this is for experts in probability and stats. There is a theorem, I have seen once:
Given a stationary analytic random process, one can show that from the values of a sample path in a finite interval one can estimate the variance of the process without error (i.e. estimation variance is zero).
Can anybody give me a reference where to find this result? I remember vaguely that it is called "theorem of B...", but the name is NOT Belyaev.
Thanks
Karl