Semi-min-stable processes (Q1201182)

Let \((Y(t), t\geq 0)\) be a stochastic process in \([0,\infty)\) such that \(\forall n\geq 1\) and for any \(\alpha\in R\) it holds \[ \left(n\min_{1\leq i\leq n} Y_ i(t),\;t\geq 0\right)\overset {d}=\bigl(Y(n^ \alpha t),\;t\geq 0\bigr), \] where \(Y_ i(.)\) are independent copies of \(Y(.)\) and \(\overset {d}=\) means equality of finite-dimensional distributions. The author characterizes this class of processes as the only possible weak limits of processes obtained by taking the minimum, pointwise over \(t\), of \(m\) independent copies of a given process and then rescaling space and time. A spectral representation of such a process is found, too. Finally, a lot of examples are exhibited. Terminology warning. The class considered here should be called min-self-similar rather than semi-min- stable. In extreme value theory the notion (max)semi-stability is being connected with d.f.s satisfying the functional equation \(F(x)=F^ r(ax+b)\) for an \(r>1\), where \(a\), \(b\) are real constants, \(a>0\).