Nonparametric estimates and limit theorems in abstract Skorokhod space (Q1193348): Difference between revisions

[For the entire collection see Zbl 0731.62005.] It is impossible to put the space of functions without discontinuities of the second kind \({\mathcal D}[0,I]\) and the multidimensional generalizations \({\mathcal D}^ k[0,I]\) in the class of linear topological spaces since the Skorokhod metric is not continuous with respect to the operation of translation. Nevertheless, extracting the characteristic properties, and thus defining an abstract Skorokhod space, we are able to prove analogues of various theorems. Thus we formulate a lemma which lets us judge the continuity of a linear functional from its kernel and some other geometric lemmas. For a sequence of Borel, independent, identically distributed random variables \(\{ \xi_ i \} \) with values in an abstract Skorokhod space we prove the alternative \[ P(\sum \xi_ i \in H ) = 0, \text{ or } P(\xi_ 1 \in H) = 1,\tag{1} \] where \(H\) is a closed linear subspace of \({\mathcal D}\). We note that the proof of the similar assertion in a Banach space is based on Mazur's theorem which does not hold in our case. The existing theorems on Banach supports of measures are generalized to the case of an abstract Skorokhod space. The results obtained are used in problems of nonparametric estimation.