A new model for record values (Q1759545)

Let \(X_1,X_2,\dots\) be independent copies of a random variable \(X\) with continuous distribution function. Fix an integer \(k\in\mathbb{N}\) and order \(X_1,\dots,X_k\) according to their size: \(X_{1:k}\leq\dots\leq X_{k:k}\). Then \(\left(X_1^{(1)},\dots,X_k^{(1)}\right):= \left(X_{1:k},\dots,X_{k:k}\right)\) is the first record vector. To define the second record vector \(\left(X_1^{(2)},\dots,X_k^{(2)}\right)\), the next \(k\) random variables exceeding \(X_{k:k}\) are needed and ordered. Proceeding like this provides a sequence of record vectors \(\left(X_1^{(n)},\dots,X_k^{(n)}\right)\), \(n\in\mathbb{N}\). Representations of record vectors are obtained in the special cases in which \(X\) follows the standard exponential or the uniform distribution on \((0,1)\). They are based on well-known representations of order statistics in these cases.