Doob martingale explained

In the mathematical theory of probability, a Doob martingale (named after Joseph L. Doob,^[1] also known as a Levy martingale) is a stochastic process that 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

Y be any random variable with E [|Y|] < inf ty . Suppose \left\{ l{F} _{0,} l{F} _{1,} ... \right\} is a filtration, i.e. l{F} _{s} \subset l{F} _{t} when s<t . DefineZ _{t} = E [Y \mid l{F} _{t],} then \left\{Z _{0,} Z _{1,} ... \right\} is a martingale,^[2] namely Doob martingale, with respect to filtration \left\{ l{F} _{0,} l{F} _{1,} ... \right\} .

To see this, note that

E [|Z _{t|]} = E [| E [Y \mid l{F} _{t]|]} \leq E [ E [|Y| \mid l{F} _{t]]} = E [|Y|] < inf ty ;E [Z _{t} \mid l{F} _{t-1} ] = E [ E [Y \mid l{F} _{t]} \mid l{F} _{t-1} ] = E [Y \mid l{F} _{t-1} ] = Z _{t-1} as l{F} _{t-1} \subset l{F} _{t} .

In particular, for any sequence of random variables

\left\{X _{1,} X _{2,} ... , X _{n\right\}} on probability space (\Omega, l{F}, P ) and function f such that E [ | f(X _{1,} X _{2,} ... , X _{n)} |] < inf ty , one could choose Y: = f(X _{1,} X _{2,} ... , X _{n)} and filtration \left\{ l{F} _{0,} l{F} _{1,} ... \right\} such that\begin{align} l{F} _{0} &: = \left\{\phi, \Omega \right\}, \\ l{F} _{t} &: = \sigma(X _{1,} X _{2,} ... , X _{t),} \forall t \geq 1 , \end{align}

i.e.

\sigma -algebra generated by X _{1,} X _{2,} ... , X _{t} . Then, by definition of Doob martingale, process \left\{Z _{0,} Z _{1,} ... \right\} where\begin{align} Z _{0} &: = E [f(X _{1,} X _{2,} ... , X _{n)} \mid l{F} _{0]} = E [f(X _{1,} X _{2,} ... , X _{n)],} \\ Z _{t} &: = E [f(X _{1,} X _{2,} ... , X _{n)} \mid l{F} _{t]} = E [f(X _{1,} X _{2,} ... , X _{n)} \mid X _{1,} X _{2,} ... , X _{t],} \forall t \geq 1 \end{align} forms a Doob martingale. Note that Z _{n} = f(X _{1,} X _{2,} ... , X _{n)} . This martingale can be used to prove McDiarmid's inequality.

McDiarmid's inequality

See main article: McDiarmid's inequality.

The Doob martingale was introduced by Joseph L. Doob in 1940 to establish concentration inequalities such as McDiarmid's inequality, which applies to functions that satisfy a bounded differences property (defined below) when they are evaluated on random independent function arguments.

A function

f: l{X} _{1} x l{X} _{2} x … x l{X} _{n} → R satisfies the bounded differences property if substituting the value of the i th coordinate x _{i} changes the value of f by at most c _{i} . More formally, if there are constants c _{1,} c _{2,} ... , c _{n} such that for all i\in[n] , and all x _{1\in} l{X} _{1,x} _{2\in} l{X} _{2,} \ldots, x _{n} \in l{X} _{n} ,\sup _{x _{i'} \in l{X} _{i}} \left|f(x _{1,} ... , x _{i-1} , x _{i,} x _{i+1} , \ldots, x _{n)} - f(x _{1,} ... , x _{i-1} , x _{i',} x _{i+1} , \ldots, x _{n)\right|} \leq c _{i.}

Notes and References

1. Doob . J. L. . 1940 . Regularity properties of certain families of chance variables . Transactions of the American Mathematical Society . 47 . 3 . 455–486 . 10.2307/1989964. 1989964 . free .
2. Book: Doob, J. L. . 1953 . Stochastic processes . New York . Wiley . 101 . 293 .