Smooth PONS (Q5929345): Difference between revisions

From MaRDI portal
Added link to MaRDI item.
Set OpenAlex properties.
 
(2 intermediate revisions by 2 users not shown)
Property / MaRDI profile type
 
Property / MaRDI profile type: MaRDI publication profile / rank
 
Normal rank
Property / cites work
 
Property / cites work: Generalized Rudin-Shapiro systems / rank
 
Normal rank
Property / cites work
 
Property / cites work: Quadrature mirror filters, low crest factor arrays, functions achieving optimal uncertainty principle bounds, and complete orthonormal sequences -- a unified approach / rank
 
Normal rank
Property / cites work
 
Property / cites work: On Golay polynomial pairs / rank
 
Normal rank
Property / cites work
 
Property / cites work: Sums of Powers of Independent Random Variables / rank
 
Normal rank
Property / cites work
 
Property / cites work: Some Theorems on Fourier Coefficients / rank
 
Normal rank
Property / OpenAlex ID
 
Property / OpenAlex ID: W3187795148 / rank
 
Normal rank

Latest revision as of 10:34, 30 July 2024

scientific article; zbMATH DE number 1584630
Language Label Description Also known as
English
Smooth PONS
scientific article; zbMATH DE number 1584630

    Statements

    Smooth PONS (English)
    0 references
    0 references
    0 references
    0 references
    30 April 2002
    0 references
    PONS stands for `Prometheus orthonormal system', a trademark of Prometheus Incorporated. The original PONS construction gave an orthonormal set \( \{p_{n}\}\) of functions taking values in \(\{0,\pm 1\}\) that are rapidly computable, but, importantly, PONS elements are well-localized: \(|p_{n}(x)|\leq \frac{C}{(1+|x|)^{1/2}}\) and \(|\widehat{p}_{n}(\xi)|\leq \frac{C}{(1+|\xi |)^{1/2}}\), a tradeoff which is optimal when \(C\) is independent of \(n\). This tradeoff is optimal in the sense that one cannot have \(|p_{n}(x)|\leq \frac{C}{(1+|x|)^{p}}\) and \(|\widehat{p}_{n}(\xi)|\leq \frac{C}{ (1+|\xi |)^{p}}\) for \(p\) strictly larger than \(1/2\). The original PONS elements are obtained by uniformly subdividing the unit interval into \(2^{n}\) equal subintervals and defining \(p_{n}\) (resp. \(q_{n}\)) to equal \(\pm 1\) on the \(i\)th subinterval according to whether the \(i\)th coefficient of the standard degree \(2^{n}-1\) Rudin-Shapiro polynomial \(P_{n}\) (resp. \(Q_{n}\)) is \(\pm 1\). Thus the PONS inherit desirable properties of of the Rudin-Shapiro polynomials, in particular, orthogonality and that their crest factors -- the ratio of their \(L^{\infty }\)-norm to their \(L^{2}\)-norm is at most \(\sqrt{2}\). However, transplanting the Rudin-Shapiro polynomials in this way does not result in a complete set. The rest of the PONS construction comes from extending the Rudin-Shapiro recursion relation to include extra `sign flips' at each step in order to establish completeness. The purpose of the present paper is to extend this PONS construction to obtain an orthonormal basis of smooth functions that retains the ideal inverse square root decay in time and frequency along with bounded crest factors. The basic observation is that one can employ the same sequence of sign flips used in the PONS construction to transform any orthonormal basis into another. If one starts with the Fourier basis (thought of as a basis for \(L^{2}\) of a subinterval of the real line) then the PONS construction yields a basis with such properties. The basis elements are not smooth on the whole real line because they are cut-off at the endpoints of the interval.
    0 references
    orthonormal basis
    0 references
    Walsh functions
    0 references
    uncertainty principle
    0 references
    Rudin-Shapiro polynomials
    0 references
    completeness
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references