On the connection of uncertainty principles for functions on the circle and on the real line (Q1416656)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2018214
Language Label Description Also known as
default for all languages
No label defined
    English
    On the connection of uncertainty principles for functions on the circle and on the real line
    scientific article; zbMATH DE number 2018214

      Statements

      On the connection of uncertainty principles for functions on the circle and on the real line (English)
      0 references
      0 references
      0 references
      0 references
      0 references
      16 December 2003
      0 references
      The authors establish a precise relationship between Fourier uncertainty and sampling, by viewing the variances of the classical position and momentum operators as limits of variances of respective periodization and sampling operators, composed with dilation. Let \(f_a(x)=\sqrt{a}f(at)\) denote the usual \(L^2\)-normalized dilate of \(f\), let \(g^{\text{per}}(t)=\sum g(t+2\pi k)\) denote the usual \(2\pi\)-periodization of \(g\) and let \(h^{\text{sam}}(t)=\sum h(k) e^{ikt}\). Poisson summation says that \(g^{\text{per}}(t)=\widehat{g}^{\text{sam}}(t)\) as elements of \(L^2(\mathbb T)\). Here the Fourier transform is defined via \(\hat f(\xi)=\int f(t) e^{- i t\xi\, dt}\). For a unit-norm element \(f\) of \(L^2(\mathbb R)\), the variation of the position operator in state \(f\) is \(\Delta x(f)= \| (\cdot - x_0) f\| _2^2\), where \(x_0 = \int x | f(x)| ^2\), while the variation of the momentum operator is \(\Delta \xi (f) =\| (\cdot -\xi_0)\hat f\| _2^2\), with \(\xi_0=\int \xi | \hat f(\xi)| ^2\). The Heisenberg product of \(f\) on \(\mathbb R\) is \(U_{\mathbb R}(f)=(\Delta x(f) \Delta \xi(f))^{1/2}\). The Heisenberg inequality states that \(U_{\mathbb R}(f)\geq 1/2\) and the only minimizers are appropriately shifted Gaussians. The authors consider compact/discrete analogues of these variances. Set \(\text{ var}_F \{c_k\} = \| (k-\kappa) c_k\| _{\ell^2(\mathbb Z)}^2\) where \(\kappa = \sum k | c_k| ^2\). The analogue of variation of position is trickier to define. For \(f\in L^2[0,2\pi)\) with unit norm, consider the angular variance \(\text{ var}_A(f)= 1/\tau(f)^2 -1\) where \(\tau(f)=\frac{1}{2\pi}\int_{-\pi}^\pi e^{it} | f(t)| ^2 \, dt\). One then defines the toral Heisenberg product \(U_{\mathbb T}(f) =(\text{var}_A(f) \text{var}_F(f))^{1/2}\). After defining \textit{admissible} (dense) classes of functions on \(\mathbb R\) and \(\mathbb T\), respectively, for which the relevant quantities are well-defined, the authors show that the toral Heisenberg product \(U_{\mathbb T}\) is also lower bounded by \(1/2\) and, in fact, there are no equalizers in this case. The main result of the paper realizes the variances for the position and momentum operators on \(\mathbb R\) as limits of corresponding toral variances, namely, for all admissible \(f\), one has \(\lim_{a\to\infty} \frac{1}{a^2} \text{ var}_F(f_a^{\text{per}})=\Delta \xi (f)\) while \(\lim_{a\to\infty} \frac{1}{a^2} \text{ var}_A(f_a^{\text{per}})=\Delta x (f)\). The connection with limits of sampled dilates of \(f\) is made by means of Poisson summation.
      0 references
      uncertainty principle
      0 references
      sampling
      0 references
      Poisson summation formula
      0 references
      toral Heisenberg product
      0 references
      0 references

      Identifiers

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