Download Almost sure invariance principles for partial sums of weakly by Walter Philipp PDF

By Walter Philipp

Similar probability books

Probability: An Introduction

This new undergraduate textual content bargains a concise creation to likelihood and random techniques. routines and difficulties variety from basic to tough, and the general therapy, even though effortless, comprises rigorous mathematical arguments. Chapters comprise middle fabric for a starting path in likelihood, a remedy of joint distributions resulting in money owed of moment-generating capabilities, the legislations of huge numbers and the significant restrict theorem, and simple random procedures.

Additional info for Almost sure invariance principles for partial sums of weakly dependent random variables

Example text

SECTION 6 Convex Hulls Sometimes interesting random processes are expressible as convex combinations of more basic processes. For example, if 0 ≤ fi ≤ 1 for each i then the study of fi reduces to the study of the random sets {ω : s ≤ fi (ω, t)}, for 0 ≤ s ≤ 1 and t ∈ T , by means of the representation 1 {s ≤ fi (ω, t)} ds. fi (ω, t) = 0 More generally, starting from fi (ω, t) indexed by T , we can construct new processes by averaging out over the parameter with respect to a probability measure Q on T : fi (ω, Q) = fi (ω, t)Q(dt).

The expectation with respect to Pσ on the right-hand side of the last expression is less than n |Fn |1 + 1 Pσ max |σ · f ∗ |. n Dnω The ﬁrst of these terms has a small expectation, because assumption (i) implies uniform boundedness of n1 P|Fn |1 . The second term is bounded by K. 1) it is also less than C max |f ∗ |2 n Dnω 2 + 2 log Mn . √ The square root √ factor contributes at most op ( n) to this bound. The other factor is of order Op ( n), because, for each point in Fnω , |f ∗ |22 = fi2 {Fi ≤ K} ≤ K i≤n Fi .

Dudley (1985) has shown that the Donsker-class property is preserved under the formation of (sequential closures of) convex hulls of classes of functions. ) This gives yet another way of handling processes representable as convex combinations of simpler processes. The same stability property is also implied by the ﬁrst theorem of Talagrand (1987). SECTION 7 Maximal Inequalities Let us now pull together the ideas from previous sections to establish a few useful maximal inequalities for the partial-sum process Sn .