«INFORMATICS AND ITS APPLICATIONS»
Scientific journal
Volume 7, Issue 1, 2013

Content | Bibliography | About  Authors

Abstract and Keywords.

ANALYTICAL MODELING OF INVARIANT MEASURE DISTRIBUTIONS IN STOCHASTIC SYSTEMS WITH DISCONTINUOUS CHARACTERISTICS .

• I.N. Sinitsyn   IPI RAN, sinitsin@dol.ru

Abstract: Based on normal approximation and statistical linearization, exact and approximate algorithms for distibutions with invariant measure analytical modeling in nongaussian stochastic systems (StS) with discontinuous characteristics are developed. Peculiarities in Poisson StS are considered. Test examples confirm the practical accuracy of algorithms.

Keywords:  analytical modeling; autocorrelated noise; distribution with invariant measure; method of normal approximation; method of statistical linearization; Poisson stochastic system; Pugachev integrodifferential equations, stochastic regime; stochastic system in Ito sense

PROBABILITY AND STATISTICAL MODELING OF INFORMATION FLOWS IN COMPLEX FINANCIAL SYSTEMS BASED ON HIGH-FREQUENCY DATA.

• V.Yu. Korolev   Faculty of Computational Mathematics and Cybernetics, M. V. LomonosovMoscow State University; IPI RAN, vkorolev@cs.msu.su
• A. V. Chertok   Faculty of Computational Mathematics and Cybernetics, M. V. LomonosovMoscow State University; Euphoria Group LLC, a.v.chertok@gmail.com
• A. Yu. Korchagin   Faculty of Computational Mathematics and Cybernetics, M. V. LomonosovMoscow State University, sasha.korchagin@gmail.com
• A.K. Gorshenin   IPI RAN, a.k.gorshenin@gmail.com

Abstract: A microstructure model is proposed for information flows in complex financial systems and stochastic nature of intensities of flows of claims which determines market price formation. The outer information flow with random intensity is considered separately within the proposed and statistically verified multiplicative model. This model makes it possible to analyze characteristics related to the intensities of the flows of claims and instant relation between the forces of buyers and sellers withoutmodeling of exogenous information background which is practically impossible to predict. Also, the generalized price processmodel is proposed whichmakes account of all the available information on flows of claims and admits further analytical interpretation.

Keywords:  financial markets; information flows; market price formation; intensity of the flow of claims; limit order book; mixture of probability distributions; generalized price

STATIONARY CHARACTERISTICS OF THE QUEUEING SYSTEM WITH LIFO SERVICE, PROBABILISTIC PRIORITY, AND HYSTERIC POLICY.

• T.A.Milovanova   People’s Friendship University of Russia, tmilovanova77@mail.ru
• A. V. Pechinkin   IPI RAN, apechinkin@ipiran.ru

Abstract: Consideration is given to single server queueing system with LIFO service discipline, probabilistic priority, and single-threshold hysteric policy. Expressions for main stationary characteristics, including stationary waiting and sojourn times distributions in terms of Laplace–Stieltjes transform, are obtained.

Keywords:  queueing system; last-in-first-out; probabilistic priority; hysteretic policy

ON CONVERGENCE IN THE SPACE Lp OF THE WORKLOAD MAXIMUM FOR A CLASS OF GAUSSIAN QUEUEING SYSTEMS.

• O. V. Lukashenko   Institute of AppliedMathematical Research of Karelian Research Center, Russian Academy of Sciences; Petrozavodsk State University, lukashenko-oleg@mail.ru
• E. V.Morozov   Institute of AppliedMathematical Research of Karelian Research Center, Russian Academy of Sciences; Petrozavodsk State University, emorozov@karelia.ru

Abstract: A class of queueing systems fed by an input containing linear deterministic component and a random component described by a centered Gaussian process is considered. The variance of the input is a regularly varying at infinity function with exponent 0 < V < 2. The conditions are found under which the maximum of stationary workload (remaining work) over time interval [0, t] converges in the space Lp as t -> (and under an appropriate scaling) to an explicitly given constant a. Asymptotics of the workload maximum in nonstationary regime is also given. The asymptotics of the hitting time of an increasing value b by the workload process is obtained.

Keywords:  Gaussian queue; workload maximum; fractional Brownian motion; asymptotical analysis; regular varying

ALGORITHMS FOR INDUCTIVE GENERATION OF SUPERPOSITIONS FOR APPROXIMATION OF EXPERIMENTAL DATA.

• G. I. Rudoy   Moscow Institute of Physics and Technology, rudoy@forecsys.ru
• V.V. Strijov   Dorodnicyn Computing Centre of RAS, strijov@ccas.ru

Abstract: The paper presents an algorithm which inductively generates admissible nonlinear models. An algorithm to generate all admissible superpositions of given complexity in finite number of iterations is proposed. The proof of its correctness is stated. The proposed approach is illustrated by a computational experiment on synthetic data.

Keywords:  symbolic regression; nonlinear models; inductive generation; models complexity

STATISTICAL TECHNIQUES OF BANS DETERMINATION OF PROBABILITY MEASURES IN DISCRETE SPACES.

• A.A. Grusho   IPI RAN, grusho@yandex.ru
• N.A. Grusho   IPI RAN, info@itake.ru
• E. E. Timonina   IPI RAN, eltimon@yandex.ru

Abstract: The method of statistical bans determination of probability measures in the discrete spaces is suggested. Consistency of the constructed estimates is shown. The application of the received estimates for testing the statistical hypotheses in the discrete spaces is constructed. It is shown that estimates of bans can generate consistent sequences of criteria in some sense.

Keywords:  consistent sequences of criteria; bans of probability measures in the discrete spaces; consistency of estimates

OPERATIONS ON THE TREE REPRESENTATIONS OF PIECEWISE QUASI-AFFINE FUNCTIONS.

• S. A. Guda   Faculty ofMathematics,Mechanics and Computer Science, Southern Federal University, gudasergey@gmail.com

Abstract: The piecewise quasi-affine (PQAF) functions and PQAF-sets are considered. Tree representation of PQAFfunction and PQAF-set and its complexity are introduced. The algorithms of union, intersection, empty check, sum, image/preimage calculation, inversion, composition, and comparison are given. The algorithms are provided with complexity estimates of the resulting objects. A theorem about the type and complexity of lexicographical extrema in the PQAF-set, depending on the parameters, has been proved.

Keywords:  piecewise quasi-affine function; quasi-affine selection tree; lexicographical extremum; Z-polytope

METHODOLOGICAL BASIS FOR THE CREATION OF INFORMATION SYSTEMS FOR THE CALCULATION OF INDICATORS OF THEMATIC LINKAGES BETWEEN SCIENCE AND TECHNOLOGY .

• V.A.Minin   RFBR, minin@rfbr.ru
• I.M. Zatsman   IPI RAN, iz_ipi@a170.ipi.ac.ru
• M.G. Kruzhkov   IPI RAN, magnit75@yandex.ru
• T. P. Norekyan   IPI RAN, izzittami@gmail.com

Abstract: International experience of indicators’ calculation of thematic linkages between science and technology is analyzed. The purpose of the analysis is to develop creation principles of information systems for calculating indicators of linkages with an allowance for structure of inherited scientific and patent information resources, which have been historically formed in Russia. This type of information systems is new to the Russian scientific and technical sphere. Their creation is necessary for evaluation of R&D programs and decision-making at all stages of programactivities. In the paper, themethodology of indicators’ calculation of thematic linkages in the Russian scientific and technical sphere is outlined to develop creation principles of these information systems.

Keywords:  linkages between science and technology; classification of the scientific domains; international patent classification; classifying scientific documents

PARALLEL TEXTS ALIGNMENT STRATEGIES: THE SEMANTIC ASPECTS.

• E.B. Kozerenko   IPI RAN, kozerenko@mail.ru

Abstract: The paper deals with the problems of design and development of the linguistically motivated mechanisms for parallel texts alignment and grammatical (functional semantic) matches extraction for design of statistical language portraits to be further incorporated into the hybrid models of machine translation. By hybrid models are meant those employing both statistical and rule-based mechanisms within one framework for natural language processing. The presented approach consists in the use of the starting (“seed”) grammar to be further enriched by the matches discovered in parallel texts. The seed grammar is used featuring the cognitive and functional characteristics of phrase structures. The grammar formalism employed is the Cognitive Transfer Grammar based on the transfemes (bilingual phrase structures). The interpreters’ transformations existing in parallel texts are of primary importance to the research efforts.

Keywords:  alignment; parallel texts; syntax; semantics; phrase structures; hybrid models; machine translation

SEMANTIC VECTOR SPACES FOR DIFFERENT KNOWLEDGE DOMAINS.

• Yu. I.Morozova   IPI RAN, yulia-ipi@yandex.ru

Abstract: The paper focuses on the actual problems of studying semantics of linguistic units using the methods of corpus linguistics. It gives a review of distributional semantics which is a new area of linguistic research. The paper proposes an enhancement to the existing distributional models by switching from lexemes to word collocations. The paper describes the methodology used to build semantic vector spaces for different knowledge domains.

Keywords:  distributional semantics; vector spaces; meaningful word combinations; collocations

INFORMATION METHOD FOR ASSESSMENT SEMANTIC ADEQUACY OF TEXTS.

• L.A. Kuznetsov   Lipetsk State Technical University, kuznetsov.leonid48@gmail.com
• V.F. Kuznetsova  Lipetsk State Technical University, kuznetsov@stu.lipetsk.ru

Abstract: A problem of automated knowledge examination is considered based on intelligent comparison of student’s answers with reference knowledge chunks. In order to increase relevance of automatically drawn conclusions, the information theory ground is used to develop the original methodology of textual content closeness evaluation. As a part of this work, this approach is applied to evaluate how close students’ essays match reference textual blocks, relating to a subject.

Keywords:  semantic similarity texts; probabilistic model of text; information theory; entropy; transmitted information of texts; knowledge estimate automatization

VARIANCE-GENERALIZED-GAMMA-DISTRIBUTIONS AS LIMIT LAWS FOR RANDOM SUMS.

• L.M. Zaks   Department ofModeling andMathematical Statistics, Alpha-Bank, lily.zaks@gmail.com
• V.Yu. Korolev   Faculty of ComputationalMathematics and Cybernetics, M.V. Lomonosov Moscow State University; IPI RAN, vkorolev@cs.msu.su

Abstract: A general theoremis proved establishing necessary and sufficient conditions for the convergence of the distributions of sums of a random number of independent identically distributed random variables to variance-mean mixtures of normal laws. As a corollary, necessary and sufficient conditions for the convergence of the distributions of sums of a randomnumber of independent identically distributed randomvariables to variance-generalized-gammadistributions are obtained. For a special case of continuous-time random walks generated by compound doubly stochastic Poisson processes, convergence rate estimates are presented.

Keywords:  random sum; generalized hyperbolic distribution; generalized inverseGaussian distribution; generalized gamma-distribution; variance-generalized-gamma-distribution; mixture of probability distributions; identifiable mixtures; additively closed family; convergence rate estimate

ON BOUNDS FOR THE CONCENTRATION FUNCTIONS OF REGULAR STATISTICS CONSTRUCTED FROM SAMPLES WITH RANDOM SIZES.

• V. E. Bening   Faculty of ComputationalMathematics and Cybernetics, M.V. Lomonosov Moscow State University; IPI RAN, bening@cs.msu.su
• N.K. Galieva   Kazakhstan Branch,M. V. Lomonosov Moscow State University, nurgul u@mail.ru
• V. Yu. Korolev   Faculty of ComputationalMathematics and Cybernetics, M.V. Lomonosov Moscow State University; IPI RAN, vkorolev@cs.msu.su

Abstract: Two-sided bounds for the concentration functions of regular statistics constructed from samples with random sizes are constructed.

Keywords:  concentration function; random sum; asymptotically normal statistic; Student distribution; Laplace distribution

ON THE ABSOLUTE CONSTANTS IN THE BERRY–ESSEEN INEQUALITY AND ITS STRUCTURAL AND NONUNIFORM IMPROVEMENTS.

• I.G. Shevtsova   Department ofMathematical Statistics, Faculty ofComputationalMathematics andCybernetics,M.V. Lomonosov Moscow State University; IPI RAN, ishevtsova@cs.msu.su

Abstract: New improved upper bounds are presented for the absolute constants in the Berry–Esseen inequality and its structural and nonuniform improvements. In particular, it is shown that the absolute constant in the classical Berry–Esseen inequality does not exceed 0.5583 in general case and 0.4690 for the case of identically distributed summands. The corresponding bounds in the Nagaev–Bikelis inequality are 21.82 and 17.36.

Keywords:  central limit theorem; convergence rate estimate; normal approximation; Berry–Esseen inequality; Nagaev–Bikelis inequality; absolute constant