Two-dimensional Shannon type expansions via one-dimensional affine and wavelet lattice actions
Abstract
It is rather unexpected, but true, that it is possible to construct reproducing formulae and orthonormal bases of L2 (R2) just by applying the standard one dimensional wavelet action of translations and dilations to the first variable x1 of the generating function (x1,x2), ∈ L2 (R2), i.e., by making use of building blocks u,s(x1,x2)=s-1/2(x1-us,x2), where u∈ R, s>0, in the case of reproducing formulae, and k,m(x1,x2)=2-k/2 (x1-2k m2k,x2 ), where k,m∈ Z, in the case of orthonormal bases. It is possible to compensate the fact, that the second variable x2 is not acted upon, by a careful selection of the generating function . Shannon wavelet tiling of the time-frequency plane R2, a standard illustration of orthogonality and completeness phenomena corresponding to the Shannon wavelet, (2km,2k(m+1)](x) 2-kI(),\, k,m∈ Z, \,I=- (1/2,1] (1/2,1], with x representing time and frequency, is substituted by a phase space tiling of R4 with unbounded, hyperboloid type blocks of the form (2km,2k(m+1)](x1)Σn,l2-kID(n,l)(1) (n,n+1](x2)(l,l+1](2),\, k,m∈ Z where Ir=2-rI, r 1, and D:Z × Z → N is a bijection, an additional parameter of the generating function, needed for the lift from L2(R) to L2(R2). Variables x1,x2 are coordinates of position and variables 1,2 of momentum.
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.