# $L^p$ norm of multivariate standard normal random variable

Given $X_i\sim \mathcal{N}(0,1)$ what is the behaviour of $$||X||_{l^p}=(\sum_{i=1}^n|X_i|^p )^{1/p}$$ as $n\rightarrow \infty$? For $p=2$ results about $\chi$-distribution tell us that $$\mathbb{P}(||X||_{l^2}\le 2n^\frac{1}{2} )\rightarrow 1.$$

I am interested in analgous statments for $p\ne1$,i.e.

$$\mathbb{P}(||X||_{l^p}\le Cn^{e(p)} ),$$ where $C$ is allowed to depend on $p$.

-
Are $X_i$ independent? –  Davide Giraudo Nov 4 '12 at 18:14
The case $p=1$ is the easiest :) –  Hans Engler Nov 4 '12 at 18:43
By the law of large numbers for i.i.d. random variables, $$\frac1n\sum_{i=1}^n|X_i|^p\to E[|X_1|^p],$$ almost surely, hence $$\frac1{n^{1/p}}\|X\|_p\to E[|X_1|^p]^{1/p},$$ almost surely. In particular, for every positive $x$, $$P[(1-x)n^{1/p}E[|X_1|^p]^{1/p}\leqslant\|X\|_p\leqslant(1+x)n^{1/p}E[|X_1|^p]^{1/p}]\to1.$$
Hint: Assuming $X_i$'s are independent you can apply the Central Limit Theorem on $$\frac{1}{n}\left\Vert X\right\Vert_{\ell^p}^p=\frac{1}{n}\sum_{i=1}^n\lvert X_i\rvert ^p.$$