No label defined (Q4731995)

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 4117686
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 4117686

      Statements

      0 references
      0 references
      1988
      0 references
      sequential selection
      0 references
      probability of correct selection
      0 references
      stopping rule
      0 references
      terminal decision rule
      0 references
      Monte Carlo experiments
      0 references
      [For the entire collection see Zbl 0638.00032.] X\({}_{ij1},X_{ij2},..\). are iid scalar random variables, \(X_{ijm}\) with pdf \(\exp [P(x_{ij})Q(\theta_{ij})+R(x_{ij})+S(\theta_{ij})]\) (1\(\leq i\leq a\), \(1\leq j\leq b)\) where \(Q(\theta_{ij})=t+u_ i+v_ j\) with \(u_ 1+...+u_ a=0\) and \(v_ 1+...+v_ b=0\). The ordered values of \(u_ i\) and \(v_ j\) are denoted by \(u_{1'}\leq...\leq u_{a'}\), \(v_{1'}\leq...\leq v_{b'}\), all these values being completely unknown. Most of the paper is devoted to the following selection problem: Select the density function which is associated with \(u_{a'}\) and \(v_{b'}\), the probability of correct selection being at least equal to \(P^*\) whenever \(u_{a'}-u_{a-1'}\geq D_ a\) and \(v_ b-v_{b- 1'}\geq D_ b,\) where \(D_ a\), \(D_ b\) and \(P^*\) are prespecified positive finite values with \((ab)^{-1}<P^*<1\). Sampling is to be sequential and taking a matrix-at-a-time: that is, at the \(m^{th}\) stage of experimentation we observe the random matrix \(X_{ijm}\) (1\(\leq i\leq a\), \(1\leq j\leq b).\) Using the general techniques given by \textit{R. E. Bechhofer}, \textit{J. Kiefer} and \textit{M. Sobel} [Sequential identification and ranking procedures.(1968; Zbl 0208.446)] a stopping rule and a terminal decision rule are constructed to meet the requirements. Monte Carlo experiments show that the procedure results in a substantial reduction in the expected total number of observations compared to procedures which use two independent single-factor experiments. Various special cases are discussed in detail, and it is pointed out that p-factor situations \((p>2)\) can be handled by the same techniques.
      0 references

      Identifiers