Doob martingale


A Doob martingale
is a mathematical construction of a stochastic process which approximates a given random variable and has the martingale property with respect to the given filtration. It may be thought of as the evolving sequence of best approximations to the random variable based on information accumulated up to a certain time.
When analyzing sums, random walks, or other additive functions of independent random variables, one can often apply the central limit theorem, law of large numbers, Chernoff's inequality, Chebyshev's inequality or similar tools. When analyzing similar objects where the differences are not independent, the main tools are martingales and Azuma's inequality.

Definition

Let be any random variable with. Suppose is a filtration, i.e. when. Define
then is a martingale
, namely Doob martingale, with respect to filtration.
To see this, note that
In particular, for any sequence of random variables on probability space and function such that, one could choose
and filtration such that
i.e. -algebra generated by. Then, by definition of Doob martingale, process where
forms a Doob martingale. Note that. This martingale can be used to prove McDiarmid's inequality.

McDiarmid's inequality

Statement

Consider independent random variables on probability space where for all and a mapping. Assume there exist constant such that for all,
Then, for any,
and

Proof

Pick any such that the value of is bounded, then, for any, by triangle inequality,
thus is bounded.
Define for all and. Note that. Since is bounded, by the definition of Doob martingale, forms a martingale. Now define
Note that and are both -measurable. In addition,
where the third equality holds due to the independence of. Then, applying the general form of Azuma's inequality to, we have
One-sided bound from the other direction is obtained by applying Azuma's inequality to and two-sided bound follows from union bound.