Eigenvalues in the large sieve inequality. II (Q614481)

From MaRDI portal





scientific article; zbMATH DE number 5831868
Language Label Description Also known as
default for all languages
No label defined
    English
    Eigenvalues in the large sieve inequality. II
    scientific article; zbMATH DE number 5831868

      Statements

      Eigenvalues in the large sieve inequality. II (English)
      0 references
      0 references
      3 January 2011
      0 references
      This paper is a continuation of the author's earlier work [Funct. Approximatio, Comment. Math. 37, Part 2, 399--427 (2007; Zbl 1181.11059)]. It investigates the distribution of the eigenvalues of the Hermitian form \[ \sum_{q\leq Q}\sum_{a:(a,q)=1}\left|\sum_{n\leq N}\phi_ne(an/q)\right|^2 \] for \(N=\sum_{q\leq Q}\phi(q)\). The usual estimate for the large sieve shows that the maximal eigenvalue is at most \(N+Q^2\sim(4.29\ldots)N\). The paper begins by improving this to a bound asymptotically \((3.547\ldots)N\), following a method suggested by Selberg. The rest of the article describes a numerical investigation into the distribution of the eigenvalues. If \(M\) is the matrix attached to the above form then the trace \(M^k\) will be the \(k\)-th moment of the eigenvalues. This was calculated up to \(k\leq 100\), for \(Q=40, 46, 50, 54\) and \(58\). Using the moments the author tries to see whether the eigenvalues might have a limiting distribution. He does this via an interpolation technique using Jackson polynomials. The numerical evidence is somewhat inconclusive, and it looks as though the limiting density, if it exists, would be rather irregular.
      0 references
      large sieve
      0 references
      eigenvalues
      0 references
      upper bound
      0 references
      numerical study
      0 references
      limiting distribution
      0 references

      Identifiers