Necessary conditions for the bootstrap of the mean of a triangular array
For ∈(0, 1) an -trimming, ∗, of a probability is a new probability obtained by re-weighting the probability of any Borel set, , according to a positive weight function, ≤1/(1−), in the way ∗()= ()(d). If , are probability measures on euclidean space, we consider the problem of obtaining the best 2-Wasserstein approximation between: (a) a fixed probability and trimmed versions of the other; (b) trimmed versions of both probabilities. These best trimmed approximations...
Page 1