Journal archive
IT management |
|
Performance management |
|
|
|
One of the main activity of a commercial bank is the issuance of loans. Loan applications to the
bank can be received very much. They are processed in special departments and evaluated for the ability
of the borrower to fulfill its obligations. Since the bank allocates a limited amount of cash for issuing
loans and sets the framework for allowable losses from the estimated income, a certain number
of applications must be rejected. The question is how to create a list of approved applications in
order to get as much revenue as possible and meet the above limitations. This paper deals with this
problem. Since it has a higher dimension, it is proposed to use a modified genetic algorithm to solve
it. The modifications were made to the initial population formation operator and the new generation
operator. The initial population is not generated randomly, but by a certain algorithm within acceptable
limits. And in the formation of a new generation, it is replenished by individuals from the array
of the best solutions in the event that the algorithm stagnates. Software was developed using this algorithm
to solve the problem. The operation of the algorithm was investigated and it was revealed at
what values of the parameters of the algorithm, it shows better results.
|
|---|---|
Software engineering |
|
Algorithmic efficiency |
|
|
The article continues learning of database hierarchical bitmap indices simulation model, that was
earlier built by author. Target of research is to find optimal time unit size of the second level index
for arbitrary distributions of new database records flow, query time length interval and granularity of
query interval edges. The optimized value is average number of OR/XOR bitstring operations to satisfy
range query. The usual way is to run model many times with different values of second level index
size in accordance with chosen optimization method. Article proposes heuristic algorithm which
allows to get very good initial approximation to optimal value with only one model run and in practice
it quite can be assigned as the final solution. This result is obtained by some facts from renewal
theory but mainly by consideration of special integral function from random model input data and replacement
model output by values of this function. Mathematical properties of this function, that help
to do such trick, are formulated and proved by means of simple number theory suggestions. It is noted,
that this approach is similar to one, in which well known slice indexes are introduced . Proposed algorithm
is confirmed by some numerical results for some kinds of input data distributions. Graphical
illustrations of results are also presented.
|
|
The article is devoted to algorithms for obtaining outlines of two-dimensional and three-dimensional
orthogonal polyhedrons. An orthogonal polyhedron is a geometric figure representing the union of
non-overlapping orthogonal objects (rectangles or parallelepipeds in the two-dimensional or three-dimensional
case, respectively) with a fixed position relative to each other, considered as a single whole
object. The need to use orthogonal polyhedrons as individual objects arises, in particular, at solving a
number of resource allocation problems. When an orthogonal polyhedron is being drawn, there should
be displayed only those edge fragments of its orthogonal objects that compose the outline of the orthogonal
polyhedron. The article contains a detailed description of the developed algorithms to obtain
a set of edges belonging to the outline of a two-dimensional or three-dimensional orthogonal polyhedron.
The algorithms are based on the idea of searching and cutting off segments located on edges belonging
to several orthogonal objects belonging to the considered orthogonal polyhedron. The developed
algorithms provide the possibility of visualization of arbitrary orthogonal polyhedrons, including
those containing any holes. The algorithms presented in the article are implemented in the developed
applied software intended to optimize the solution of resource allocation problems of various dimension,
including the orthogonal packing and rectangular cutting problems.
|
|
The article is devoted to the description of the developed method and algorithm for finding the intersection
of a ray with a triangle based on kd-trees. The algorithm allows to reduce the number of
viewed items when searching, to speed up the process of searching for the intersection of the beam
with a certain area. Here are considered the advantages and disadvantages of kd-trees, the most popular
types of trees, their advantages and disadvantages. The author introduced the necessary abstractions
for optimizing structures that allow accelerating ray tracing, and describes the implementation of
the ray tracing algorithm based on kd-trees. The quality assessment of the constructed tree was made.
An example is given of determining the best location of the section plane for a group of elements limited
to one volume only. The algorithm for counting the number of elements to the left and right of
the partition plane is described. Analyzed aspects of the implementation of the algorithm in terms of
assessing its complexity. An algorithm for traversing a kd-tree, based on recursion, has been developed.
These algorithms can significantly accelerate the process of modeling the ray tracing in the optical
system of a light device for further analysis of its lighting characteristics.
|
|
The Aivika simulation framework is represented in this article, where a distributed module implementing
the Time Warp method is considered in detail. It is shown that main discrete event simulation
paradigms including the process-oriented one can be used in the distributed simulation. Also it is
represented how a GPSS-like embedded domain-specific language can be used in simulation. The important
details of the implementation are des
|
|
The growing complexity of information systems raises popularity of techniques and methods of
information security that are aimed at overcoming information asymmetry between an attacker and
defense. Methods for protecting information systems from reconnaissance by an external intruder are
among the areas in which such solutions are developed. Reconnaissance protection can be based on
misinformation techniques or continuous modification of the system. Thus, information about the key
parameters obtained by an adversary becomes obsolete either in the next instant of time or when applied
to another software copy. Evaluation methods to be applied to the results of implementing the
protection tools in terms of evaluating enhancement of the system’s security level in general are a matter
of difficulty. The present paper presents a formalized procedure for evaluation of Reconnaissance
Protection Techniques (RPT) based on mathematical models of Cyber Epidemic Dynamics. Additional
terms and definitions, such as «network cohesion components», «border nodes», and «radially independent
networks» were introduced into the model. It enabled calculating the relative level of the system’s
security from being compromised based on the security parameters of the system’s border elements
and graph connectivity components. The presented model enables defining the overall system
security level with consideration to graph topology and pre-calculated security parameters of elements
and subsystems. Consequently, we created a formalized evaluation method, which enables deriving
relative security indices at the stage of designing architecture of information systems.
|
|
|
The article discusses some new results obtained in the problem solving process of developing new
tools for assessing the viability of large innovative multiprojects in the metallurgical industry for the
new metal products issue. Accordingly, methods have been created for predicting the performance indicators
of enterprises participat-ing in such multiprojects. The results obtained are important to development
of economic-mathematical and instrumental methods for analyzing the effectiveness of
in-novation in the dynamic environment. A model has been created for the implementation of a large
innovative multiproject as part of the new metal production creation with the ability to carry out this
project stages by several participants. An assessment was made of the various components of the metallurgical
enterprises innovation potential and their external environment using dynamic growing pyramidal
networks and Kalman filtering. A computer actor-simulation model for implementation of
the organizational stage of production in the metallurgical complex has been developed. An organizational-
computational procedure and an information system for decision-making support to managing
a large innovative multiproject in the metallurgy and approving technical tasks for its subprojects. An
organizational-computational procedure and an information system for decision-making support to
managing a large innovative multiproject in the metallurgy and approving technical tasks for its subprojects.
The created economic and mathematical tools were used to develop recommendations for
the implementation of an innovative project carried out at the stage of semi-finished production for
future metal-made products.
|
|
|
This article discusses the server indexing system based on the FTP protocol. The article describes
a set of functions for indexing files, search engine and database. The authors have shown the possibility
of FTP server indexing, possible problems and the ways of solving them. Despite the evolution of
file storages through the 34 years passed after the creation of FTP protocol this issue remains relevant
and useful because of big amount of data, stored on FTP servers and wide functionality of developed
search system. The search system developed by the project consists of three main parts: indexing system,
database and Web-service. Indexing system is the most important and difficult part of the project,
it indexes various types of servers, which are fundamentally different from each other in the format of
data on files, and avoids link traps and upload indexed data to database. To ensure multi-threaded operation
and efficiency, the indexing system is implemented with Python. Web-service accepts requests
from the end user, makes a request to the database and returns the search result. The user can create a
query specifying the full or partial file name, file type and size. The unique difference of this system is
the storage of the change history of files and servers that can be tracked for each file, directory or server.
|
|
In this paper, using the Berlekamp-Massey algorithm, the linear complexity of the signals generated
by communication systems was estimated, the modeling of which was carried out in [5]. In addition,
the linear complexity of the communication system with simple signals represented by a sine wave
was evaluated [13]. As a result of the calculations, it was found that of the considered communication
systems, the most linear complexity is possessed by communication systems with chaotic signals (with
the exception of a communication system based on chaotic masking, in which the Rossler attractor is
used as a generator of chaotic signals). Among the considered systems of communication with chaotic
signals the most linear complexity has a communication system based on chaotic modulation. The least
linear complexity has a communication system based on simple signals. Communication systems based
on "classical" noise-like signals have a linear complexity much greater than the communication system
with simple signals, but less than the communication system with chaotic signals. It is established that
communication systems with chaotic signals in general are more preferable for secure communication
systems than communication systems based on "classical" noise-like signals. In general, it should be
noted that in order to increase the security of transmitted information using communication systems
based on noise-like signals, it is necessary either to complicate the structure of the transmitted signal,
making it looks like white noise, or to use a potentially infinite number of sets of different classes
of expanding sequences and their periodic change, subject to certain requirements for expanding
sequences.
|
|
Management of distributed systems that cannot be covered visually is usually done through the
formation of a virtual model of such a system. A model that provides both the development of control
actions and the management of their implementation. The analysis shows that in the pre-informational
epoch such models were formed in a mental form, based on the logical and analytical reasoning
of managers, according to the data of the gathering environment. Subsequently, the automation tools
provided for the collection of information, its processing and presentation, simplifying the formation
of a system model, but ensured this task to the end, without bringing it to a logical conclusion. The
solution of the problem has come closer with the advent of “virtual reality” technologies, which allow
us to form models of the required detail and efficiency in an almost automatic mode. Taking into
account the needs of the management process and the possibilities of virtualization technologies, the
article formulates proposals for describing and using the managed system model in decision support
systems. Two main areas of application of virtualization are proposed and algorithmically described:
when planning actions and managing the implementation of plans, as well as when training personnel.
|
