Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

While reading lecture notes on Stein's method, there's one example that I cannot prove myself.

For iid random variables $X_1, ..., X_n$ where $\Pr(X_i=1) = \Pr(X_i=-1) = 1/2$, define $S_n=\frac{1}{\sqrt{n}}\sum_{i=1}^n X_i$. Then $S_n$ converges to normal distribution due to Central Limit Theorem. Still, $d_{TV}(S_n,Z)=1$ for all $n$ where $d_{TV}$ is the total variation distance and $Z\sim N(0,1)$.

It's an example of 'total variation distance is often too strong to be useful', but I don't know how to prove it. Any hint or suggestion?

share|cite|improve this question
HINT Try to show the more general fact that the total variation distance between a continuous distribution and a discrete distribution is $1$. – Srivatsan Dec 19 '11 at 18:39

What's the probability that $S_n$ is an integer multiple of $1/\sqrt{n}$?

share|cite|improve this answer
Let $\mathcal{E}$ be the event that $S_n$ is an integer multiple of $1/\sqrt{n}$. This event corresponds to finitely many points, and finitely many points in a continuous distribution amount to zero. I guess this is what you meant, right? – Federico Magallanez Dec 19 '11 at 22:39

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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