Central limit theorems for multicolor urns with dominated colors (Q983172)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Central limit theorems for multicolor urns with dominated colors
scientific article

    Statements

    Central limit theorems for multicolor urns with dominated colors (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    3 August 2010
    0 references
    The urn contains \(a_j>0\) balls of color \(j \in \{1,\ldots,d\}\), where \(d \geq 2\). At each time \(n \geq 1\), a ball is drawn and then replaced with a random number of balls of the same color. Let \(X_{n,j}\) denote the indicator of \{ball of color \(j\) at time \(n\)\}. Say that \(A_{n,j} \geq 0\) balls of color \(j\) are added to the urn when \(X_{n,j}=1\). Assume that \[ EA_{n,j}=EA_{n,1}, \quad n \geq 1, 1 \leq j \leq d_0, \] \[ \liminf_n EA_{n,1}>\limsup_n EA_{n,j}, \quad j>d_0, \] for some integer \(1 \leq d_0 \leq d\). When \(d_0<d\) some colors (those labeled from \(d_0+1\) to \(d\)) are dominated by the others. Let \( N_{n,j}=a_j+ \sum_{k=1}^n X_{k,j} A_{k,j}\) be the number of balls of color \(j\) in the urn at time \(n\) and \(Z_{n,j}=N_{n,j}/ \sum_{i=1}^d N_{n,i}\), \(Z_{n,j}^{\ast}=N_{n,j}/ \sum_{i=1}^{d_0}N_{n,i}\), \(M_{n,j}^{\ast}=\sum_{k=1}^n X_{k,j} /(1+\sum_{i=1}^{d_0} \sum_{k=1}^n X_{k,i})\). Fix \(j\leq d_0\) and \(n \to \infty\). Under some conditions, \(Z_{n,j} \to Z_{(j)}\) a.s., where \(Z_{(j)}\) is a random variable such that \(Z_{(j)}>0\) a.s. The authors give conditions for \[ (C_{n,j}^\ast, D_{n,j}^\ast) \to \mathcal N(0,U_j) \times \mathcal N(0,V_j) \text{ stably,} \] where \( C_{n,j}^\ast= \sqrt n (M_{n,j}^\ast-Z_{n,j}^\ast), D_{n,j}^\ast=\sqrt n (Z_{n,j}^\ast-Z_{(j)})\) and \(U_j,V_j\) are certain random variables. (\(\mathcal N(a,b)\) denote the one-dimensional Gaussian law with mean \(a\) and variance \(b\)).
    0 references
    0 references
    central limit theorem
    0 references
    clinical trials
    0 references
    random probability measure
    0 references
    stable convergence
    0 references
    urn model
    0 references
    0 references
    0 references