A local limit theorem on one-sided large deviations for dominated-variation distributions (Q1366368)

From MaRDI portal
scientific article
Language Label Description Also known as
English
A local limit theorem on one-sided large deviations for dominated-variation distributions
scientific article

    Statements

    A local limit theorem on one-sided large deviations for dominated-variation distributions (English)
    0 references
    0 references
    0 references
    28 October 1997
    0 references
    Let \(X_n\), \(n\in\mathbb N\), be i.i.d. random variables with values in \(\mathbb Z\), and define \(S_n=X_1+\dots+X_n\). Assume that \(E[X_1]\) exists and, for ease of notation, assume that it is zero. Define \(p_m=P(X_1=m)\) for \(m\in\mathbb Z\) and assume that \(p_{m-u}=p_m(1+o(1))\) as \(m\to\infty\) for every \(u\in\mathbb N\). Under additional assumptions on this heavy-tailed distribution, which are weaker than the ones used by \textit{R. A. Doney} [Math. Proc. Camb. Philos. Soc. 105, No. 3, 575-577 (1989; Zbl 0674.60031)], the author proves that \(P(S_n=m)=np_m(1+o(1))\) where \(m=m(n)\) and \(n\) tends to infinity in such a way that \(\liminf_{n\to\infty}m/n>0\). This behaviour of \(P(S_n=m)\) is determined by the \(n\) disjoint events numbered by \(k\) from \(1\) to \(n\), where \(X_k\) attains an extraordinary large value near \(m\) and \(X_1\), \(\dots\), \(X_{k-1}\), \(X_{k+1}\), \(\dots\), \(X_n\) show their `` typical'' behaviour. For results in the case \(E| X_1| =\infty\) see \textit{R. A. Doney} [Probab. Theory Relat. Fields 107, No. 4, 451-465 (1997; Zbl 0883.60022)].
    0 references
    0 references
    large deviations
    0 references
    heavy tails
    0 references
    dominated-variation distribution
    0 references
    regularly oscillating
    0 references
    integer-valued random walk
    0 references
    0 references