Talk:Random field

From Wikipedia, the free encyclopedia

Xi in {0,1,...,G-1}? Shouldn't continuous variable and continuous fields also be allowed? fnielsen 11:21, 3 Mar 2005 (UTC)

[edit] hyperlinks missing

This article refers to Markov random field and Gibbs random field, neither of which exist in Wikipedia, however this article discussed MRF's and GRF's within itself. Shouldn't the other entries at least be stubbed-out with a short description? —Preceding unsigned comment added by 149.169.178.128 (talk • contribs)

[edit] warning

This entry is incorrect. If there is a domain where such definition is used it is certainly not probability theory. The reference quoted treats a very special (although important) case of random fields. In modern probability theory the notion of random field is almost synonymous with stochastic process, except that the term "random field" is more often used if the underlying index set is a subset of \Bbb{R}^n. Slawekk 16:54, 14 March 2006 (UTC)

[edit] Expert attention needed

I agree. In current literature a random field is usually (in order of increasing abstraction) 1. collection of random variables (the "pointwise values" on its domain, a subset of \Bbb{R}^n) 2. a random variable with values in a functional space over that domain 3. a probability measure on that functional space. These concepts are equivalent in a suitable sense. The article ought to be completely rewritten. Jmath666 00:05, 19 March 2007 (UTC)