Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $\xi_{tn}(\theta),t=1,\dots,n$ be a real-valued martingale difference array indexed by a parameter $\theta \in \Theta \subset R$, where the set $\Theta$ is compact. Now, for all fixed $\theta \in \Theta$, the law of large numbers,

$\sum_{t=1}^n \xi_{tn} (\theta) \rightarrow 0$ in probability,

is assumed to hold. Then, what conditions make the following UNIFORM law of large numbers hold?

$\sup_{\theta \in \Theta} | \sum_{t=1}^n \xi_{tn}(\theta) | \rightarrow 0$ in probability.

Are there any articles or books about the result?

share|improve this question
    
How is supposed $\xi_{tn}(\theta)$ behave when $\theta$ changes? –  Davide Giraudo Feb 9 '13 at 10:19
    
For all $\theta \in \Theta$, $\xi_{tn} (\theta)$ is assumed martingale difference. –  fkh Feb 10 '13 at 3:38

2 Answers 2

up vote 1 down vote accepted

Even if $\xi$'s are independent (a specific case of your martingale-difference stitation), uniform LLN sometimes holds and sometimes it does not. This type of questions has been studied in Machine learning and, specifically, in Vapnik-Chervonenkis theory. The Glivenko-Cantelli theorem (see wikipedia) describes one situation where the convergence is uniform. More general results can be formulated in terms of VC (Vapnik-Chervonenkis) classes. Perhaps some literature is available for martingales, too.

share|improve this answer

An extension of the uniform Glivenko-Cantelli notion to martingale differences can be found here: www-stat.wharton.upenn.edu/~rakhlin/papers/emp_proc_dep.pdf

share|improve this answer
    
Thanks. This seems to be beneficial. –  fkh Feb 21 '13 at 19:01

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.