Optimal dual pairs of frames for erasures
S. Arati, P. Devaraj, Shankhadeep Mondal · 2024
Read Optimal dual pairs of frames for erasures by S. Arati, P. Devaraj, Shankhadeep Mondal (2024) 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-2412-15709.txt.
Introduction
Arati, Devaraj, and Shankhadeep Mondal wrote this mathematics paper, published in 2024. It studies pairs of frames, the redundant coordinate systems used in signal processing, that lose the least information when some of the signal's pieces are erased.
Frames spread a signal across many measurements, so losing a few need not be fatal. But the recovery partner matters: some dual frames rebuild far better than others after erasures. The authors introduce a new score for this, based on the size of the error operator in the Frobenius sense. They characterize the best dual pairs under that score for any number of erasures. They also redo the analysis under two other scores, the spectral radius and the numerical radius. Finally, they give explicit recipes for building certain optimal pairs in finite-dimensional spaces.
Catalog record
| CREATOR | S. Arati, P. Devaraj, Shankhadeep Mondal |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-15709
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Optimal dual pairs of frames for erasures S. Arati, P. Devaraj, Shankhadeep Mondal Abstract The study involves characterizations of dual pairs of frames which are optimal to handle erasures among all dual pairs for a finite dimensional Hilbert space. A new optimality measure using the Frobenius norm of the error operator has been introduced and the corresponding optimal dual pairs have been analyzed for any number of erasures. Also, other measures of the error operator, namely the spectral radius and the numerical radius, have been considered for the analysis. Besides, explicit construction of certain optimal dual pairs has been provided. 2010 MSC 42C15, 46C05, 15A60 1 Introduction Duffin and Schaeffer [ 11 ] introduced the notion of frames which generalizes the concept of bases in a Hilbert space. A vector in a Hilbert space can have several representations using a sequence of frame elements. Erasures of frame coefficients may occur when transmitting data, especially under challenging global network conditions like congestion and limitations in the transmission channel. The redundancy features of frames plays a crucial role in easing the effect of these erasures and improving the precision of reconstructing the original signal. Recent years have witnessed substantial research endeavors aimed at tackling erasure-related challenges through the application of frame theory. Goyal et al. [ 12 ] have analyzed the optimality of unit norm tight frames for handling erasures in a coding theory perspective. They have considered the mean-square error as a measure of optimality and shown that for one erasure, a unit norm tight frame is optimal among all unit norm frames, both in average and worst-case scenarios. Casazza and Kovačević [ 4 ] have extensively investigated equal norm tight frames and analyzed their construction and robustness against erasures. In a related study, Holmes and Paulsen [ 14 ] have introduced the concept of utilizing the operator norm of the error operator as an optimality measure and have classified those dual pairs which minimize the worst case error among all dual pairs of the form (F,F) consisting of Parseval frames and their canonical duals. A similar problem but with an average type optimality measure, considering various possible locations of erasures of a particular length, has been dealt with in [ 3 ] . In the context of a given frame, Lopez and Han [ 21 ] have provided a sufficient condition for the canonical dual to be the only optimal dual with respect to the operator norm. Furthermore, their study involves the analysis of various topological properties of the collection of all optimal duals associated with a given frame. Moreover, in the context as described above, additional characterizations for optimality have been provided in [ 16 ] . Pehlivan et al. [ 22 ] have taken the spectral radius of the error operator as the measure of optimality and have given characterizations of optimal dual frames of a given frame for one erasure. The same problem associated with the occurrence of two erasures has been looked into in [ 23 , 10 ] . Subsequently, many other optimality measures have been considered for the analysis of optimal dual frames. For instance, numerical radius has been introduced as a measure in [ 1 ] , the average of numerical radius and operator norm of the error operator has been considered in [ 9 ] and seminorms of the error operator have been used in [ 15 ] . Interestingly, the erasure problem has also been formulated by incorporating probabilities and weights associated with the erasures. We refer to [ 17 , 18 , 20 , 19 , 2 ] in this connection. In the works discussed above, most of them deal with finding an optimal dual frame for a given fixed frame, under various measures of optimality. Some of them consider the optimal dual pair among pairs of Parseval frames and their canonical duals. In [ 2 ] , the optimality has been analyzed, while taking into account the larger collection of all possible pairs of frames and their duals. However, their study is confined to the cases of spectral radius and operator norm as the optimality measure. In this paper, we further extend the study of the deterministic erasure problem considering all dual pairs to other contexts, such as the Frobenius norm, which have not been studied so far as per our knowledge. More precisely, we give many characterizations of dual pairs which, among all pairs of frames and their duals, minimize the maximum of the Frobenius norm of the error operators associated with all possible locations of erasures of any arbitrary but fixed length. The characterizing conditions pertaining to the particular case of optimality of tight frames and their canonical duals, among all pairs, involve the well-known concept of equiangular frames, which is prominent in quantum information theory. Furthermore, we provide some results related to the spectral radius measure, which turn out to be an improvement of the corresponding results that can be obtained from [ 2 ] . In addition, we have made attempts to construct spectrally optimal dual pairs for both one and two erasures. Besides, we analyse the scenario wherein the numerical radius of the error operator measures the optimality, in the case of one erasure. These problems, associated with the three different measures, have been addressed in Sections 3, 4 and 5 respectively. A comparitive analysis between the three optimal duals is provided in Section 6. We shall now turn to Section 2 to recall the necessary background in connection with the problem of erasures and optimality in reconstruction. 2 Preliminaries on erasures We shall denote an n- dimensional Hilbert space by \mathcal{H} . A finite frame for \mathcal{H} is a finite sequence F=\{f_{i}\}_{i=1}^{N} in \mathcal{H} such that \displaystyle{A\left\|f\right\|^{2}\leq\sum\limits_{i=1}^{N}\left|\langle f,f_{i}\rangle\right|^{2}\leq B\left\|f\right\|^{2}},\forall f\in\mathcal{H}, where A and B are some positive constants and are called lower and upper frame bounds respectively. The supremum over all such lower bounds and the infimum over all such upper bounds are called the optimal frame bounds. If there exists a constant \alpha such that \left\|f_{i}\right\|=\alpha for all i , then the frame F is called an equal norm frame . An equal norm frame F is said to be an equiangular frame if there exists a constant c such that |\langle f_{i},f_{j}\rangle|=c for all i\neq j. A frame F for \mathcal{H} is called a tight frame if there exists a positive constant A such that \sum\limits_{i=1}^{N}\big|\langle f,f_{i}\rangle\big|^{2}=A\left\|f\right\|^{2} for all f\in\mathcal{H}, and is called a Parseval frame if \sum\limits_{i=1}^{N}\big|\langle f,f_{i}\rangle\big|^{2}=\left\|f\right\|^{2} for all f\in\mathcal{H}. There are three important operators, associated with each frame F for \mathcal{H} : 1. Analysis operator \Theta_{F}:\mathcal{H}\to{\mathbb{C}^{N}} given by \Theta_{F}(f)=\{\langle f,f_{i}\rangle\}_{i=1}^{N} , 2. Adjoint of \Theta_{F} , called synthesis operator \Theta_{F}^{*}:\mathbb{C}^{N}\to\mathcal{H} given by \Theta_{F}^{*}\left(\{c_{i}\}_{i=1}^{N}\right)=\sum\limits_{i=1}^{N}{c_{i}f_{i}}. 3. Frame operator S_{F}:\mathcal{H}\to\mathcal{H} defined by S_{F}=\Theta_{F}^{*}\Theta_{F}. The explicit action of the frame operator is S_{F}f=\sum_{i=1}^{N}\langle f,f_{i}\rangle f_{i} , f\in\mathcal{H} . It is an invertible, self-adjoint and positive operator on \mathcal{H}. It is well-known that \{S_{F}^{-1}f_{i}\}_{i=1}^{N} is a frame for \mathcal{H} and is known as the canonical or standard dual of F . It provides the following reconstruction formula: f=\sum\limits_{i=1}^{N}\langle f,S_{F}^{-1}f_{i}\rangle f_{i}, for all f\in\mathcal{H}. In general, for a frame F for \mathcal{H}, a sequence G=\{g_{i}\}_{i=1}^{N} in \mathcal{H} satisfying f=\sum\limits_{i=1}^{N}\langle f,g_{i}\rangle f_{i} for all f\in\mathcal{H},