law of large numbers for $P\{X_k = \pm 2^k\} = \frac{1}{2}$;

545 Views Asked by At

I want to show that the law of large numbers holds/doesn't hold for the sequence of independent random variables $P\{X_k = \pm 2^k\} = \frac{1}{2}$

A sufficient condition is $\frac{s_n}{n} \rightarrow 0$ where $s_n$ is the sum of the variances of the variables. However, this is clearly not true so I'm trying to show that the law of large numbers fails to hold.

1

There are 1 best solutions below

0
On

Assume they are independent, and $S_n=\sum\limits_{k=1}^n X_k$.

Actually for this simple example we can list all possibilities of the outcome: using induction or binary system, we know $S_n$ is symmetric with $2^{n-1}$ positive outcomes with equal probability $1/2^{n}$: $2,6,10,\cdots,2^{n+1}-2$. So there are at most $\lfloor (n+2)/4\rfloor<n$ number of the outcomes can be larger than $n$, which means $$P(|\frac{S_n}{n}|>1)\geq 1-\frac{n}{2^{n}}\rightarrow 1$$

if $n\rightarrow\infty$. So it cannot satisfy the weak law of large numbers.

If they are dependent, I wonder whether there is an example that WLLN holds--will edit later if I find one or disprove it.