- Doob martingale
-
A Doob martingale (also known as a Levy 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.
Contents
Definition
A Doob martingale (named after J. L. Doob)[citation needed] is a generic construction that is always a martingale. Specifically, consider any set of random variables
taking values in a set A for which we are interested in the function
and define:
where the above expectation is itself a random quantity since the expectation is only taken over
- Xi + 1,Xi + 2,...,Xn,
and
- X1,X2,...Xi
are treated as random variables. It is possible to show that Bi is always a martingale regardless of the properties of Xi. Thus if one can bound the differences
- | Bi + 1 − Bi | ,
one can apply Azuma's inequality and show that with high probability
is concentrated around its expected value
McDiarmid's inequality
One common way of bounding the differences and applying Azuma's inequality to a Doob martingale is called McDiarmid's inequality.[citation needed] Suppose
are independent and assume that f satisfies
(In other words, replacing the i-th coordinate xi by some other value changes the value of f by at most ci.)
It follows that
and therefore Azuma's inequality yields the following McDiarmid inequalities for any ε > 0:
and
and
See also
- Markov inequality
- Chebyshev's inequality
- Bernstein inequalities (probability theory)
References
- McDiarmid, Colin (1989). "On the Method of Bounded Differences". Surveys in Combinatorics 141: 148–188. http://www.stats.ox.ac.uk/__data/assets/pdf_file/0009/4113/bdd-diffs89.pdf.
This probability-related article is a stub. You can help Wikipedia by expanding it.