4.2d. Bebutov [1; 2] as well as Krylov & Bogoliubov [1; 2; 3] investigated processes with discretetime under various restrictions superimposed on the topological or differential-geometric nature of thephase space <strong>and</strong> on the differentiability of the transition probabilities (at least concerning theexistence of their densities with respect to some measure).4.2e. Especially numerous studies are devoted to the case of continuous time with the phase spacebeing a differential manifold when (t) is continuous <strong>and</strong> the transition probabilities are adequatelydifferentiable. Kolmogorov showed that these probabilities obeyed here parabolic or hyperbolicpartial differential equations that had first appeared in Fokker’s <strong>and</strong> Planck’s works on the Brownianmotion. Kolmogorov [19], also see Yaglom [2], offered the general expressions for these equations.The investigations of the behavior of (t) are reduced here to various boundary value problems forthe appropriate partial differential equations. For examples of applying this method see Petrovsky [2]or Kolmogorov & Leontovich [1]. Bogoliubov & Krylov developed important applications of thistool in statistical physics, but their works are beyond the scope of our review.4.2f. For one-dimensional processes with independent increments Kolmogorov <strong>and</strong> Lévyascertained in all generality the analytical form of the transition probabilities. These are representedby means of infinitely divisible laws (§1). In a number of works Gnedenko <strong>and</strong> Khinchin studied moresubtle problems about processes with independent increments.The success of Kolmogorov <strong>and</strong> Lévy was assured by the applicability here of the method ofcharacteristic functions. However, their result is not inseparably connected with it, <strong>and</strong> thedecomposition of the process into a step-wise part <strong>and</strong> a continuous (for independent increments,inevitably a Gaussian) part, accomplished by these authors, can probably be generalized to aconsiderably more general case of Markov processes. It seems that under some very generalconditions the Smoluchowski equation can be replaced by a mixed integro-differential equation of thetype indicated at the end of Kolmogorov’s contribution [32] on the general theory of Markovprocesses.4.3. Stationary Processes. For Markov processes, the transition probabilitiesP(t 1 ; t 2 ; ; ) at a given state <strong>and</strong> moment t 1 play the main role. Generally speaking, the veryexistence of unconditional probabilities for some course of the process is not here assumed. We shallnow adopt another viewpoint: The distribution of probabilities in the space of functions (t) is givenwith t taking all real values <strong>and</strong> the values of belonging to the phase space .The dem<strong>and</strong> of time-homogeneity is here transformed into a dem<strong>and</strong> for stationarity: theprobability P[(t) ∈ A] that the function (t), describing the course of the process in time, belongs tosome set A does not change if A is replaced by set H A appearing after each (t) from A is replacedby (t – ).The study of stationary processes with phase space is tantamount to investigating dynamicsystems with invariant measure in space R of functions (t) of a real variable with values from .From among the general results the first place is occupied here by the celebrated Birkhoff – Khinchinergodic theorem (Khinchin [25; 27; 52]).4.4. The Correlation Theory of Processes Stationary in the Broad Sense. The stationarity of theprocess of the variation of (t) leads, for any functions f() <strong>and</strong>g() having finite expectations E{f[(t)} 2 <strong>and</strong> E{g[(t)} 2 , to the expectationsE{f [(t) g[(t + )} = B fg ( )being functions only of , <strong>and</strong> to the constancy of the expectations E{f[(t)} = A f ,E{g[(t)} = A g .When only turning our attention to the results that can be expressed through these first <strong>and</strong> secondmoments for a finite number of functions f(), g(), …, it is natural to pass on to the followingconcept of processes stationary in the broad sense: The phase space is an n-dimensional space ofvectors x = (x 1 , x 2 , …, x n ); the r<strong>and</strong>om process is given by the distribution of probabilities in the spaceof vector-functions x(t) = [x 1 (t), x 2 (t), …, x n (t)], that is, by the joint law of distribution of n r<strong>and</strong>om
functions 1 (t), 2 (t), …, n (t), <strong>and</strong> stationarity is restricted by dem<strong>and</strong>ing that E k (t) = A k do notdepend on time <strong>and</strong>E[ i (t) j (t + )] = B ij ( )only depend on .Without loss of generality it is possible to assume that A k = 0 <strong>and</strong> B kk = 1, k = 1, 2, …, n. Then B ij ( )is nothing but the correlation coefficient between i (t) <strong>and</strong> j (t + ). Khinchin [32] outlined thegeneral patterns emerging from these propositions of the correlation theory of r<strong>and</strong>om processes.Kolmogorov [47] reviewed its further development in which Slutsky, he himself, V.N. Zasukhin <strong>and</strong>M.G. Krein had participated.4.5. Other Types of R<strong>and</strong>om Processes. A number of studies is devoted to various special typesof r<strong>and</strong>om processes that do not belong to the patterns described above. These include the so-calledgeneralized chains introduced already by Markov which can be reduced (as the number of statesincreases) to usual simple Markov chains. Stochastic bridges 6 introduced by Bernstein [21], whichgeneralize the notion of Markov process, would have deserved a more detailed investigation. It isnatural that the study of stationary processes leads to processes with stationary increments.Bogoliubov carried out important investigations of the limiting transitions from some types ofr<strong>and</strong>om processes to other ones (for example, he studied the conditions for a limiting transition of anon-Markov process depending on a parameter into a Markov process).5. A New Approach to Limiting Theorems for Sums of a Large Number ofTermsIt is well known from the history of the classical analysis that a limiting transition from relationsbetween finite differences to the corresponding differential expressions often leads to considerablysimpler results as compared with those accomplished by a direct investigation of the initial relations.In a similar way, most asymptotic formulas derived by long reasoning in the classical studies of sumsof a large number of r<strong>and</strong>om terms, or of results of a large number of trials, appear as precisesolutions of naturally <strong>and</strong> simply formulated problems in the theory of r<strong>and</strong>om processes withcontinuous time. This idea that goes back to Poincaré <strong>and</strong> Bachelier is convincingly developed bymeans of simplest illustrations, see for example Khinchin [33]. It offers a leading principle forconstructing new proofs <strong>and</strong> for discovering new formulations of the limit theorems of the classicaltype. Some foreign authors from Courant’s school (Pólya, Lüneberg) were first to apply such amethod <strong>and</strong> Kolmogorov [32] provided its general outline. In his abovementioned works [13; 18] healso applied it for proving the classical Liapunov limit theorem <strong>and</strong> limit theorems of a new type forsums of independent terms.The method was further developed according to two methodologically different viewpoints.Bernstein systematically develops the method of stochastic differential equations but does not connect{it with} the limiting differential equations of the limiting probability-theoretic pattern of a r<strong>and</strong>omprocess with continuous time. On the contrary, Petrovsky <strong>and</strong> Khinchin always bear in mind thislimiting r<strong>and</strong>om process.As to actual results, Bernstein [22; 27; 31; 41] examined considerably more fully the difficultiesthat occur in cases of a non-compact phase space (such, for example, as even the usual real straightline with which he indeed has to do), whereas Petrovsky [1] <strong>and</strong> Khinchin [33] more fully studiedproblems with boundaries.We note that the totality of the available theoretically rigorous contributions of this type does not atall cover the cases in which similar limit transitions are made use of in applications. Further intensivework in this direction is therefore desirable.The abovementioned works of the three authors are mostly concerned with limit theoremsconnected with continuous r<strong>and</strong>om processes governed by the Fokker – Planck equations. OnlyKhinchin [33] generalizes the Poisson limit theorem for step-wise processes with continuous time.This theorem was the foundation of all the investigations of limit theorems for sums leading toinfinitely divisible laws <strong>and</strong> expounded in detail in §1. It should be noted in this connection that there
- Page 4 and 5:
[3] I bear in mind the well-known p
- Page 6 and 7:
successes of physical statistics. B
- Page 8 and 9:
classes of independent facts whose
- Page 10 and 11:
distribution is a corollary of the
- Page 12 and 13:
examine in the first place the curv
- Page 14 and 15: 12. According to Bortkiewicz’ ter
- Page 16 and 17: generality, the similarities taking
- Page 18 and 19: one on another, as well as the corr
- Page 20 and 21: is inapplicable because the right s
- Page 22 and 23: Instead, Slutsky introduced new not
- Page 24 and 25: abandoned in August 1936, but it is
- Page 26 and 27: last decades, mathematicians more o
- Page 28 and 29: charged with making the leading ple
- Page 30 and 31: motion and a number of others) are
- Page 32 and 33: phenomena. It is self-evident that
- Page 34 and 35: Such new demands were formulated in
- Page 36 and 37: The addition of independent random
- Page 38 and 39: automatic lathes, etc. Here, the ma
- Page 40 and 41: 11. Kolmogorov, A.N. Grundbegriffe
- Page 42 and 43: period 1 and remained, until the ap
- Page 44 and 45: of the analytical tool rather than
- Page 46 and 47: with probability approaching unity,
- Page 48 and 49: logic. The ensuing vagueness in his
- Page 50 and 51: 2. Gnedenko, B.V. (1949), On Lobach
- Page 52 and 53: will be sufficient, although not ne
- Page 54 and 55: nlimk = 1P(| k (n) - m k (n) | > H
- Page 56 and 57: favorite classical issue as the gam
- Page 58 and 59: and some quite definite (not depend
- Page 60 and 61: influenced by a construction that a
- Page 62 and 63: P ij (1) = p ij (1) , P ij (t) =kP
- Page 66 and 67: are yet no limit theorems correspon
- Page 68 and 69: described, from the viewpoint that
- Page 70 and 71: conditional variance and determined
- Page 72 and 73: Romanovsky [45] and Kolmogorov [46]
- Page 74 and 75: Let S be the general population wit
- Page 76 and 77: Part 1. Russian/Soviet AuthorsAmbar
- Page 78 and 79: 2. On necessary and sufficient cond
- Page 80 and 81: Gnedenko, B.V., Groshev, A.V. 1. On
- Page 82 and 83: 52. ( (Mathematical Principl
- Page 84 and 85: Kozuliaev, P.A. 1. Sur la répartit
- Page 86 and 87: Obukhov, A.M. 1. Normal correlation
- Page 88 and 89: 30. Généralisations d’un théor
- Page 90 and 91: 22. Alcune applicazioni dei coeffic
- Page 92 and 93: 10. A.N. Kolmogorov. The Theory of
- Page 94 and 95: Kuznetsov, Stratonovich & Tikhonov
- Page 96 and 97: In the homogeneous case H s t = H t
- Page 98 and 99: to such a generalization. He only s
- Page 100 and 101: In the particular case of a charact
- Page 102 and 103: as it is usual for the modern theor
- Page 104 and 105: 1. {The second reference to Pugache
- Page 106 and 107: Smirnov, Romanovsky and others made
- Page 108 and 109: determined the precise asymptotic c
- Page 110 and 111: for finite values of N, M and R 2 .
- Page 112 and 113: Mikhalevich’s findings by far exc
- Page 114 and 115:
Uch. Zap. = Uchenye ZapiskiUkr = Uk
- Page 116 and 117:
Khinchin, A.Ya. 43. Math. Ann. 101,
- Page 118 and 119:
7. DAN 115, 1957, 49 - 52.Pinsker,
- Page 120 and 121:
Anderson, T.W., Darling, D.A. (1952
- Page 122 and 123:
Statistical problems in radio engin
- Page 124 and 125:
observations for its power with reg
- Page 126 and 127:
securing against mistakes (A.N. Kry
- Page 128 and 129:
of the others, then its distributio
- Page 130 and 131:
In Kiev, in the 1930s, N.M. Krylov