Metrical results on the discrepancy of Halton-Kronecker sequences (Q431282)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Metrical results on the discrepancy of Halton-Kronecker sequences
scientific article

    Statements

    Metrical results on the discrepancy of Halton-Kronecker sequences (English)
    0 references
    0 references
    0 references
    26 June 2012
    0 references
    For a sequence \((\mathbf{z}_n)_{n\geq 0}\) in the \(d\)-dimensional unit cube \([0,1)^d\) the discrepancy of the sequence is defined by \[ D_N=\sup_{B\subseteq [0,1)^d}|A_N(B)/N-\lambda(B)|, \] where \(A_N(B)=\#\{n:0\leq n<N, a_n\in B\}\) and \(B=\prod_{i=1}^d [a_i,b_i)\) with \(0\leq a_i<b_i\leq 1\) for \(i=1,\ldots d\). If the supremum is extended over all subintervals \(B=\prod_{i=1}^d [0,b_i)\) for \(i=1,\ldots d\), then it is called the star discrepancy \(D^*_N\). Let \(b\geq 2\) be an integer. For \(n=\sum_{i=0}^\infty n_ib^i\) with \(n_i=0,\ldots, b-1\) \((i\geq 0)\) the radical inverse function in base \(b\) is defined by \(\psi_b(n)=\sum_{i=0}^\infty n_ib^{-i-1}\). Let \(b_1, \ldots, b_s\) be pairwise coprime integers and let \(\mathbf{\alpha}=(\alpha_1,\ldots, \alpha_t)\in[0,1)^t\). The sequence \((\mathbf{z}_n)_{n\geq 0}\) defined by \(\mathbf{z}_n=(\psi_{b_1}(n),\ldots, \psi_{b_s}(n),\{\alpha_1n\},\ldots,\{\alpha_tn\})\), \(n=0,1,\ldots\) is called Halton-Kronecker sequence. The authors obtain metric results for the discrepancy of a Halton-Kronecker sequence. They show that for almost all \(\mathbf{\alpha}=(\alpha_1,\ldots, \alpha_t)\in[0,1)^t\) the discrepancy of the Halton-Kronecker sequence \((\mathbf{z}_n)\) satisfies for every \(\varepsilon>0\), \[ ND_N=O\left(\log^{s+t+1+\varepsilon}(N)\right) \] with an implied constant only depending on \(s, t, b_1,\ldots, b_s, \mathbf{\alpha}\) and \(\varepsilon\). Furthermore, they show that for almost all \(\alpha\in [0,1)\) the discrepancy of the sequence defined by \((\psi_{b_1}(n),\ldots, \psi_{b_s}(n),\{\alpha n\})\) satisfies for every \(\varepsilon>0\), \[ ND_N=O\left(\log^{s+1+\varepsilon}(N)\right) \] with an implied constant only depending on \(s, b_1, \ldots, b_s, \alpha\) and \(\varepsilon\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    discrepancy
    0 references
    hybrid sequence
    0 references
    Kronecker sequence
    0 references
    Halton sequence
    0 references
    quasi-Monte Carlo method
    0 references
    0 references