On eigenvalues of the kernel \(\frac{1}{2} +\lfloor \frac{1}{xy} \rfloor -\frac{1}{xy}\) (Q2199510)

From MaRDI portal





scientific article; zbMATH DE number 7246532
Language Label Description Also known as
default for all languages
No label defined
    English
    On eigenvalues of the kernel \(\frac{1}{2} +\lfloor \frac{1}{xy} \rfloor -\frac{1}{xy}\)
    scientific article; zbMATH DE number 7246532

      Statements

      On eigenvalues of the kernel \(\frac{1}{2} +\lfloor \frac{1}{xy} \rfloor -\frac{1}{xy}\) (English)
      0 references
      0 references
      11 September 2020
      0 references
      For \( 0< x, y\le 1 \), define the kernel \[ K(x,y)=\dfrac{1}{2}-\left\{\dfrac{1}{xy}\right\}, \] where \( \{t\} \in [0,1) \) denotes the fractional part of \( t\in \mathbb{R} \). It is well-known that every such a kernel has at least one eigenvalue \( \lambda \). That is, there exists a number \( \lambda \ne 0 \), and an associated eigenfunction \(\phi(x)\), with \( 0< \int_0^1 |\phi(x)|^{2}\,dx < \infty\), satisfying \[ \phi(x)=\lambda \int_{0}^{1}K(x,y)\phi(y) \,dy \] almost everywhere, with respect to Lebesgue measure, in \( [0,1] \). Since, the kernel \( K \) is symmetric, that is, \( K(x,y)=K(y,x) \), all eigenvalues of \( K \) are real and so there is no essential loss of generality in considering those eigenfunctions of \( K \) that are real valued. In the paper under review, the author considers the set \[ \mathcal{S}(K) = \{ \lambda: \lambda ~ \text{is an eigenvalue of}~ K\}. \] From the general theory, it is well-known that the set \( \mathcal{S}(K) \) is countable, and it is not hard to see that \( \mathcal{S} (K) \) cannot be a finite set. The main result of the author is the following. Theorem 1. The sets \( \mathcal{S}(K) \cap (-\infty, 0) \) and \( \mathcal{S}(K) \cap (0, \infty) \) are infinite. The interest of the author in the kernel \( K \) is motivated by the appearance of the quadratic form \[\sum_{m=1}^{N}\mu (m) \sum_{n=1}^{N}\mu(n)K\left( \frac{m}{N}, \frac{n}{N}\right)\] in an identity involving the Mertens function \( M(x)=\sum_{m\le x} \mu(n)\), where \( \mu(n) \) is the well-known Möbius function. The proof of Theorem 1 follows from a clever combination of techniques in analytic number theory and some results from the Hilbert-Schmidt theory.
      0 references
      Mertens function
      0 references
      eigenvalue
      0 references
      symmetric kernel
      0 references

      Identifiers