Stochastic processes and statistical inference
Ulf Grenander · Arkiv för matematik · 1950
By ULF GRENANDER v. aRENAr~I)ER, Stochastic processes and statistical inference If /(o~) is a real function defined on ~9 and measurable with respect to P, / (@ is called a stochastic variable.The mean value operator E is defined as E 1 = f / d P (w) :2 if /((o) is integrable with respect to P. The modifications necessary in order to treat complex valued stochastic variables are evident.Let /1(co), ]2(o~),.../~(w) be n stochastic variables defined on ~.If it is true for every choice of n Borelsets El, E2 .... E~ on the real axis that P {/i(w) eEi; i = 1, 2 .... n}r ----~[P {/i (w) e Ei}(., i=1 the variables are said to be independent.Let A be an arbitrary measurable set and xl (@, x2 (09) .... x,~ (~o) n stochastic variables.If M is a "cylinderset" with an arbitrary measurable set in the n-dimensional Euclidean space Rn(x~, x2,.., xn) as basis, there exists one, and but for equivalence, only one function P(A Ix1, x2,.., x,) that satisfies P (A M) = f P (A I x.z2 .... x~) d P (x.x2 .... x.,) M for every M with the said properties (see KOL~OCOROrF 1, p. 42).P (A Ix1, x2 .... x,~)is called the conditional (with regard to xl, x2,.., x~) probability of A. The conditional expectation is defined in an analogous way (see KOLMOGOROFF 1, p. 47).It has been shown that the conditional probability has usually the same properties as the absolute probabilities.Let A be a fixed set.Then we have
Cited by 419 works. doi:10.1007/bf02590638 Open-access PDF
Authors
References
- Extrapolation, Interpolation, and Smoothing of Stationary Time Series
- Mathematical Methods of Statistics.
- Fourier Transforms in the Complex Domain
- Korrelationstheorie der station�ren stochastischen Prozesse
- Theory Of The Integral
- The Brownian Movement and Stochastic Equations
- Grundbegriffe der Wahrscheinlichkeitsrechnung
- The Elementary Gaussian Processes
- Stochastic Processes Depending on a Continuous Parameter
- On the Theory of Stationary Random Processes