The Lifting Property for Frame Multipliers and Toeplitz Operators
Peter Balazs, Karlheinz Gröchenig · 2025
Read The Lifting Property for Frame Multipliers and Toeplitz Operators by Peter Balazs, Karlheinz Gröchenig (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-18441.txt.
Introduction
Peter Balazs and Karlheinz Gröchenig prove that a broad class of operators built from frames can always be undone. Frame multipliers generalize Toeplitz operators: they analyze a signal against a frame, multiply by a symbol, and rebuild. Their boundedness on the Banach spaces tied to frames is well understood, but invertibility was open. The catalog title carries funding text, so the clean head title governs here.
For multipliers with a positive symbol, the 2025 paper shows the operator is an isomorphism between the matching coorbit spaces, a lifting result in the style of classical theorems for smoothness spaces. Applied to Gabor frames, it yields a new lifting theorem between modulation spaces; applied to Fock spaces, it matches weighted versions of them. The proofs lean on localized frames and on matrix algebras that stay closed under inversion.
Catalog record
| CREATOR | Peter Balazs, Karlheinz Gröchenig |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-18441
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Lifting Property for Frame Multipliers and Toeplitz Operators Thanks: This research was funded in whole or in part by the Austrian Science Fund (FWF) LOFT - Grant-DOI 10.55776/P34624 . For open access purposes, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission. Peter Balazs, Karlheinz Gröchenig Abstract . Frame multipliers are an abstract version of Toeplitz operators in frame theory and consist of a composition of a multiplication operator with the analysis and synthesis operators. Whereas the boundedness properties of frame multipliers on Banach spaces associated to a frame, so-called coorbit spaces, are well understood, their invertibility is much more difficult. We show that frame multipliers with a positive symbol are Banach space isomorphisms between the corresponding coorbit spaces. The results resemble the lifting theorems in the theory of Besov spaces and modulation spaces. Indeed, the application of the abstract lifting theorem to Gabor frames yields a new lifting theorem between modulation spaces. A second application to Fock spaces yields isomorphisms between weighted Fock spaces. The main techniques are the theory of localized frames and existence of inverse-closed matrix algebras. 2020 Mathematics Subject Classification 47A60,42C15 1. Introduction A lifting theorem establishes an isomorphism between spaces of functions, distributions or sequences and their weighted versions. The basic example is the multiplication operator with a multiplier sequence (\mu_{k}) , given by (1.1) c=(c_{k})_{k\in K}\mapsto(\mu_{k}c_{k})_{k\in K}\,. It maps the weighted \ell^{p} -space \ell^{p}_{\mu}(K) with norm \|c\|_{\ell^{p}_{\mu}}=\Big(\sum_{k\in K}|c_{k}|^{p}\mu_{k}^{p}\Big)^{1/p} to the unweighted \ell^{p}(K) over the index set K . The multiplication operator establishes an isomorphism between those spaces and is called a lifting operator. By means of a lifting operator one may often reduce the study of functions or operators to the unweighted case. Although this example is not more than a definition, it becomes interesting when a function space is described by the behavior of its coefficients with respect to a basis. For example, the usual Sobolev space on the torus is characterized by its Fourier coefficients as \mathbb{H}^{s}(\mathbb{T})=\{f\in L^{2}(\mathbb{T}):\sum_{k\in\mathbb{Z}}|\hat{f}(k)|^{2}(1+4\pi^{2}|k|^{2})^{s}<\infty\} [ 44 ] . Let \mu_{k}=(1+4\pi^{2}|k|^{2})^{\nu/2} be a polynomial weight function, then the operator M_{\mu}f(x)=\sum_{k\in\mathbb{Z}}\mu_{k}\hat{f}(k)e^{ikx}=(\mathrm{I}-\frac{d^{2}}{dx^{2}})^{\nu/2}f(x) is a bijection between the Sobolev spaces \mathbb{H}^{s}(\mathbb{T}) and \mathbb{H}^{s-\nu}(\mathbb{T}) . For the Sobolev space \mathbb{H}^{s}(\mathbb{T}) the underlying basis is the Fourier basis, and a function in \mathbb{H}^{s}(\mathbb{T}) is characterized by the size of its coefficients with respect to this basis. The lifting operator can be described as a multiplication operator on the coefficients, i.e., as a Fourier multiplier, or directly as a a fractional power of the derivative. Similar phenomena can be observed for Besov spaces and modulation spaces (see below). The situation becomes more subtle when a function space is no longer described with respect to a basis, but with respect to a frame, which is a form of overcomplete system with stable series expansions. To illustrate this substantial difference, assume first that \Psi=\{\psi_{k}:k\in K\} is an orthonormal basis. Then the mapping f\mapsto\langle f,\psi_{k}\rangle_{k\in K} is a bijection between the spaces \mathcal{H} and \ell^{2}(K) . Thus for bounded symbols \mu=\left(\mu_{k}\right)_{k\in K} the sequence \mu_{k}\langle f,\psi_{k}\rangle,k\in K, is always in the range of the analysis operator defined by the sequence \left(C_{\Psi}f\right)_{k}=\left<f,\psi_{k}\right> . Consequently, there is always h\in\mathcal{H} , such that \mu_{k}\langle f,\psi_{k}\rangle=\langle h,\psi_{k}\rangle . By contrast, if \Psi is a redundant frame, then \mathrm{ran}\,C_{\Psi}\neq\ell^{2}(K) and in general (\mu_{k}\langle f,\psi_{k}\rangle)\not\in\mathrm{ran}\,C_{\Psi} . In this case it is still meaningful, but more intricate, to describe a space by the size of its coefficients with respect to the given frame \Psi . The additional difficulties introduced by the redundancy can also be seen by the fact that the (synthesis) operator c\mapsto\sum c_{k}\psi_{k} possesses a non-trivial kernel and coefficients in the series expansion are no longer unique. For the description of associated spaces one therefore uses a special set of coefficients, namely those with respect to the dual frame \{\tilde{\psi_{k}}:k\in K\} , so that every f possesses a (non-unique) series expansion f=\sum_{k\in K}\langle f,\tilde{\psi_{k}}\rangle\,\psi_{k} . Whereas initially such expansions hold only in a Hilbert space, it is often possible to define associated spaces - \mathcal{H}^{p}_{m} - by imposing a weighted \ell^{p} -norm on these canonical coefficients via (1.2) \|f\|_{\mathcal{H}^{p}_{m}}=(\big(\sum_{k\in K}|\langle f,\tilde{\psi_{k}}\rangle|^{p}m_{k}^{p}\Big)^{1/p}\,. It is not completely obvious how and when this is a meaningful definition. The decisive assumption is that the frame is localized, and the theory of these spaces is treated in the context of localized frames [ 30 ] . We will provide the necessary background in Section 3 . It is now tempting to follow the example of Sobolev spaces and consider a map (1.3) \mathit{M}_{\mu,\Psi}f=\sum_{k\in K}\mu_{k}\,\langle f,\tilde{\psi_{k}}\rangle\,\psi_{k}\,. Once \mathcal{H}^{p}_{m} is well-defined, it is easy to see that \mathit{M}_{\mu,\Psi} is bounded between \mathcal{H}^{p}_{m\mu} and \mathcal{H}^{p}_{m} , but it is unclear whether \mathit{M}_{\mu,\Psi} is indeed an isomorphism as in the case of a basis. As stated above the difficulty lies in the fact that the multiplied coefficient sequence \mu_{k}\,\langle f,\tilde{\psi_{k}}\rangle need not be of the form \langle h,\tilde{\psi_{k}}\rangle and might even belong to the kernel of the synthesis operator. An understanding of this phenomenon has been a major challenge for us, and this paper is entirely devoted to obtain a deeper insight into the isomorphism property of \mathit{M}_{\mu,\Psi} . Operators of the form ( 1.3 ) are of interest in their own right and have been investigated both as abstract frame multipliers [ 6 ] and for special frames, e.g., under the name Gabor multipliers [ 28 ] . Especially Gabor multipliers have numerous applications, e.g., as quantization operators in physics [ 2 ] and for time-variant filtering [ 43 ] in signal processing. They are applied in adaptive sound analysis [ 42 ] , noise reduction [ 16 ] , psychoacoustic modeling [ 11 ] , sound morphing [ 45 ] and synthesis [ 22 ] . The perturbation of invertible frame multipliers is studied e.g. in [ 47 , 48 ] , but it remains difficult to show that a given frame multiplier is invertible. Our investigation offers some answers to this question. Our main theorem asserts that a frame multiplier with a positive multiplier sequence {\mu} is a Banach space isomorphism between the Banach spaces associated to the frame. To offer a flavor of our main result, we state a provocatively inexact version of such a result. Result 1 . Under suitable conditions on the frame {\Psi} and the weights m and \mu , the frame multiplier \mathit{M}_{\mu,\Psi} is a Banach space isomorphism between \mathcal{H}_{m\sqrt{\mu}}^{p} and \mathcal{H}^{p}_{m/\sqrt{\mu}} , and for some constants m,M>0 one has the norm equivalence m\|f\|_{\mathcal{H}_{m\sqrt{\mu}}^{p}}\leq\|\mathit{M}_{\mu,\Psi}f\|_{\mathcal{H}^{p}_{m/\sqrt{\mu}}}\leq M\|f\|_{\mathcal{H}_{m\sqrt{\mu}}^{p}}\,. The result as stated is, of course, very imprecise. Our main contribution is to identify appropriate abstract conditions on the frame {\Psi} so that all symbols in the envisioned theorem are meaningful and the results holds rigorously under precisely stated assumptions. This will be done after some preliminary work in Section 4, Theorem 11 . By and large, the proof is motivated by the proof of the lifting theorems for modulation spaces in [ 36 ] . Many analytic arguments there can be translated into the language of abstract frame theory. The additional ideas comes from the theory of localized frames and the spectral invariance of matrix algebras and will be presented in Sections 2 and 3.