We have some IID sequence, $\left\{ {{X_n}} \right\}_{n = 1}^\infty $, of standard normal random variable on the probability space $\left( {\Omega ,\mathcal{F},\mathbb{P}} \right)$. Also $\left\{ {{\xi _n}} \right\}_{n = 1}^\infty $ is an orthonormal basis for ${L^2}\left( {[0,\infty ),\mathcal{B}\left( {[0,\infty )} \right),\mu } \right)$. Mu refers to the Lebesgue measure. I have been trying to prove that the following sequence converges in $L^2(\mathbb{P})$ without much success. I have been thinking of proving that it is a Cauchy sequence and exploiting the completeness of the space but that hasn't paned out. Any ideas?

$$Y_t^{(k)} = \sum\limits_{n = 1}^k {{X_n}\int\limits_0^\infty {{\xi _n}(u){1_{[0,t]}}(u)d\mu } } (u)$$

  • $\begingroup$ Do you want to prove that you have a Cauchy sequence for each fixed $t$? $\endgroup$ – Davide Giraudo May 23 at 17:16
  • $\begingroup$ Yes, I am trying to prove that it converges for every $t \in [0,\infty )$. $\endgroup$ – Anonymous May 23 at 17:19
  1. The series $\sum_{n\geqslant 1}c_nX_n$ is convergent in $\mathbb L^2$ if $\sum_{n\geqslant 1}c_n^2$ converges (show that the sequence is Cauchy in $\mathbb L^2$ using the fact that $\sum_{n=N}^Mc_nX_n$ has a centered normal distribution with variance $\sum_{n=N}^Mc_n^2$.

  2. For a fixed $t$, define $c_n:=\int\limits_0^\infty \xi _n(u) 1_{[0,t]} (u)d\mu (u)$. Then $c_n=\langle \xi_n,\mathbf 1_{[0,t]}\rangle$ and square summability of $\left(c_n\right)_{n\geqslant 1}$ follows from Parseval's inequality.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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