---
title: Cyclic Frames in Finite Hilbert Spaces
url: https://www.emergentmind.com/topics/cyclic-frames
type: topic
---

# Cyclic Frames in Finite Hilbert Spaces

Searching arXiv for recent and foundational papers on cyclic frames in frame theory.
arxiv.search(query="cyclic frames finite-dimensional Hilbert spaces harmonic frames", max_results=10, sort_by="relevance")【อ่านข้อความเต็มjson
[{"arxiv_id":"2508.17088","version":"v1","title":"Cyclic frames in finite-dimensional Hilbert spaces","authors":["Ole Christensen","Ritu Redhu","Vaibhav Kumar Shukla"],"categories":["math.FA"],"published":"2025-08-23","abstract":"Generalizing a definition by Kalra \\cite{Kalra}, the purpose of this paper is to analyze cyclic frames in finite-dimensional Hilbert spaces. Cyclic frames form a subclass of the dynamical frames introduced and analyzed in detail by Aldroubi et al. in \\cite{ACM} and subsequent papers; they are particularly interesting due to their attractive properties in the context of erasure problems. By applying an alternative approach, we are able to shed new light on general dynamical frames as well as cyclic frames. In particular, we provide a characterization of dynamical frames, which in turn leads to a characterization of cyclic frames."},{"arxiv_id":"1611.07121","version":"v2","title":"On the number of harmonic frames","authors":["A. G. R. Dayal","D. G. Ralston"],"categories":["math.NT","math.CA"],"published":"2016-11-22","abstract":"There is a finite number $h_{n,d}$ of tight frames of $n$ distinct vectors for $\\mathbb{C}^d$ which are the orbit of a vector under a unitary action of the cyclic group $\\mathbb{Z}_n$. These cyclic harmonic frames (or geometrically uniform tight frames) are used in signal analysis and quantum information theory, and provide many tight frames of particular interest. Here we investigate the conjecture that $h_{n,d}$ grows like $n^{d-1}$. By using a result of Laurent which describes the set of solutions of algebraic equations in roots of unity, we prove the asymptotic estimate $$ h_{n,d} \\approx {n^d \\over \\varphi(n)}\\ge n^{d-1}, \\qquad n\\to\\infty. $$ By using a group theoretic approach, we also give some exact formulas for $h_{n,d}$, and estimate the number of cyclic harmonic frames up to projective unitary equivalence."},{"arxiv_id":"1209.0153","version":"v1","title":"The number of harmonic frames of prime order","authors":["D. G. Mixon","M. C. Parshall"],"categories":["math.FA"],"published":"2012-09-02","abstract":"Harmonic frames of prime order are investigated. The primary focus is the enumeration of inequivalent harmonic frames, with the exact number given by a recursive formula. The key to this result is a one-to-one correspondence developed between inequivalent harmonic frames and the orbits of a particular set. Secondarily, the symmetry group of prime order harmonic frames is shown to contain a subgroup consisting of a diagonal matrix as well as a permutation matrix, each of which is dependent on the particular harmonic frame in question."},{"arxiv_id":"1705.11127","version":"v1","title":"Finite equal norm Parseval Wavelet Frames over Prime Fields","authors":["Sh. Rahimi","K. A. Seddighi"],"categories":["math.FA","42C40"],"published":"2017-05-30","abstract":"In the framework of wave packet analysis, finite wavelet systems are particular classes of finite wave packet systems. In this paper, using a scaling matrix on a permuted version of the discrete Fourier transform (DFT) of system generator, we derive a locally-scaled version of the DFT of system genarator and obtain a finite equal-norm Parseval wavelet frame over prime fields. We also give a characterization of all multiplicative subgroups of the cyclic multiplicative group, for which the associated wavelet systems form frames. Finally, we present some concrete examples as applications of our results."},{"arxiv_id":"1909.06223","version":"v2","title":"Full Spark Frames in the Orbit of a Representation","authors":["Bernhard G. Bodmann","Mitra Ghandehari","Sina Jafarpour","A. Powell"],"categories":["math.FA"],"published":"2019-09-13","abstract":"We present a new infinite family of full spark frames in finite dimensions arising from a unitary group representation, where the underlying group is the semi-direct product of a cyclic group by a group of automorphisms. The only previously known algebraically constructed infinite families were the harmonic, Gabor and Dihedral frames. Our construction hinges on a theorem that requires no group structure. Additionally, we illustrate our results by providing explicit constructions of full spark frames."},{"arxiv_id":"1509.05087","version":"v2","title":"Group Frames with Few Distinct Inner Products and Low Coherence","authors":["Matthew Fickus","Dustin G. Mixon","Christopher A. Nelson","Yang Wang"],"categories":["cs.IT","math.IT"],"published":"2015-09-17","abstract":"Frame theory has been a popular subject in the design of structured signals and codes in recent years, with applications ranging from the design of measurement matrices in compressive sensing, to spherical codes for data compression and data transmission, to spacetime codes for MIMO communications, and to measurement operators in quantum sensing. High-performance codes usually arise from designing frames whose elements have mutually low coherence. Building off the original \"group frame\" design of Slepian which has since been elaborated in the works of Vale and Waldron, we present several new frame constructions based on cyclic and generalized dihedral groups. Slepian's original construction was based on the premise that group structure allows one to reduce the number of distinct inner pairwise inner products in a frame with $n$ elements from $\\frac{n(n-1)}{2}$ to $n-1$. All of our constructions further utilize the group structure to produce tight frames with even fewer distinct inner product values between the frame elements. When $n$ is prime, for example, we use cyclic groups to construct $m$-dimensional frame vectors with at most $\\frac{n-1}{m}$ distinct inner products. We use this behavior to bound the coherence of our frames via arguments based on the frame potential, and derive even tighter bounds from combinatorial and algebraic arguments using the group structure alone. In certain cases, we recover well-known Welch bound achieving frames. In cases where the Welch bound has not been achieved, and is not known to be achievable, we obtain frames with close to Welch bound performance."},{"arxiv_id":"2211.12540","version":"v3","title":"Finite frames and expansions by bounded operators","authors":["Ole Christensen","Ritu Redhu","Vaibhav Kumar Shukla"],"categories":["math.FA"],"published":"2022-11-22","abstract":"If a frame $(f_k)_{k=1}^n$ in a finite-dimensional Hilbert space is linearly independent, then it can be represented as iterates of a linear operator $T$ as ${f_k} = \\, (T^{k-1}f_1)_{k=1}^{n}$; however, as soon as the frame is overcomplete, it was an open problem to determine whether such a representation is possible or not. In this paper we solve this problem completely and characterize the finite-dimensional frames that can be represented as $(T^{k-1}f_1)_{k=1}^{n}$ for a linear operator $T$."},{"arxiv_id":"1602.09012","version":"v1","title":"Spark deficient Gabor frames","authors":["Ioannis G. Malikiosis"],"categories":["math.CA"],"published":"2016-02-29","abstract":"The theory of Gabor frames of functions defined on finite abelian groups was initially developed in order to better understand the properties of Gabor frames of functions defined over the reals. However, during the last twenty years the topic has acquired an interest of its own. One of the fundamental questions asked in this finite setting is the existence of full spark Gabor frames. The author proved the existence, as well as constructed such frames, when the underlying group is finite cyclic. In this paper, we resolve the non-cyclic case; in particular, we show that there can be no full spark Gabor frames of windows defined on finite abelian non-cyclic groups. We also prove that all eigenvectors of certain unitary matrices in the Clifford group in odd dimensions generate spark deficient Gabor frames. Finally, similarities between the uncertainty principles concerning the finite dimensional Fourier transform and the short-time Fourier transform are discussed."},{"arxiv_id":"2110.10630","version":"v1","title":"Full spark Gabor systems from difference sets","authors":["Bernhard G. Bodmann","Navin Singla"],"categories":["math.FA"],"published":"2021-10-20","abstract":"Difference sets are special subsets of finite cyclic groups. We use them to construct Gabor systems in dimensions given by powers of odd primes that are full spark. Our approach gives a natural generalization of the known fact that Gabor systems generated by inverse Fourier transforms of Chu or P4 sequences are full spark in prime dimensions. Specifically, for each prime power $q$, we exhibit a vector $g \\in \\mathbb{C}^q$ with the property that the set of time-frequency shifts of $g$ indexed by any affine line in the vector space $\\mathbb{F}_q^2$ spans a simplex in its eigenspace."},{"arxiv_id":"0712.3291","version":"v1","title":"Dynamical sampling and frame representations with bounded operators","authors":["Akram Aldroubi","Carlos Cabrelli","U. Molter"],"categories":["math.FA","42C15","94A12"],"published":"2007-12-20","abstract":"This paper deals with a dynamical sampling model on a Hilbert space, and is motivated by advances in signal processing ਅਤੇ control theory, as well as by some emerging questions in frame theory. In this setup, the question is when a function $f$ in a Hilbert space may be recovered from given spatial samples of the iterates $A^n f$, where $A$ is a bounded operator on the Hilbert space. The main results consist of necessary and sufficient conditions on the operator $A$ and the set of sampling functionals, so that the ensemble of sampled values allows stable recovery of functions in the Hilbert space. They also characterize the spanning properties of sequences of iterates of vectors by bounded operators."}]
Searching for related foundational work on dynamical frames generated by operator iterates.
arxiv.search(query="finite frames expansions by bounded operators cyclic frames dynamical frames", max_results=10, sort_by="relevance")【อ่านข้อความเต็มjson
[{"arxiv_id":"2211.12540","version":"v3","title":"Finite frames and expansions by bounded operators","authors":["Ole Christensen","Ritu Redhu","Vaibhav Kumar Shukla"],"categories":["math.FA"],"published":"2022-11-22","abstract":"If a frame $(f_k)_{k=1}^n$ in a finite-dimensional Hilbert space is linearly independent, then it can be represented as iterates of a linear operator $T$ as ${f_k} = \\, (T^{k-1}f_1)_{k=1}^{n}$; however, as soon as the frame is overcomplete, it was an open problem to determine whether such a representation is possible or not. In this paper we solve this problem completely and characterize the finite-dimensional frames that can be represented as $(T^{k-1}f_1)_{k=1}^{n}$ for a linear operator $T$."},{"arxiv_id":"0712.3291","version":"v1","title":"Dynamical sampling and frame representations with bounded operators","authors":["Akram Aldroubi","Carlos Cabrelli","U. Molter"],"categories":["math.FA","42C15","94A12"],"published":"2007-12-20","abstract":"This paper deals with a dynamical sampling model on a Hilbert space, and is motivated by advances in signal processing এবং control theory, as well as by some emerging questions in frame theory. In this setup, the question is when a function $f$ in a Hilbert space may be recovered from given spatial samples of the iterates $A^n f$, where $A$ is a bounded operator on a Hilbert space. The main results consist of necessary and sufficient conditions on the operator $A$ and the set of sampling functionals, so that the ensemble of sampled values allows stable recovery of functions in the Hilbert space. They also characterize the spanning properties of sequences of iterates of vectors by bounded operators."},{"arxiv_id":"2508.17088","version":"v1","title":"Cyclic frames in finite-dimensional Hilbert spaces","authors":["Ole Christensen","Ritu Redhu","Vaibhav Kumar Shukla"],"categories":["math.FA"],"published":"2025-08-23","abstract":"Generalizing a definition by Kalra \\cite{Kalra}, the purpose of this paper is to analyze cyclic frames in finite-dimensional Hilbert spaces. Cyclic frames form a subclass of the dynamical frames introduced and analyzed in detail by Aldroubi et al. in \\cite{ACM} and subsequent papers; they are particularly interesting due to their attractive properties in the context of erasure problems. By applying an alternative approach, we are able to shed new light on general dynamical frames as well as cyclic frames. In particular, we provide a characterization of dynamical frames, which in turn leads to a characterization of cyclic frames."},{"arxiv_id":"1903.09493","version":"v1","title":"Dynamical Frames with an Operator Representation","authors":["Marzieh Ehler","Ole Christensen","Mourad Hasannasab"],"categories":["math.FA","42C15"],"published":"2019-03-22","abstract":"We consider frames for separable Hilbert spaces represented as the iterates $(T^k\\varphi)_{k=0}^{\\infty}$ of a bounded linear operator $T$. Such representations have roots in dynamical sampling, and they are highly relevant in connection with a recent operator-theoretic approach to wavelet and Gabor analysis. Our main results characterize when a frame of the form $\\{f_k\\}_{k=1}^{\\infty}$ can be represented in the form $f_k=T^{k-1}f_1$ with a bounded operator $T$; a key criterion is linear independence and invariance of the kernel of the synthesis operator under the right-shift. The results have direct implications for shift-invariant systems, Gabor frames, and wavelet frames."},{"arxiv_id":"2404.12340","version":"v1","title":"Dynamical frame preservation under linear transformations for finite and infinite-dimensional Hilbert spaces","authors":["Mourad Hasannasab"],"categories":["math.FA"],"published":"2024-04-18","abstract":"In frame theory and signal processing, a common task is to reconstruct vectors in a Hilbert space from known data associated with a dynamical frame generated by powers of a bounded operator. This study investigates the properties of the transformed sequence $(A T^n f_0)_{n=0}^{\infty}$, showing that if it remains a frame for the transformed space, it is again a dynamical frame. A key result demonstrates that the preservation of frame properties under linear transformations depends critically on the surjectivity of the transformation, regardless of whether the operator $T$ is bounded or unbounded. As a corollary, a sufficient condition is identified under which any finite frame in a finite-dimensional Hilbert space can be transformed into a dynamical frame, by using an onto map into a Hilbert space of suitable dimension."},{"arxiv_id":"2207.04384","version":"v2","title":"Wave packet frames generated by hyponormal operators on finite-dimensional Hilbert spaces","authors":["Ole Christensen","Mourad Hasannasab"],"categories":["math.FA"],"published":"2022-07-10","abstract":"The notion of wave packet frames is a natural extension of Gabor systems and wavelet systems. We characterize the finite-dimensional frames that can be represented in terms of finite wave packet systems; the result significantly improves an earlier sufficient condition in the literature. Furthermore, we prove that if a frame $(f_k)$ in a finite-dimensional Hilbert space can be represented as a wave packet system generated by a hyponormal operator, then the frame in fact can be represented in terms of an isometric operator. We provide several concrete examples to illustrate the results."},{"arxiv_id":"2204.02215","version":"v2","title":"Wave packet systems in finite-dimensional vector spaces","authors":["Ole Christensen","Mourad Hasannasab"],"categories":["math.FA"],"published":"2022-04-05","abstract":"Motivated by wavelet and Gabor systems, we investigate wave packet systems in finite-dimensional Hilbert spaces. Via a constructive analysis of the synthesis matrix of a frame $(f_k)_{k=1}^N$, we prove that every frame in a finite-dimensional Hilbert space can be represented as a finite wave packet system $(D^j T^k \varphi)_{j=0,\ldots,J,\; k=0,\ldots,K}$ for suitable linear operators $T,D$ and a vector $\varphi$. The result demonstrates a sharp contrast to the infinite-dimensional case, where overcomplete wave packet frames typically require the generating operators to be unbounded."},{"arxiv_id":"1701.09019","version":"v1","title":"Finite Unit Norm Tight Frames and Some Categorical Aspects of Frames","authors":["Stefan Waldmann"],"categories":["math.OA","math.FA"],"published":"2017-01-31","abstract":"We consider finite unit norm tight frames in finite-dimensional Hilbert spaces over the reals and complex numbers. Due to their relation with isometric embeddings and finite sums of rank-one projectors, these objects have applications in signal processing, geometric functional analysis, and symplectic geometry. In the first part we obtain a normal form for finite unit norm tight frames with a clear geometric interpretation. In the second part we show that finite tight frames and Parseval frames can be put into a category-theoretic context by regarding them as finite-dimensional analogues of Hilbert $C^*$-modules."},{"arxiv_id":"2110.10630","version":"v1","title":"Full spark Gabor systems from difference sets","authors":["Bernhard G. Bodmann","Navin Singla"],"categories":["math.FA"],"published":"2021-10-20","abstract":"Difference sets are special subsets of finite cyclic groups. We use them to construct Gabor systems in dimensions given by powers of odd primes that are full spark. Our approach gives a natural generalization of the known fact that Gabor systems generated by inverse Fourier transforms of Chu or P4 sequences are full spark in prime dimensions. Specifically, for each prime power $q$, we exhibit a vector $g \\in \\mathbb{C}^q$ with the property that the set of time-frequency shifts of $g$ indexed by any affine line in the vector space $\\mathbb{F}_q^2$ spans a simplex in its eigenspace."},{"arxiv_id":"2303.13094","version":"v1","title":"Frames of translates with prescribed fine structure in shift invariant spaces","authors":["Mourad Hasannasab","Alejandro Mosquera Ucha"],"categories":["math.FA"],"published":"2023-03-23","abstract":"This paper explores the problem of constructing wave packet systems with prescribed fine structure in shift invariant spaces, motivated by applications in signal processing and data analysis. The authors prove that every shift generated frame of translates in $L^2(\\mathbb{R}^d)$ can be transformed via a bounded invertible operator into a shift generated frame of translates with any prescribed spectrum, provided certain spectral conditions are met. The results address the existence of finite-generator shift generated systems with arbitrary nonnegative pointwise trace and suggest applications to frame design with specified local spectral profiles."}]
Cyclic frames are finite frames generated by repeated application of a single operator. In finite-dimensional Hilbert spaces, a dynamical frame has the form \(f_k=T^{k-1}f_1\), and it is called a cyclic frame of length \(n\) when \(T^n=I_{\mathbb{C}^d}\) [2508.17088]. A central special class is formed by cyclic harmonic frames, also called geometrically uniform tight frames, which arise as orbits under a unitary action of the cyclic group \(\mathbb{Z}_n\) and admit an equivalent Fourier-character description [1611.07121]. The subject sits at the intersection of operator-generated systems, finite Fourier analysis, finite group representations, and structured frame design.

## 1. Definitions and basic models

Frame theory begins with the standard finite-dimensional notion: a sequence \((f_k)_{k=1}^n\subset\mathbb{C}^d\) is a frame for \(\mathbb{C}^d\) if there exist constants \(0<A\le B<\infty\) such that
\[
A\|f\|^2 \le \sum_{k=1}^n |\langle f,f_k\rangle|^2 \le B\|f\|^2
\]
for every \(f\in\mathbb{C}^d\) [2508.17088]. If \(A=B\), the frame is tight; if all vectors have the same norm, it is equal-norm. In the harmonic setting, a sequence \(\{v_j\}_{j=1}^n\) is \(A\)-tight when
\[
\sum_{j=1}^n |\langle f,v_j\rangle|^2 = A\|f\|^2,
\]
equivalently when the frame operator \(S f := \sum_{j=1}^n \langle f,v_j\rangle v_j\) satisfies \(S=A I\) [1611.07121].

A cyclic frame is a special dynamical frame. Given a linear operator \(T:\mathbb{C}^d\to\mathbb{C}^d\) and a vector \(f_1\in\mathbb{C}^d\), the finite sequence
\[
f_k=T^{k-1}f_1,\qquad k=1,\dots,n,
\]
is a dynamical frame if it is a frame, and it is cyclic if in addition \(T^n=I_{\mathbb{C}^d}\) [2508.17088]. If \(n\) is the minimal positive integer with \(T^n=I\), the sequence is called a minimal cyclic frame [2508.17088]. This places cyclic frames within the broader operator-iterate perspective developed in dynamical sampling and frame representations by bounded operators [0712.3291].

Cyclic harmonic frames are the unitary-group counterpart. Fix \(n\ge 1\), let \(\zeta=\exp(2\pi i/n)\), and identify \(\mathbb{Z}_n=\{0,1,\dots,n-1\}\). Such a frame can be described in two equivalent ways: an orbit picture, \(v_k=U^k v_0\) for a unitary \(U\) with \(U^n=I\), or a character picture obtained by choosing \(d\) distinct characters \(\xi_j(k)=\zeta^{a_jk}\) and setting
\[
v_k=(\xi_1(k),\dots,\xi_d(k))^T,\qquad k\in\mathbb{Z}_n.
\]
Every cyclic harmonic frame is unitarily equivalent to one of these Fourier-submatrix constructions [1611.07121].

In the prime-order case \(G\cong \mathbb{Z}_p\), choosing \(d\) distinct generators \(n_1,\dots,n_d\in\mathbb{Z}_p\) gives the classical DFT-FUNTF
\[
\varphi_m=\frac{1}{\sqrt d}\bigl(e^{2\pi i m n_1/p},\dots,e^{2\pi i m n_d/p}\bigr)^T,\qquad m\in\mathbb{Z}_p,
\]
which is the basic model for prime cyclic harmonic frames [1209.0153].

## 2. Characterization by operator structure, eigenstructure, and kernel invariance

The modern finite-dimensional theory gives several exact characterizations of when a frame is cyclic. A useful starting point is the parametrization of dynamical frames: if \(f_1,\dots,f_d\) is any basis of \(\mathbb{C}^d\), \(\varphi\in\mathbb{C}^d\) is arbitrary, and \(n\ge d\), then defining
\[
T f_k=f_{k+1},\quad k=1,\dots,d-1,\qquad T f_d=\varphi
\]
produces a frame \(\{T^{k-1}f_1\}_{k=1}^n\); conversely, every dynamical frame with \(n\ge d\) arises in this way [2508.17088]. This result complements the finite-dimensional representation theorem showing that the overcomplete case can also be characterized completely [2211.12540].

For cyclic frames, the decisive criterion is spectral. Fix integers \(d<n\). If \(T\) is diagonalizable with \(d\) distinct eigenvalues \(\omega_1,\dots,\omega_d\), each an \(n\)th root of unity, and
\[
T=U\,\mathrm{diag}(\omega_1,\dots,\omega_d)\,U^{-1},
\]
then \(\{T^{k-1}\varphi\}_{k=1}^n\) is a cyclic frame whenever \(\varphi=U f_1\) and the coordinates of \(f_1\) are all nonzero. Conversely, if \(\{T^{k-1}\varphi\}_{k=1}^n\) is a cyclic frame, then \(T\) must be diagonalizable with distinct eigenvalues that are \(n\)th roots of unity, and the coordinates of \(\varphi\) in the eigenbasis are all nonzero [2508.17088]. The proof reduces the first \(d\) iterates to a Vandermonde system, so distinct roots of unity are exactly the nondegeneracy condition.

A second characterization is formulated in terms of the synthesis operator. Let \((f_k)_{k=1}^n\subset\mathbb{C}^d\) be any frame with synthesis operator \(\Theta:\mathbb{C}^n\to\mathbb{C}^d\). Then \((f_k)\) is a cyclic frame of length \(n\) if and only if \(\ker\Theta\) is invariant under the right-shift
\[
R(x(1),\dots,x(n))=\bigl(x(n),x(1),\dots,x(n-1)\bigr).
\]
This kernel-invariance criterion is one of the clearest structural signatures of cyclicity [2508.17088].

There is also a constructive circulant formulation. For \(d<n\), choose \(a\in\mathbb{C}^n\) with exactly \(n-d\) nonzero entries, let \(c=\check a\) be its inverse discrete Fourier transform, and form the circulant matrix \(\mathcal{C}_n(c)\). If \(M=\mathrm{range}\,\mathcal{C}_n(c)\subset\mathbb{C}^n\), \(\dim M=n-d\), and \(v_1,\dots,v_d\) is any basis of \(M^\perp\), then the columns of \(V^T\), where \(V=[v_1\ \cdots\ v_d]\), form a cyclic frame of length \(n\) [2508.17088]. This places circulant linear algebra and Fourier duality directly inside the classification.

Low-dimensional examples illustrate the theory. In \(\mathbb{C}^2\), the Mercedes-Benz frame is generated by a rotation matrix \(T\) with \(T^3=I\), yielding a tight, equal-norm, equiangular frame. In \(\mathbb{C}^3\), there is also a cyclic frame of length \(4\) generated by a matrix \(T\) with \(T^4=I\) [2508.17088].

## 3. Harmonic and Fourier-analytic realizations

The harmonic theory specializes cyclicity to finite abelian group characters. For a finite abelian group \(G\) of order \(N\) with characters \((\xi_1,\dots,\xi_N)\), choosing an index set \(J\subset\{1,\dots,N\}\) of size \(d\) and a unitary \(U\) on \(\mathbb{C}^d\) yields a harmonic frame
\[
\Phi=\left\{U(\xi_j(g))_{j\in J}: g\in G\right\}\subset\mathbb{C}^d.
\]
When \(G\cong\mathbb{Z}_N\), these are cyclic harmonic frames, and in the cyclic case every such frame is unitarily equivalent to a Fourier-submatrix construction [1209.0153].

The harmonic picture is especially effective because it turns frame equivalence into a problem about group actions on subsets of frequency indices. For cyclic harmonic frames of \(n\) distinct vectors in \(\mathbb{C}^d\), one counts \(d\)-element subsets \(J\subset \mathbb{Z}_n\) that generate \(\mathbb{Z}_n\), modulo the action of the unit group \(\mathbb{Z}_n^*\). Burnside’s lemma gives the exact formula
\[
m_{n,d}=\frac1{\varphi(n)}\sum_{a\in\mathbb{Z}_n^*}\bigl|\{J\subset \mathbb{Z}_n: |J|=d,\ \langle J\rangle=\mathbb{Z}_n,\ a\cdot J=J\}\bigr|,
\]
and \(h_{n,d}=m_{n,d}\) up to a negligible exceptional correction [1611.07121]. In the prime case, no exceptional cases occur, and one obtains
\[
h_{p,d}=\frac1{p-1}\sum_{j\mid \gcd(p-1,d)} \binom{(p-1)/j}{d/j}\varphi(j).
\]

The asymptotic behavior is equally explicit. For fixed \(d\ge 2\) and any \(\varepsilon>0\),
\[
h_{n,d}=\frac{n^d}{d!\,\varphi(n)}\prod_{p\mid n}(1-p^{-d})\bigl(1+O(n^{-1+\varepsilon})\bigr),
\]
hence
\[
h_{n,d}\asymp \frac{n^d}{\varphi(n)}\ge n^{d-1},\qquad n\to\infty.
\]
The proof separates non-exceptional unitary equivalences, which are exactly \(\mathbb{Z}_n^*\)-orbits of generating \(d\)-subsets, from exceptional torsion-point solutions controlled via Laurent’s theorem [1611.07121].

For prime order \(p\), a more refined exact enumeration is available through orbit sizes under the multiplicative action of \((\mathbb{Z}_p)^\times\) on unordered generator classes
\[
\mathcal{A}_p^d=\{[n_1,\dots,n_d]: n_j\in\mathbb{Z}_p\ \text{distinct}\}.
\]
Proposition 4.3 establishes a bijection between the orbits of this action and unitary-equivalence classes of harmonic frames [1209.0153]. If \(c>1\) divides \(p-1\) and satisfies \(c\mid d\) or \(c\mid(d-1)\), the numbers \(\alpha_c\) are defined recursively by
\[
\alpha_c =
\frac{(p-1)(p-1-c)\cdots\bigl(p-1-(\tfrac d c -1)c\bigr)}{c^{\tfrac d c -1}\,(d/c)!}
-\frac c{p-1}\sum_{\substack{b>c\\ b\mid d}} \frac{p-1}{b}\,\alpha_b,
\qquad c\mid d,
\]
and
\[
\alpha_c =
\frac{(p-1)(p-1-c)\cdots\bigl(p-1-(\tfrac{d-1}c -1)c\bigr)}{c^{\tfrac{d-1}c -1}\,((d-1)/c)!}
-\frac c{p-1}\sum_{\substack{b>c\\ b\mid d-1}} \frac{p-1}{b}\,\alpha_b,
\qquad c\mid(d-1),
\]
with
\[
\alpha_1=\frac1{p-1}\binom p d
-\sum_{\substack{c>1\\ c\mid d}}\frac{\alpha_c}{c}
-\sum_{\substack{c>1\\ c\mid(d-1)}}\frac{\alpha_c}{c}.
\]
The total number of inequivalent prime-order harmonic frames is then
\[
\alpha_1+\sum_{\substack{c>1\\ c\mid d}}\alpha_c+\sum_{\substack{c>1\\ c\mid(d-1)}}\alpha_c
\]
[1209.0153].

This enumeration resolves a basic growth problem: the exact recursion yields the asymptotic \(\alpha_1=O(p^{d-1})\), settling the conjectural \(O(p^{d-1})\) behavior for inequivalent prime-order harmonic frames [1209.0153].

## 4. Equivalence, symmetry, and intrinsic geometry

Unitary equivalence is the natural classification relation for finite harmonic frames. Two FUNTFs \(\Phi,\Psi\subset\mathbb{C}^d\) of the same size are equivalent if there exists \(U\in U(d)\) with \(U\Psi=\Phi\) as sets [1209.0153]. In the prime cyclic case, if the generators of \(\Phi\) are \(n=(n_1,\dots,n_d)\) and those of \(\Psi\) are \(n'=(n'_1,\dots,n'_d)\), then \(\Phi\approx\Psi\) if and only if there exist permutations \(\sigma_1\in S_p\), \(\sigma_2\in S_d\) such that
\[
\phi_m(k)=\psi_{\sigma_1(m)}(\sigma_2(k)),\qquad \forall m,k.
\]
This replaces an apparently analytic equivalence problem by a combinatorial one on exponent tuples [1209.0153].

The symmetry group of a prime-order harmonic frame admits a canonical subgroup. If \(\Phi_n\) corresponds to an orbit of size \((p-1)/c\), then its full symmetry group contains the subgroup generated by the diagonal unitary
\[
D=\mathrm{diag}(\omega^{n_1},\dots,\omega^{n_d}),\qquad \omega=e^{2\pi i/p},
\]
and a permutation matrix \(Q\) of order \(c\) that cyclically permutes each \(c\)-cycle among the generators \(n_k^c\). Thus \(\langle D,Q\rangle\) is a subgroup of \(\mathrm{Sym}(\Phi_n)\) isomorphic to \(\mathbb{Z}_p\times \mathbb{Z}_c\) in the generic case \(c>1\), while for \(c=1\) the symmetry reduces to the diagonal subgroup \(\langle D\rangle\) of order \(p\) [1209.0153].

For general cyclic frames, tightness imposes strong operator-theoretic consequences. If \(\{T^{k-1}f_1\}_{k=1}^n\) is a cyclic frame with bounds \(A,B\), then
\[
1\le \|T\|\le \sqrt{B/A},\qquad 1\le \|T^{-1}\|\le \sqrt{B/A}.
\]
Moreover, if the frame is tight and cyclic, then \(T\) is unitary, and therefore the frame is equal-norm. In this case it is equiangular if and only if \(|\langle T^\ell f_1,f_1\rangle|\) is constant for \(\ell=1,\dots,n-1\) [2508.17088]. The canonical dual is again cyclic:
\[
\{S^{-1}T^{k-1}f_1\}=\{(S^{-1}TS)^{k-1}S^{-1}f_1\},
\]
with \(S^{-1}TS=(T^*)^{-1}\) [2508.17088].

These facts clarify a common misconception: cyclicity alone does not force the generator \(T\) to be unitary. Unitarity is guaranteed in the tight cyclic case, while the general characterization only requires diagonalizability with distinct roots of unity as eigenvalues [2508.17088]. A plausible implication is that the operator-theoretic notion of cyclic frame is strictly broader than the harmonic tight-orbit notion, even though the two theories share many algebraic mechanisms.

## 5. Coherence, spark, and related orbit constructions

Cyclic group structure is particularly useful for controlling coherence. For a cyclic group \(C_n\) generated by a unitary \(U\) with \(U^n=I\), the orbit
\[
\{v,\ Uv,\ U^2v,\ \dots,\ U^{n-1}v\}\subset\mathbb{C}^m
\]
is a cyclic frame [1509.05087]. In the prime-order diagonal construction, if \(m\mid (n-1)\), \(r=(n-1)/m\), and \(K\subset(\mathbb{Z}/n\mathbb{Z})^\times\) is the unique subgroup of size \(m\), then the vectors
\[
f_\ell=\frac1{\sqrt m}\begin{pmatrix}
\omega^{k_1\ell}\\
\omega^{k_2\ell}\\
\vdots\\
\omega^{k_m\ell}
\end{pmatrix},\qquad \ell=0,\dots,n-1,
\]
form a unit-norm tight frame in \(\mathbb{C}^m\), with exactly \(r=(n-1)/m\) distinct nontrivial inner-product values [1509.05087]. If the off-diagonal magnitudes take exactly \(r\) values equally often, then
\[
\mu\le \sqrt r\sqrt{\frac{n-m}{m(n-1)}},
\]
and in the prime cyclic case a sharper bound is obtained:
\[
\mu \le \frac1r\Bigl(\frac1m+(r-1)\sqrt{\frac1m\Bigl(r+\frac1m\Bigr)}\Bigr)
\]
[1509.05087]. In special difference-set cases, the cyclic harmonic frame attains the Welch bound [1509.05087].

Spark phenomena show a parallel dependence on cyclic structure. For \(G=\mathbb{Z}_n\rtimes H\), where \(H\) is a subgroup of automorphisms, irreducible unitary representations produce orbit frames \(\{\pi(g)v:g\in G\}\). When \(n\) is prime and \(H<(\mathbb{Z}_n)^\times\) is arbitrary, every non-character irreducible \(\pi\) of \(\mathbb{Z}_n\rtimes H\) on \(\mathbb{C}^{|H|}\) is full spark, and almost every \(v\in\mathbb{C}^{|H|}\) yields a full spark equal-norm tight frame of size \(n|H|\) [1909.06223]. This extends the previously known algebraic families of harmonic, Gabor, and Dihedral frames.

A closely related dichotomy appears for finite Gabor frames. If the underlying finite abelian group \(G\) is cyclic, then almost all windows generate a full-spark Gabor frame; if \(G\) is finite abelian and non-cyclic, then every nonzero window generates a spark-deficient Gabor frame [1602.09012]. This shows that cyclicity is not a superficial group-theoretic convenience: it can determine whether maximal linear independence is generically possible.

Prime-field wavelet constructions provide another orbit-based analogue. Let \(G=(\mathbb{Z}/p\mathbb{Z})^\times\) be cyclic of order \(p-1\), let \(H\le G\) be a cyclic subgroup of order \(M\), and let \(W(g;H)=\{\psi_{m,k}=T_kD_m g: m\in H,\ k\in\mathbb{Z}/p\}\). If \(\widehat g(0)\neq 0\) and the Fourier transform \(\widehat g\) does not vanish on any coset of \(H\), then \(W(\Psi^*e_0;H)\) is a Parseval frame for \(\mathbb{C}^p\), and all its elements have the same norm [1705.11127]. This is not usually called a cyclic frame, but it uses the same cyclic-subgroup and Fourier-block mechanisms that dominate the harmonic theory.

Applications track these structural advantages. Cyclic harmonic frames are used in signal analysis and quantum information theory [1611.07121]. More broadly, group and cyclic constructions are used in robust transmission under erasures, compressive sensing, spherical coding, MIMO communications, and quantum sensing [1509.05087]. For Parseval frames, uniformity minimizes worst-case \(1\)-erasure error, while equiangular Parseval frames minimize worst-case \(2\)-erasure error; tight cyclic frames therefore occupy a natural position in erasure-robust design [2508.17088].

## 6. Extensions, boundaries, and open problems

Several boundaries of the theory are now explicit. The prime-order harmonic case is unusually clean: every harmonic frame is unitarily equivalent to a DFT-FUNTF, no exceptional equivalences occur in the exact counting formula, and symmetry groups admit the explicit subgroup \(\langle D,Q\rangle\) described above [1209.0153] [1611.07121]. Beyond prime order, new phenomena arise because \((\mathbb{Z}_N)^\times\) need not be cyclic, multiple abelian subgroups can generate harmonic frames, and one must handle composite-order roots of unity [1209.0153].

One open problem concerns symmetry. In the prime case, it was conjectured that the subgroup \(\langle D,Q\rangle\) is in fact the full symmetry group \(\mathrm{Sym}(\Phi_n)\) even when \(c>1\) [1209.0153]. More generally, enumerating and describing symmetry groups of non-prime-order harmonic frames remains open [1209.0153].

Another frontier is the operator theory of cyclic frames themselves. The kernel-invariance criterion strongly suggests an infinite-dimensional extension to reproducing kernel Hilbert spaces; classification beyond unitary \(T\) and the behavior of cyclic frames under \(3\) or more erasures are also identified as open directions [2508.17088]. Since cyclic frames are a subclass of dynamical frames, a plausible implication is that future progress will continue to move in both directions: from finite Fourier-group constructions toward abstract operator-generated systems, and from operator criteria back toward concrete structured families.

Taken together, the current literature presents cyclic frames not as a single isolated construction but as a framework for understanding operator orbits, cyclic group actions, Fourier-submatrix models, exact orbit counting, coherence control, and spark behavior within one algebraic-analytic vocabulary [2508.17088] [1611.07121].

Source: https://www.emergentmind.com/topics/cyclic-frames