2
$\begingroup$

I have a two-index succession of real-valued random variables $x_{t,n}$ such that $\lim_{n\to\infty} x_{t,n} = x_t$, for all $t$ and suitable limit r.v. $x_t$.

I would like to prove that $$\lim_{n\to\infty}\frac1n \sum_{t=1}^n x_{t,n} =\lim_{n\to\infty}\frac1n\sum_{t=1}^n x_t. \tag{*}$$

Of course this does not hold without additional hypotheses, even if the $x_{t,n}$ were numbers instead of random variables. In addition, my random variables are quite nasty: they are neither independent, nor identically distributed, nor bounded in norm (they have a Gaussian-like "tail" at infinity).

What additional hypotheses or criteria are there that could enable me to prove (*)?

$\endgroup$
2
  • 2
    $\begingroup$ For a start you could try searching for "laws of large numbers for triangular arrays" and seeing if that turns up results that help. $\endgroup$ May 10, 2012 at 14:32
  • $\begingroup$ Seem interesting, I'll study the results. Thanks! $\endgroup$ May 10, 2012 at 17:54

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.