Decompositions of Rational Gabor Representations

Abstract

Let = Tk,Ml:k∈Zd,l∈ BZ% d be a group of unitary operators where Tk is a translation operator and Ml is a modulation operator acting on L2( Rd) . Assuming that B is a non-singular rational matrix of order d, with at least one rational non-integral entry, we obtain a direct integral irreducible decomposition of the Gabor representation which is defined by the isomorphism π:( Zm× BZd) d→ where π( θ,l,k) =e2π iθMlTk. We also show that the left regular representation of ( Zm× BZ% d) d which is identified with via π is unitarily equivalent to a direct sum of card( [ ,] ) many disjoint subrepresentations: L0,L1,·s,Lcard( [ ,] ) -1. It is shown that for any k≠ 1 the subrepresentation Lk of the left regular representation is disjoint from the Gabor representation. Furthermore, we prove that there is a subrepresentation L1 of the left regular representation of which has a subrepresentation equivalent to π if and only if B ≤1. Using a central decomposition of the representation π and a direct integral decomposition of the left regular representation, we derive some important results of Gabor theory. More precisely, a new proof for the density condition for the rational case is obtained. We also derive characteristics of vectors f in L2(R)d such that π()f is a Parseval frame in L2(R)d.

0

Turn this paper into a lesson

ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.

Discussion (0)

Sign in to join the discussion.

Loading comments…