Numbers of near bivariate record-concomitant observations (Q716170)

Let \(Z_1=(X_1,Y_1)\), \(Z_2=(X_2,Y_2)\), \dots be independent copies of the bivariate random vector \(Z=(X,Y)\) with continuous distribution function. The sequence of record values \(X(n)\) and record times \(L(n)\) is defined by \(L(1)=1\), \(L(n+1)=\min\{j:\,j>L(n), X_j>X_{L(n)}\}\), \(X(n)=X_{L(n)}\), \(n\geq 1\). The sequence of record values \(X(n)\) induces the sequence of their concomitants \(Y(n)=Y_i\) if \(X_i=X(n)\). The bivariate record-concomitant observations are \(Z(n)=(X(n),Y(n))\). \(Z_j=(X_j,Y_j)\) is called a near \(n\)\,th bivariate record-concomitant observation if \(L(n)<j<L(n+1)\) and \(Z_j\) belongs to the rectangle \((X(n)-a,X(n))\times (Y(n)-b_1,Y(n)+b_2)\), where \(a\,,b_1,\,b_2>0\). The number of near \(n\)\,th bivariate record-concomitant observations is denoted by \(\xi_n\). In this paper the probability mass function \(P(\xi_n=k)\), \(k\geq 0\), is computed and the asymptotic behavior of \(\xi_n\) as \(n\) increases is determined under different conditions. Various examples are given. New simulation techniques for generating bivariate record-concomitant observations, the number of near record observations and the number of near bivariate record-concomitant observations conclude this article.