Rates of merge in generalized St. Petersburg games (Q4815550)

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 2098971
Language Label Description Also known as
default for all languages
No label defined
    English
    Rates of merge in generalized St. Petersburg games
    scientific article; zbMATH DE number 2098971

      Statements

      0 references
      7 September 2004
      0 references
      St. Petersburg game
      0 references
      semistable law
      0 references
      central limit theorem
      0 references
      rate of merge
      0 references
      Rates of merge in generalized St. Petersburg games (English)
      0 references
      For \(0<p<1\), \(\alpha>0\), consider the discrete distribution \(P(X=r^{k/\alpha})=q^{k-1}p\), \(k=1,2,\dots\), where \(q=1-p\), \(r=1/q\). The case \(\alpha=1\), \(p=1/2\) corresponds to the St. Petersburg game. In general, \(E(X^\alpha)=\infty\), but \(E(X^\beta)<\infty\) for \(0<\beta<\alpha\). Hence in the case \(\alpha\geq 2\), this distribution is in the domain of attraction of the normal law, with infinite variance if \(\alpha=2\). For \(0<\alpha<2\) however this is not in the domain of attraction of any law, though the law of large numbers holds if \(1<\alpha<2\). Let \(S_n=X_1+X_2+\dots+X_n\), where \(\{X_i\}\) is an i.i.d. sequence of random variables with the above distribution. In the case \(0<\alpha<2\) limiting distribution exists only along subsequences. Classes of certain infinitely divisible distribution functions \(\{G_{\gamma}^{\alpha,p}(\cdot):\, q<\gamma\leq 1\}\) are defined such that suitably centered and normalized \(S_n\) can be well approximated by appropriate members of these classes. Sharp rates of these approximations are given.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references