|
|
 Выпуск 79
-
- Agasandyan G. On peculiarities of families of risk-preference functions for CC-VaR
The work investigates theoretical and qualitative properties of parametric families of risk-preference functions (r. p. f.) of an investor, who upholds the continuous VaR-criterion (CC-VaR). The investor in problems with such a criterion selects not family but only one function. Nevertheless, the knowledge of families' properties has to help the investor to better formalize risk preferences. The conception of families' correctness that is connected with their yield and important for applying CC-VaR is introduced. One-parametric families are correct, if their yields are monotone functions of the parameter at arbitrary possible answer of the market. The families' analysis is prosecuted on base of normalized r. p. f., for which the integral in its domain is independent of parameter. The theorem about necessary and sufficient condition of families' correctness with some useful consequences is formulated and proved. An example of two-parametric superfamily of linear r. p. f. with one fracture that generates one-parametric correct families with special property of symmetry is considered. The example substantiates the hypothesis of quality type that more «profitable» r. p. f. as compared with a rival one generates lower incomes near zero and higher incomes near one. Analytical investigations are accompanied by computations and diagrams.
- Agieva M., Babicheva J., Okulist N., Ougolnitsky G. Analysis and forecasting problems in the control of target audience in marketing
Some networks analysis and forecasting models are considered in the paper relative to marketing. A brief survey of the control and influence models on social networks is presented. Namely, the problems include determination of the strong subgroups and satellites, calculation of quantitative characteristics of the network, determination of the final opinions of the members of the target audience based on their initial opinions and mutual interactions. For the solution of the forecasting problem an original algorithm is developed and implemented by means of the programming language R. The algorithm is described in details and considers a specific case of a strongly connected digraph and the general case. The complexity of the algorithm is estimated. Other named problems can be solved by embedded functions of the language that is illustrated by test examples together with the results of the authors' algorithm. The model examples use influence digraphs with 10-15 and more than 100 vertices. Such closeness measures as degree centrality, closeness centrality, and betweenness centrality are calculated for the model example, their interpretation is given. Solution of the analysis and forecasting problems essentially facilitates the problem of control of the opinions of target audience in marketing. The directions of future research in building and analysis of the models of opinion control in marketing are outlined.
- Akinfiev V. Two approaches to dynamic capacity expansion problem in oligopoly
The productive capacity is determinant of a company’s success once it allows meeting the current and future demand. This article proposes quantitative models for determining investments in the development of the productive capacity in competitive oligopolistic markets, based on the Nash Equilibrium, formulated as an optimization problem. Two types of markets are considered: the market with elastic demand (Cournot market) and the market with inelastic demand. The second type of market is characteristic of commodity markets (oil, gas, metallurgical coal, etc.), which are located at the beginning of the production chain for the final product creation. In the first case, the problem is formulated as a set of interrelated quadratic optimization problems. To solve this problem, a method of converting the original problem to the mixed complementarity problem (MCP) is proposed. In the second case, methods based on multi-agent simulation and matrix games analyses are proposed. A simulation model of the company (agent) has been developed, which models the mutual influence of agents investment and market dynamics. The proposed method allows obtaining an approximate solution of the original problem, since the heuristic principles and algorithms of the behavior of agents are used in the construction of the model.
- Shumov V. Hierarchy of models of military actions and border conflicts
The substantive aspects of wars, battles, military and border conflicts and the views on them of military theorists are discussed. Three levels of models are highlighted. At the lower level, a probabilistic conflict function based on the definition of a battle (a set of coordinated targets, place and time of strikes, fire and maneuver of troops) is used to simulate the combat of subunits, which takes into account the numbers of combat units of the parties and the parameter of combat superiority. His statistical estimate was found by the maximum likelihood method. Military superiority is moral superiority (taken into account through the percentages of bloody losses sustained by the parties) and technological (superiority in the organization of interaction, in maneuver, reconnaissance and fire capabilities). The game-theoretic problem of 26 resource allocation among the objects of defense has been solved. Models of the middle level are extensions of the classical Osipov – Lanchester fight equations that take into account the moral potentials of the parties using the basic law of psychophysics. The results of the calculations do not contradict the provisions of the military science of the combat capabilities of the troops. In the top-level models, the social costs of the course and outcome of wars. In particular, the loss of the United States in the Vietnam War can be explained by the lack of readiness of American society to bear casualties in a war with unclear objectives.
- Galayev A., Maslov E., Yahno V., Abramyants T. A moving object evasion from detection in the threat environment
Formalization and solution results of a number of a moving object evasion problems from detection by homogeneous or collage of observers on a plane are presented. The object is detected by signals of primary (passive mode) and secondary (active mode) field signals irradiated by it. Simulation results are also presented.
- Melnikov S. Distributed databases data architecture of continuous corporate learning process control system in the lcc gazprom transgaz nizhny novgorod
Minimization costs is one of the key tasks in the organization of the corporate training process in a large company, which can be partially solved by the introduction of specialized automated systems. Increasing the automation level here should be carried out very carefully, as excessive automation can make the system uncomfortable and not flexible enough. An information technology infrastructure of LCC Gazprom transgaz Nizhny Novgorod currently has a number of significant limitations and requirements for the information system that makes it impossible to use online access to the unified database and using two-phase transactions. This paper presents the development of an original synchronization mechanism of distributed databases data based on the method of asynchronous delivery of the transaction log. The proposed principle of construction of the system minimizes the work on setting up the database on remote nodes. The application problem of ensuring the integration of the local databases residing in some nodes of the computer network to a user running on any node in the network has been solved. The architecture of interaction at different levels is described.
|
|
