Quantile clocks (Q655573)

For a positive random variable \(R\) with continuous increasing distribution function \(F\), let \(Q_R\) denote the inverse function of \(F\). Setting \[ T_R(t):=\int_{[0,t]}Q_R(1-t/s)L(ds), \;\;t\geq 0, \] where \((L(t))_{t\geq 0}\) is a subordinator, the authors investigate some properties of sample paths and the one-dimensional distributions of the process \((T_R(t))_{t\geq 0}\), which they call quantile clock. The quantile clock is an instance of a convoluted subordinator as defined by \textit{C. Bender} and \textit{T. Marquardt} [``Integrating volatility clustering into exponential Lévy models'', J. Appl. Probab. 46, No. 3, 609--628 (2009; Zbl 1188.91239)]. The authors point out conditions that ensure that the sample paths of a quantile clock are continuous and increasing with probability one. They also show that, for each fixed \(t\), \(T_R(t)\) has the same law as \(X(t)\), where \((X(t))\) is a subordinator with explicitly given Lévy measure. Extending this line the authors relate the one-dimensional distribution of the composition of independent quantile clocks to those of the composition of independent subordinators. Under certain conditions on the law of \(R\) the authors demonstrate how to choose the law of \((L(t))\) in order to guarantee that \(X(t)\) has a desired law from Jurek's class \(\mathcal{U}_\delta\) of generalized \(s\)-selfdecomposable distributions. Finally, an interesting decomposition of the gamma subordinator found by the authors is worth mentioning. A number of possible applications is discussed, and many selected illustrating examples are given. For instance, one section is devoted to examples of laws of independent random variables \(R\) and \(Y\) such that \(RY\) has the beta law with parameters \(\delta\) and \(1\).