Computational Science and Techniques

Šiame darbe sudarytas rekurentinis paslėptųjų Markovo modelių parametrų vertinimo algoritmas. Paslėptieji Markovo modeliai modeliuojami Gauso skirstiniu, kurio parametrai pasiskirstę pagal daugiamatį normalųjį dėsnį su nežinomais vidurkių vektoriumi ir kovariacijų matrica. Nežinomų parametrų įverčiai gaunami didžiausio tikėtinumo metodu. Rekurentinis algoritmas sudarytas remiantis didžiausio tikėtinumo metodu išvestomis formulėmis ir klasikiniu EM algoritmu. Kadangi rekurentinio algoritmo vykdymo laikas yra proporcingas apdorojamų stebėjimų skaičiui, tai jis gali būti naudojamas modelio parametrų vertinimui realiu laiku. Realizuoto rekurentinio EM algoritmo savybės buvo ištirtos kompiuteriniu eksperimentu klasterizuojant duomenis. Jis taip pat gali būti taikomas duomenų klasifikavimo ir atpažinimo realiu laiku uždaviniams spręsti.

This paper presents the protons and neutrons distributions in atomic nucleus shells calculation algorithm which may be used for ab initio no-core nuclear shell model computations. The problem of enumeration of many-particle states is formulated on energetic basis instead of application of the traditional scheme for states classification. The algorithm provides calculations of protons and neutrons occupation restrictions for nuclear shells for an arbitrary number of oscillator quanta. The reported results show that the presented algorithm significantly outperforms the traditional approach and may fit the needs of state-of-the-art no-core shell model calculations of atomic nuclei.

The article discusses the possibilities of Klaipeda region historic cemeteries destruction risk assessment using multiple criteria analytic hierarchy process (AHP). The proposing original assessment methodology developed by combining information from scientific literature on historical artefacts preservation topic with the data collected by scientists of Institute of Baltic Region History and Archaeology during their field expeditions to Klaipeda region Evangelical Lutheran Cemeteries.The results show that the process of historical cemeteries destruction risk assessment can be formalized and fully automated using AHP and modern software.

This paper addresses the issue of finding the most efficient estimator of the normal population mean when the population “Coefficient of Variation (C. V.)” is ‘Rather-Very-Large’ though unknown, using a small sample (sample-size ≤ 30). The paper proposes an “Efficient Iterative Estimation Algorithm exploiting sample “C. V.” for an efficient Normal Mean estimation”. The MSEs of the estimators per this strategy have very intricate algebraic expression depending on the unknown values of population parameters, and hence are not amenable to an analytical study determining the extent of gain in their relative efficiencies with respect to the Usual Unbiased Estimator X ̅(sample mean ~ Say ‘UUE’). Nevertheless, we examine these relative efficiencies of our estimators with respect to the Usual Unbiased Estimator, by means of an illustrative simulation empirical study. MATLAB 7.7.0.471 (R2008b) is used in programming this illustrative ‘Simulated Empirical Numerical Study’.

The paper presents the results on the dimensionality reduction technique which is based on radial basis function (RBF) theory. The technique uses RBF for mapping multidimensional data points into a low-dimensional space by interpolating the previously calculated position of so-called control points. This paper analyses various ways of selection of control points (regularized orthogonal least squares method, random and stratified selections). The experiments have been carried out with 8 real and artificial data sets. Positions of the control points in a low-dimensional space are found by principal component analysis. Combinations of RBF technique with random and stratified selections outperformed RBF with regularized orthogonal least squares algorithm regarding to computation time analysing all data sets. We demonstrate that random and stratified selections of control points are efficient and acceptable in terms of balance between projection error (stress) and time-consumption.

This paper proposes methodology for companies’ assessment. There is suggesting assessing the company’s prospect, not only according to share price, forecasts of the analysts, but also on the basis of position of each company in two-dimensional space in respect of the other companies. Seeking to describe the share prices of a company during the year, the parameters of skew t distribution are calculated. Then they are used in the inputs of random forest algorithm. Proximity matrices are stored during classification, and they are displayed in two-dimensional space. Thus, two clusters are obtained: one of the companies with upgrade trend, another one – with downgrade trend. This method may be useful those investors who are important to choose the most promising companies of all industry without wasting a lot of time.

The Lithuanian coast of the south eastern part of the Baltic Sea represents a generic type of more or less straight, high-energy (in the Baltic Sea conditions), actively developing coasts that contain a relatively large amount of finer, mobile sediments, are open to predominating wind directions and are exposed to wave activity for a wide range of wave approach directions. The combination of the angular distribution of winds and the geometry of the coast are such that the wave-induced long-shore sediment transport is, in average, to the north over the entire Curonian spit and the mainland coast of Lithuania. Analysis of the field data performed by the G. Žilinskas (2008) from 1976-2007 revealed that the length of accumulative sections has been considerably reduced. Accordingly, the length of the gradually eroding sectors has increased in the end of the 20th century. In this study were analised potential variations in the long-shore sediment transport rate due changes of the wind wave directions.

Physical and chemical parameters were measured in a mostly freshwater estuarine lagoon in the SE Baltic. Present paper demonstrates an attempt to trace the sources and analyse the seasonal and spatial patterns of distribution of POC, DIC and DOC in the Curonian lagoon mostly by the isotopic content in different forms of carbon. Samples were collected in 2012-2014 in 9 stations in the Curonian lagoon including riverine and marine input/output stations. Riverine inputs and summarizing outflow to the Baltic sea locations (Nemunas river delta and Klaipeda channel stations were sampled monthly, while POC, DIC and DOC samples in other stations were collected on a seasonal basis. The observed results allow easily differentiate between estuarine and riverine POM samples, while the differences in DOC δ13C content between sampling stations were found to be not statistically reliable. The high biological productivity of the Nemunas river along with the minor contribution of the Baltic Sea inflows to the overall hydrodynamics of the lagoon explain similarity of content between riverine and estuarine material in the spring and autumn. However, the δ13C content of DIC and DOC could serve as indicator of external inputs only in connection with seasonal water residence variations.

This paper proposes an algorithm for construction of C2 surface. The input is a 3D quadrangular surface which doesn't have adjacent extraordinary points. The difference between this algorithm and the regular C2 forming algorithms is a possibility to choose the natural number of points of the output grid. For example using the Catmull-Clark algorithm the same possibility corresponds exponential natural numbers where the basis is 2. The changing infinite Bezier surfaces generation process to finite calculation of necessary points leads the ability to get more results. The C2 surface generation algorithm was realized and the quality results of output surfaces were performed using reflection lines.

Lack of ICT specialists is a big problem in Lithuania for many years. Even ICT specialists are prepared both by universities and colleges, need of those specialists is increasing every year. Students’ enrolment to the ICT related study programmes is increasing a little every year, but only around 51% of them complete their studies. There are very different reasons of students drop-out: health problems, characteristics of personality (lack of responsibility, self-doubts, etc.), socio-economic factors, organisation of study process (Barkauskaitė & Gudžinskienė 2006). In this article a different approach is used to academic arrears – using statistical calculations, academic arrears of students of Vilnius University, Faculty of Mathematics and Informatics from 1991 till 2012 are analysed. At the end of the article, justification of a theoretical model is presented, using which students’ drop-out peculiarities in different study programmes can be observed.