11.07.2015 Views

7. Probability and Statistics Soviet Essays - Sheynin, Oscar

7. Probability and Statistics Soviet Essays - Sheynin, Oscar

7. Probability and Statistics Soviet Essays - Sheynin, Oscar

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Let E be some linear space with elements x, <strong>and</strong> P(A) – the probability that element xbelongs to set A. We denote linear functionals of x by f. Then a characteristic function ofdistribution P(A) is a function of the functional(f) = E xe ifx P(dE x ).If (f) can be exp<strong>and</strong>ed into a Taylor series(f) = 1 + h 1 (f 1 ) + h 2 (f 1 ; f 2 ) + h 3 (f 1 ; f 2 ; f 3 ) + …,the multilinear forms h n (f 1 ; f 2 ; …; f n ) provide the moments of the distribution P(A). Until now,only a very small number of corollaries were deduced from the indicated definitions.However, they {definitions or corollaries?} promise to be interesting, in particular for anumber of physical problems.3c. The Fokker – Planck differential equation <strong>and</strong> its conjugate underpin the derivation ofa number of new limit theorems concerning sums of a large number of r<strong>and</strong>om variables.Several methods for passing from integral equations determining the laws of distribution offinite sums to the corresponding differential equations were offered. The most elegant, as itseems to me, is the Petrovsky method which is a modification of the Perron method of upper<strong>and</strong> lower functions. However, it was Bernstein who obtained the fullest results in somedirections by applying another method for passing to the Fokker –Planck equations.Another series of limit theorems can be connected with the equations of step-wisestochastic processes (§3a). Whereas the first series of these propositions generalizes theLaplace – Liapunov theorem, the second one does the same with respect to the Poisson limittheorem. The theorems of the second type were gotten by Khinchin <strong>and</strong> Bavli.Introductory Literature1. Khinchin, A. (1933), Asymptotische Gesetze der Wahrscheinlichkeits-rechnung. Berlin.Note 1. {Fisher’s Statistical Methods for Research Workers first appeared in 1925.}4. A.Ya. Khinchin. The Theory of <strong>Probability</strong> in Pre-Revolutionary Russia <strong>and</strong> in the<strong>Soviet</strong> UnionFront Nauki i Tekhniki, No. 7, 1937, pp. 36 – 46Foreword by TranslatorKhinchin described the findings of the Moscow school of the theory of probability <strong>and</strong>argued that <strong>Soviet</strong> mathematics was far more advanced in these years (about 1937) thanbefore 191<strong>7.</strong> And he attempted to show that this fact was due to the favorable atmospherecreated in the <strong>Soviet</strong> Union for scientists. Khinchin’s high st<strong>and</strong>ing is the only reason whysuch rubbish deserved to be translated. If proof is needed, see Kolman (1931) <strong>and</strong> <strong>Sheynin</strong>(1998). Kolman (a minor mathematician <strong>and</strong> a high-ranking Parteigenosse who much laterescaped to the West) named Vernadsky, Sergei Vavilov, Ramsin et al <strong>and</strong> mathematiciansSchmidt, Kagan <strong>and</strong> Egorov as the bad guys; Egorov, for example, was indeed declared asaboteur, exiled <strong>and</strong> died soon afterwards. The horrible situation existing then (<strong>and</strong> bothearlier <strong>and</strong> later) in the <strong>Soviet</strong> Union is now widely known so that either Khinchin was forcedto state nonsense or he had been completely blind. True, the Luzin case was suddenly

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!