# Show convergence using CLT

Let $x_1,\ldots,x_n$ be i.i.d. Bernoulli random variables with parameter $1/2$. Let $S=\sum_{i=1}^nx_i$. Using the Central Limit Theorem, show that $$\frac{|2S-n|}{\sqrt n}$$ is convergent to a standard normal random variable.

Thank you.

-
You don't want the absolute values there. – Robert Israel Jul 11 '12 at 2:24
I am interesting in the result with absolute values. Is it possible to show? – AlexK Jul 11 '12 at 2:34
With absolute values you get convergence to a half-normal distribution – Henry Jul 11 '12 at 2:41

The expectation of $S$ is $\frac{n}{2}$ and its variance is $\frac{n}{4}$ so $$\frac{S - \frac{n}{2}}{\sqrt{\frac{n}{4}}} = \frac{2S - n}{\sqrt{n}}$$ converges to a standard normal distribution by the Central Limit Theorem.
The absolute value of a normal distribution with mean $0$ is a half-normal distribution so $$\frac{|2S - n|}{\sqrt{n}}$$ converges to a half-normal distribution with mean $\sqrt{\frac{2}{\pi}}$.