable Hilbert space such that the point spectrum \(\sigma _p(T)\) of T contains E, the series \(\sum _{n}n^{\alpha -1}\Vert T^n\Vert ^{-2}\) converges. A partial converse of this result has been obtained by El-Fallah and Ransford. Namely they constructed, for any \(\alpha \) strictly greater than the upper box dimension of E, an operator T on a separable Hilbert space such that \(\sigma _p(T)\) contains E and \( \frac{1}{n} \sum _{k=0}^{n-1}\left\| T^k\right\| ^2\lesssim n^{\alpha }\). In this paper, we improve on this latter result for regular sets. Indeed, for any Ahlfors–David regular set E and for any \(\alpha \) strictly greater than the Hausdorff dimension of E there exists an operator T on a separable Hilbert space such that \(\sigma _p(T)\) contains E and \(\Vert T^n\Vert ^2\asymp n^{\alpha }\).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let T be a bounded linear operator on a complex Banach space X. The unit circle is denoted by \(\mathbb {T}\), while \(\sigma _p(T):=\{\lambda \in \mathbb {C}:\,\, \ker (T-\lambda I)\not =\emptyset \}\) represents the point spectrum of T. Jamison [10] showed that if X is separable and T is power-bounded, then \(\sigma _p(T)\cap \mathbb {T}\) is at most countable. Later on, several authors (see, e.g., [1, 2, 4, 15, 16] and the references therein) have shown interest in the study of the relationship between the size of the set \(\sigma _p(T)\cap \mathbb {T}\) and the growth of \(\Vert T^n\Vert \) as \(n\rightarrow \infty \).
In the case of separable Hilbert spaces, Nikolski [13] proved that if \(\sigma _p(T)\cap \mathbb {T}\) has a positive \(\gamma \)-capacity, where \(\gamma : \mathbb {T}\rightarrow (0,\infty )\) is integrable with positive Fourier coefficients, then there exists \(N\in \mathbb {N}\) such that \(\sum _n\widehat{\gamma }(n+N)\Vert T^n\Vert ^{-2}\) converges (see [11, Chapter 3, p. 31] for the definition of \(\gamma \)-capacity). As a by-product of this, if \(\sigma _p(T)\cap \mathbb {T}\) contains a subset E of \(\mathbb {T}\) with \(\dim _H(E)>\alpha >0\), where \(\dim _H(E)\) is the Hausdorff dimension of E, then the series \(\sum _nn^{\alpha -1}\Vert T^n\Vert ^{-2}\) converges. El-Fallah-Ransford [8] proved that, as a partial converse of the preceding result, if \(\alpha \) is strictly greater than the upper box dimension of E, then there exists an operator T on a separable Hilbert space such that \(\sigma _p(T)\cap \mathbb {T}\) contains E and the series \(\sum _nn^{\alpha -1}\Vert T^n\Vert ^{-2}\) diverges. (See [9, p. 41] for more details about the box dimension). Precisely, they constructed an operator T such that \(\sigma _p(T)\) contains E and
where \(\omega : \mathbb {Z} \rightarrow (0, \infty )\) is a regular weight function satisfying \(\sum _n \frac{1}{\omega (n)^2}<\infty \), \(E_{1 / n}:=\{\zeta \in \mathbb {T}: \,\, {{\,\textrm{dist}\,}}(E,\zeta )<\frac{1}{n}\}\) with \({{\,\textrm{dist}\,}}(.,.)\) being the arc-length distance, and \(\left| E_{1 / n}\right| \) is its Lebesgue measure. In particular, one can obtain
The main result of this paper is the following theorem.
Theorem 1.1
Assume that \(E\subset \mathbb {T}\) is a closed Ahlfors–David regular set. If \(\alpha >\dim _H(E)\), then there exists an operator T on a separable Hilbert space such that \(\sigma _p(T)\cap \mathbb {T}\) contains E and
The definitions of Ahlfors–David regular sets and Hausdorff dimension are recalled in Sect. 4. Theorem 1.1 is a corollary from the following more general result.
Theorem 1.2
Let E be closed subset of \(\mathbb {T}\), if there exist an increasing function \(\Lambda : [0,1]\rightarrow [0,+\infty )\) such that \(\frac{\Lambda (t)}{t^c}\) is decreasing for some \(c>0\), and a positive finite Borel measure \(\mu \) on \(\mathbb {T}\) satisfying
where \(\mu (\overline{\zeta },t)=\mu (\overline{\zeta } e^{-it},\overline{\zeta } e^{it}),\) then there exists an operator T on a separable Hilbert space such that \(E \subset \sigma _p(T)\) and
In Sect. 2, we study the point spectrum of adjoint of the shift operator acting on some weighted Dirichlet spaces. In Sect. 3, we determine the growth of power of the adjoint of this shift operator. In Sect. 4, we give the proofs of the main theorems.
2 Point Spectrum
Let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be a positive function, let \(\mathbb {D}\) be the unit disc, and let \(\mu \) be a positive finite Borel measure on the unit circle \(\mathbb {T}\). The weighted Dirichlet integral of \(f\in \textrm{Hol}(\mathbb {D})\) associated with \(\Lambda \) and \(\mu \) is defined as follows:
where dA denotes the normalized area measure on \(\mathbb {D}\), and \(P_\mu \) is the Poisson integral of \(\mu \) on \(\mathbb {T}\) given by
The associated weighted Dirichlet spaces \({\mathcal {D}}_{\Lambda ,\mu }\) consist of all analytic functions on \(\mathbb {D}\) with finite weighted Dirichlet integral, i.e.,
We associate to \({\mathcal {D}}_{\Lambda ,\mu }\) the following inner product
where
\({\mathcal {D}}_{\Lambda ,\mu }\) is a reproducing kernel Hilbert space, and denote K (or \(K_{\Lambda ,\mu }\) if necessary) its reproducing kernel. The standard weighted Dirichlet spaces on \(\mathbb {D}\), denoted \({\mathcal {D}}_\alpha \), correspond to \(\Lambda (t)=t^{\alpha }\) and \(\mu =m\) the normalized arc measure on \(\mathbb {T}\). If \(\Lambda =1\), then \({\mathcal {D}}_{\Lambda ,\mu }\) is the harmonically weighted Dirichlet spaces. (See, e.g., [6, 7]). Note that, in general, \({\mathcal {D}}_{\Lambda ,\mu }\) is not contained in the Hardy space \(\mathrm {H^2 }\). In the following proposition, using a reasoning similar to that in [3], we establish the density of polynomials in \({\mathcal {D}}_{\Lambda ,\mu },\) for some regular weights \(\Lambda \).
Proposition 2.1
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t)}{t^c}\) is decreasing for some \(c>0\). Then the polynomials are dense in \({\mathcal {D}}_{\Lambda ,\mu }.\)
Proof
The proof uses the fact that the dilations \(f_r(z):=f(r z)\) for \(r\in [0,1)\) tend to f in the norm, i.e.,
To this end, it is sufficient to show that
Since \(\Lambda (2t)\asymp \Lambda (t)\), we have
where \(\rho (w)=(1-|w|)/2.\) Hence,
Denote
Using the following inequality (cf. [14, Lemma 2.5]):
we obtain
\(\square \)
Let \(f\in \textrm{Hol}(\mathbb {D})\), and let \(\zeta \in \mathbb {T}\), we write \(f^*(\zeta ):=\lim _{r\rightarrow 1}f(r\zeta )\) its radial limit (if it exists) at \(\zeta \). Note that, under conditions of Proposition 2.1, if \(\sup _{0\le r<1}K(r\zeta ,r\zeta )<\infty \), then \(f^*(\zeta )\) exists for every \(f\in {\mathcal {D}}_{\Lambda ,\mu }\).
We denote \(S_{\Lambda ,\mu }\) (or simply S) the shift operator acting on \({\mathcal {D}}_{\Lambda ,\mu }\) defined as follows:
and \(S^*\) is its adjoint operator.
Theorem 2.1
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t)}{t^c}\) is decreasing for some \(c>0\), then
To prove Theorem 2.1, we need the following lemma.
Lemma 2.1
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t)}{t^c}\) is decreasing for some \(c>0\), then
The proof of Lemma 2.1 is inspired from [5]. For the sake of completeness, we include the proof here.
Proof
We have
Let \(f\in {\mathcal {D}}_{\Lambda ,\mu }\), and let \(z=r\in [\frac{1}{2},1)\). Consider
We have
Using the following change of variables:
we obtain
Now, for \(u+iv \in \Delta _r\), we have \( \begin{aligned} |u+i v|&\le u+| v | \le u+\frac{1-u}{2} =\frac{1+u}{2}. \end{aligned} \) Then \(1-|u+i v| \geqslant 1-\frac{1+u}{2}=\frac{1-u}{2}.\) Hence, , and Therefore,
Moreover, the disc \(D\left( r, \frac{1-r}{4}\right) \) is included in \(\left\{ z=x+i y,|x-r| \lesssim \frac{1-r}{4},|y| \lesssim \right. \left. \frac{1-x}{2}\right\} .\) Thus
Therefore,
and we get
\(\square \)
Proof of Theorem 2.1
We have for any \(\zeta \in \mathbb {T}\), and \(t\in (0,1)\). Then
Let \(\zeta \in \mathbb {T}\) such that \(\int _0^1\frac{dx}{\Lambda (x)\mu (\overline{\zeta },x)}<\infty \). Combining inequality (2) with Lemma 2.1, we obtain \(\sup _{0\le r<1} K(r\overline{\zeta },r\overline{\zeta })<\infty \). Consider now \(L_{\overline{\zeta }}: {\mathcal {D}}_{\Lambda ,\mu }\rightarrow \mathbb {C}, f\mapsto f^*(\overline{\zeta })\). Since \(L_{\overline{\zeta }}\) is continuous, it follows from Riesz representation theorem that there exists \(k_{\overline{\zeta }}\in {\mathcal {D}}_{\Lambda ,\mu }\) such that \(f^*(\overline{\zeta })=\left<f,k_{\overline{\zeta }}\right>_{\Lambda ,\mu }.\) Hence, \(S^*k_{\overline{\zeta }}=\zeta k_{\overline{\zeta }}\). Indeed, we have
\(\square \)
3 Growth of Power of Shift Operator
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let S be the shift operator acting on the Dirichlet space \({\mathcal {D}}_{\Lambda ,\mu }\) associated with \(\mu \) and a positive function \(\Lambda \).
Theorem 3.1
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t) }{t^c}\) is decreasing for some \(c>0\), and
We have
To prove Theorem 3.1, we require the following lemmas.
Lemma 3.1
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t) }{t^c}\) is decreasing for some \(c>0\). We have
Proof
Let \(n\ge 1\). We have
Now,
where \(\textbf{B}\) is Beta function. Using equality (4), we obtain
\(\square \)
In the case of \(\mu =\delta _\zeta \), the Dirac measure at \(\zeta \in \mathbb {T}\), the local weighted Dirichlet integral of \(f\in \textrm{Hol}(\mathbb {D})\) is given by
Suppose that \(\Lambda \equiv 1\). Let \(f\in \textrm{Hol}(\mathbb {D})\), say \(f(z)=\sum _{n\in \mathbb {N}}a_nz^n\), we have
see [7, Theorem 7.2.6].
For the rest of the paper, we suppose that \(\int _0\frac{dt}{\Lambda (t)}<\infty .\) Therefore, according to Lemma 2.1, the reproducing kernel of \({\mathcal {D}}_{\Lambda ,\zeta }\) satisfies \(\sup _{0\le r<1}K_{\Lambda ,\zeta }(r\zeta ,r\zeta )<\infty \). Then the following space
is closed in \({\mathcal {D}}_{\Lambda ,\zeta }.\) In order to extend formula (5) to \({\mathcal {D}}_{\Lambda ,\zeta },\) we endow the space \({\mathcal {D}}^0(\Lambda ,\zeta )\) with the following norm:
Additionally, we consider the weighted Bergman space \(\mathcal {A}^2_{(1-|z|^2)\Lambda (1-|z|^2)}\) equipped with the following norm:
Lemma 3.2
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t) }{t^c}\) is decreasing for some \(c>0\), and
For \(f\in {\mathcal {D}}_{\Lambda ,\zeta }\) write \(f(z)=\sum _{n\in \mathbb {N}}a_nz^n\). We have
Proof
Without loss of generality, we may assume that \(\zeta =1\). We make the following identification
The operator \(\textbf{T}\) is a surjective isometry. Indeed, let \(f\in {\mathcal {D}}^0(\Lambda ,1),\) and denote \(||f||_0:=||f||_{ 0, \Lambda , 1}\), we have
then \(\textbf{T}\) is an isometry. To prove that \(\textbf{T}\) is surjective, we consider
where \(\mathbb {C}[z]\) is the set of polynomials. Let \(p\in \mathbb {C}[z]\), we have
and
Since \((1-|z|^2)\Lambda (1-|z|^2)\) is a radial weight, we get that polynomials are dense in \(\mathcal {A}^2_{(1-|z|^2)\Lambda (1-|z|^2)}.\) It follows from equality (6) that \(\textbf{V}\) extends to an isometry \(\mathbf {\tilde{V}}\) on \(\mathcal {A}^2_{(1-|z|^2)\Lambda (1-|z|^2)}\). Using equality (6), we obtain
Thus, \(\textbf{T}\) is surjective. Moreover, since \((1-|z|^2)\Lambda (1-|z|^2)\) is a radial weight, we get that \(\left( e_n(z):=\frac{z^n}{||z^n||}\right) \) is an orthonormal basis of \(\mathcal {A}^2_{(1-|z|^2)\Lambda (1-|z|^2)}\). Hence, the following sequence
is an orthonormal basis of \({\mathcal {D}}^0(\Lambda ,1).\) Therefore, there exists a sequence \((c_n)_{n\ge 1}\) of complex numbers such that
Thus,
According to identification (6), we obtain
Then
\(\square \)
Lemma 3.3
Let \(\mu \) be a positive finite Borel measure on \(\mathbb {T}\), and let \(\Lambda : [0,1]\rightarrow [0,+\infty )\) be an increasing function such that \(\frac{\Lambda (t) }{t^c}\) is decreasing for some \(c>0\), and
For \(f\in {\mathcal {D}}_{\Lambda ,\zeta }\) write \(f(z)=\sum _{n\in \mathbb {N}}a_nz^n\). Then
Proof
We have
By Theorem 3.2, we obtain
Since \(||z^{n-1}||\le ||z^{n-1-p}||\), we get
Let \(A_n=\sum _{k=0}^nk a_k\), we have
and
Then
It follows from the inequality \(|a_0|\le ||f||_{\Lambda ,\delta _{\zeta }}\) that
In addition,
Moreover,
Then
\(\square \)
We are now in a position to prove Theorem 3.1.
Proof of Theorem 3.1
Let \(p\in \mathbb {N}\). By definition, \(\Vert S^{*p}\Vert =\sup _{\Vert f\Vert _{\Lambda ,\mu }=1}\Vert S^{*p}f\Vert _{\Lambda ,\mu }\). Then we get \(\Vert S^{*p}\Vert ^2\ge \Vert z^p\Vert ^2_{\Lambda ,\mu }\asymp p\Lambda \left( \frac{1}{p}\right) \). Now, let \(n\in \mathbb {N}\). Similar to the proof of Lemma 3.1, we have
Then
Furthermore,
Combining inequalities (6), (7), and Lemma 3.3, we obtain
4 Proofs of the Main Theorems
We recall some definitions which will be used in what follows. Let \(E\subset \mathbb {T}\), let \(\delta >0\), and let \(d\in [0,\infty )\). Consider
where the infimum is taken over all countable intervals covering E. The Hausdorff outer measure of dimension d is given by \(H^d(E)=\lim _{\delta \rightarrow 0}H^d_{\delta }(E)\), and the Hausdorff dimension of E is defined by
A closed subset E of \(\mathbb {T}\) is an Ahlfors-David regular set if there exists a measure \(\mu \) supported on E such that
for all \(\zeta \in {{\,\textrm{supp}\,}}\mu \) and \(t\in [0,1]\). In this case, we have \(\dim _H(E)=d\) (see [12] for more details).
Proof Theorem 1.2
Assume that there exist \(\mu \) and \(\Lambda \) satisfying the conditions of Theorem 1.2. Consider the operator \(T=S^*\) on \({\mathcal {D}}_{\Lambda ,\mu }\). We have \(\sigma _p(T)\supset E\). Indeed, for \(\zeta \in E\), we have \( \int _0\frac{dt}{\Lambda (t)\mu (\overline{\zeta },t)}<\infty ,\) it follows from Theorem 2.1 that \(\zeta \in \sigma _p(T).\) The first assertion is proved. Since \(\mu (\overline{\zeta },t)\Lambda (t)\lesssim \Lambda (t)\), condition (3) holds and the last assertion comes from Theorem 3.1.
We are now ready to prove Theorem 1.1.
Proof of Theorem 1.1
Since E is an Ahlfors–David set, \(E^*:=\{\zeta \in \mathbb {T}\,: \overline{\zeta }\in E\}\) is too. Thus, there exists a measure \(\mu \) supported on \(E^*\) such that \(\mu (\zeta ,t)\asymp t^d\) for any \(\zeta \in E^*\), and \(d=\dim _H(E)\). Let \(\alpha >d\) and consider the function \(\Lambda (t)=t^{1-\alpha }\), \(t\in (0,1)\). We have
According to Theorem 1.2, we deduce the result.
Data availability
This paper has no associated data.
References
Badea, C., Grivaux, S.: Size of the peripheral point spectrum under power or resolvent growth conditions. J. Funct. Anal. 246(2), 302–329 (2007)
Badea, C., Grivaux, S.: Unimodular eigenvalues, uniformly distributed sequences and linear dynamics. Adv. Math. 211(2), 766–793 (2007)
Bourass, M., Marrhich, I.: Littlewood–Paley estimates with applications to Toeplitz and integration operators on weighted Bergman spaces. Banach J. Math. Anal. 17(1), 10 (2023)
Eisner, T., Grivaux, S.: Hilbertian Jamison sequences and rigid dynamical systems. J. Funct. Anal. 261(7), 2013–2052 (2011)
El-Fallah, O., Elmadani, Y., Kellay, K.: Kernel and capacity estimates in Dirichlet spaces. J. Funct. Anal. 276(3), 867–895 (2019)
El-Fallah, O., Elmadani, Y., Labghail, I.: Extremal functions and invariant subspaces in Dirichlet spaces. Adv. Math. 408, 108604 (2022)
El-Fallah, O., Kellay, K., Mashreghi, J., Ransford, T.: A Primer on the Dirichlet Space, vol. 203. Cambridge University Press, Cambridge (2014)
El-Fallah, O., Ransford, T.: Peripheral point spectrum and growth of powers of operators. J. Oper. Theory 8, 89–101 (2004)
Falconer, K.: Fractal geometry. Math. Found. Appl. (1990)
Jamison, B.: Eigenvalues of modulus 1. Proc. Am. Math. Soc. 16(3), 375–377 (1965)
Kahane, J., Salem, R.: Ensembles Parfaits et series trigonometriques. Hermann, Paris (1994)
Mackay, J.M., Tyson, J.T.: Conformal Dimension: Theory and Application, vol. 54. American Mathematical Society (2010)
Nikolskii, N.: Selected problems of weighted approximation and spectral analysis. Trudy Mat. Inst. Akad. Nauk SSSR, p. 120 (1974)
Ortega, J., Fàbrega, J.: Pointwise multipliers and corona type decomposition in \( bmoa\). Annales de l’institut Fourier 46, 111–137 (1996)
Ransford, T.: Eigenvalues and power growth. Israel J. Math. 146(1), 93–110 (2005)
Ransford, T., Roginskaya, M.: Point spectra of partially power-bounded operators. J. Funct. Anal. 230(2), 432–445 (2006)
Acknowledgements
We are very grateful to the referee for the thoughtful comments and detailed suggestions which have considerably contributed to the enhancement of our work.
Funding
No funds, grants, or other support was received.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
El-Fallah, O., Elmadani, Y. & Labghail, I. Ahlfors–David Regular Sets, Point Spectrum and Dirichlet Spaces. Results Math 79, 74 (2024). https://doi.org/10.1007/s00025-023-02098-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00025-023-02098-9