Phase retrieval for nilpotent groups (Q6170361)

A sequence of vectors \((x_i)_{i \in I}\) in a Hilbert space \(\mathcal{H}\) is said to do phase retrieval if \(f = c g\) for \(f , g \in \mathcal{H}\) and unitary \(c \in \mathbb{C}\), if and only if \(|\langle f , x_i \rangle | = |\langle g , x_i \rangle |\) for all \(i\in I\). Informally speaking, this means that the knowledge of the modulus (with respect to the frame \((x_i)_{i \in I}\)) allows the recovery of the phase. Phase retrieval originated in crystallography and has applications to signal processing, among other areas. The explicit construction of vectors, or frames, doing phase retrieval is still considered an open problem. In the context of representation theory, for a locally compact group \(G\) phase retrieval may be considered via a continuous unitary representation \(\pi\) acting on \(\mathcal{H}_\pi\). Concretely, a vector \(\eta\), as well as the group \(G\), is said to do phase retrieval if the sequence \((\pi(x)\eta)_{x \in G}\) does phase retrieval. Two examples of groups with phase retrieval are the finite Heisenberg groups and the Heisenberg (Lie) groups with the corresponding Schrödinger representation. In both cases, the existence of the vector \(\eta\) doing phase retrieval corresponds to certain ``non-vanishing'' properties of a so-called ambiguity function (see Examples 1 and 2). In this paper, the authors extend the class of groups with phase retrieval. In particular, they show that simply connected, connected nilpotent Lie groups with irreducible representations do phase retrieval (Theorem 12). The case of finite nilpotent group is also considered, where they give sufficient criteria for phase retrieval for \(p\)-groups (Corollary 33). In the nilpotent Lie group case, using Kirillov lemma, the authors use a presentation of the group \(G\) as a semidirect product \(G=G_0 \rtimes \exp(\mathbb{R} X)\) (for certain \(X \in \mathfrak{g}\)) and a representation \(\tau\) of \(G_0\) such that \(\pi =\mathrm{Ind}_{G_0}^G \tau\) to show that \(\pi\) acts partially as a Schrödinger representation in the semidirect product (cf. equation (7)). Then the existence of the vector \(\eta\) and the proof of the phase retrieval property are proved in a similar way to the Heisenberg Lie group case. It is perhaps the finite nilpotent group case that is more interesting. In order to use a similar approach to the Lie group case, the authors prove a \(p\)-group analog of Kirillov lemma (Lemma 22) which appears to be new. Moreover, to give the nonvanishing property mentioned above, for a finite group with unitary representation \((\pi,\mathcal{H}_\pi)\) the authors define the quantities \begin{align*} p_0(\pi,\eta) &= \max_{f \in \mathcal{H}_\pi \setminus \{0\}} \left( 1 - \frac{|\mathrm{supp} (V_\eta f)|}{|G|} \right), \\ p_0(\pi) &= \min\{ p_0(\pi,\eta) \, | \, \eta \in \mathcal{H}_\pi\setminus\{0\} \}, \end{align*} for \(\eta \in \mathcal{H}_\pi\) and where \(V_\eta \psi (g) = \langle \psi, \pi(g) \eta \rangle\). In order to guarantee the phase retrieval property, it is necessary to bound the quantity \(p_0\) for the analog of the representation \(\tau\) of the Lie group case (see Lemma 29 and Corollary 33). Furthermore, the quantity \(p_0(\pi,\eta)\) appears to be of independent interest as the authors show a relation with the full spark property for finite frames, which has applications to quantum information theory and signal processing. The paper is self-contained, contains detailed proofs and remarks, and may be used as an introduction to phase retrieval in representation theory for non-specialists.