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

Let $X_n, n\in\mathbb{Z}$ be a sequence of independent random variables (finite, let's say the size of the alphabet is $2$ to simplify things) with mean zero and variance less than $1$. Is there a large deviations type result for this case? i.e. Does $P\left[|\frac{X_1+\cdots+X_n}{n}|\ge t\right]$ fall exponentially?

The motivation is this: Let $(X_n,Y_n),n\in \mathbb{Z}$ be a sequence of iid random variables with joint distribution $p_{XY}=p_YW_{X|Y}$ for some transition kernel $W_{X|Y}$. Can we state a "Conditional Large Deviations" result for $X^n$ given the outcome $Y^n=y^n$ where $y^n$ is a likely outcome of $Y^n$ (Let's say the empirical distribution of $y^n$ is close to $P_Y$ with respect to some metric) Here $Y$ can take values from $\mathbb{R}$ but $X$ is finite.

share|cite|improve this question
up vote 1 down vote accepted

One can realize each $X_n$ as $\rho(Y_n,Z_n)$ for some i.i.d. sequence $(Z_n)_n$ independent of $(Y_n)_n$. Then the result holds (in a somewhat greater generality) for almost every realization of $(Y_n)_n$, see Stochastic Sub-Additivity Approach to the Conditional Large Deviation Principle by Zhiyi Chi.

share|cite|improve this answer
Thanks for your help again! – Jim Jan 9 '13 at 20:23

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.