Extreme value laws in dynamical systems for non-smooth observations (Q625531): Difference between revisions

Consider a dynamical system \((\mathcal X, \mathcal B, \mu, f)\), a random variable \(\varphi:\mathcal X \to \mathbb R\cup \{\pm \infty\}\), and a stochastic process \(X_0, \, X_1,\dots \) given by \[ X_n=\varphi\circ f^n, \quad n=0,1,\dots\;. \] Build a new stochastic process \(M_0, \, M_1,\dots\), where \(M_n=\max\{X_0, \, X_1,\dots , X_{n-1} \}\). We say that we have an extreme value law (EVL) for \(M_n\) if there exist sequences \((a_n)\) and \((b_n)\), such that \( a_n(M_n-b_n) \) converges in distribution to a non-degenerate distribution function. Consider now a sequence of sets \((U_n)\), and define the first hitting time \(r_n(x)\) to be \[ r_n(x)=\min \{j\in \mathbb N\cup \{+\infty \}:f^j(x) \in U_n\}. \] We say that the system has hitting time statistics (HTS) for \((U_n)\) if, under a suitable renormalization, \(r_n\) converges in distribution to a non-degenerate distribution function. In this paper, the authors prove the equivalence between the existence of a non-trivial HTS law and EVL in the case of dynamical systems with measures which are not absolutely continuous with respect to the Lebesgue measure. This is a counterpart to the result by the same authors in [Probab. Theory Relat. Fields 147, No. 3--4, 675--710 (2010; Zbl 1203.37021)], where they treat the absolutely continuous case.