Wyniki wyszukiwania - Biblioteka Nauki

1

New Algorithm Permitting the Construction of an Effective Spanning Tree

100%

Belkacem S. N. , Abbasy M.

|

tom Vol. 48, No. 3

313--329

EN

In this paper, we have done a rapid and very simple algorithm that resolves the multiple objective combinatorial optimization problem. This, by determining a basic optimal solution, which is a strong spanning tree constructed, according to a well-chosen criterion. Consequently, our algorithm uses notions of Bellman’s algorithm to determine the best path of the network, and Ford Fulkerson’s algorithm to maximise the flow value. The Simplex Network Method that permits to reach the optimality conditions manipulates the two algorithms. In short, the interest of our work is the optimization of many criteria taking into account the strong spanning tree, which represents the central angular stone of the network. To illustrate that, we propose to optimize a bi-objective distribution problem.

2

A minimum spanning tree analysis of the Polish stock market

86%

Tomeczek A. F.

|

tom 44

EN

Aim/purpose - This article aims to explore the network topology of the stock market in Poland during the COVID-19 pandemic. Design/methodology/approach - Kruskal's algorithm was used to find the minimum spanning trees (MST) of three undirected correlation networks: MST1 (December 2019 - August 2021), MST2 (February 2020 - April 2020), and MST3 (June 2021 - August 2021). There were123 firms included in all three networks representing three key indexes (WIG20, mWIG40, and sWIG80). Findings - The comovements of stock prices varied between various periods of the pandemic. The most central firms in Poland were PEO, UNT, SPL, PKO, KGH, CCC, and PZU. WIG20 was the most influential stock index for all networks. During the turbulent period represented by MST2, many of Poland's largest companies have clustered around KGH at the center of the network. In contrast, MST3 is the least compact of the three networks and is characterized by the absence of a single strongly influential node. Research implications/limitations - Correlation networks are efficient at quantitatively describing the degree of interdependence of a stock. MST finding algorithms are a crucial method of analysis for correlation networks. However, a limitation of the study, inherent to undirected correlation networks, is the inability to determine the direction of influence that stocks have on each other. Originality/value/contribution - The results of the article contribute to the economic analysis of stock markets in several ways. First, it expands on Gałązka (2011) by including additional centralities and the dynamic aspect of changes in the topology during the COVID-19 pandemic. Second, it broadens the MST-based empirical research of stock markets by showing the emergence of the star topology during the period of high uncertainty in Poland. Third, it has practical applications for systemic risk assessment and portfolio diversification.

3

Understanding the Impact of COVID–19 on Global Financial Network Using Graph Based Algorithm: Minimum Spanning Tree Approach

86%

Hatipoğlu V. F.

|

tom Vol. 46, No. 1

111--123

EN

In this paper effects of COVID–19 pandemic on stock market network are analyzed by an application of operational research with a mathematical approach. For this purpose two minimum spanning trees for each time period namely before and during COVID–19 pandemic are constructed. Dynamic time warping algorithm is used to measure the similarity between each time series of the investigated stock markets. Then, clusters of investigated stock markets are constructed. Numerical values of the topology evaluation for each cluster and time period is computed.

4

Segmentacja obrazów obiektów w postaci skupisk na przykładzie obrazów kometowych

72%

Bal A.

Pomiary Automatyka Kontrola

|

2012

|

tom R. 58, nr 1

55-58

PL

W pracy przedstawiono propozycję metody segmentacji obiektów będących skupiskami, przykładem takich obiektów są tzw. komety będące wynikiem jednokomórkowej elektroforezy żelowej. Prezentacja nowej metody została poprzedzona przedstawieniem wyników segmentacji tych obrazów metodami standardowymi. Opracowana metoda działa dwuetapowo: etap 1. to segmentacja służąca wyznaczeniu fragmentów składowych obiektów, etap 2 wykorzystuje minimalne drzewo rozpinające do określenia zbioru fragmentów tworzących poszczególne obiekty.

EN

This paper deals with the problem of segmentation of aggregate objects i.e. objects which are formed by the set of unconnected elements smaller than the object. Images of such objects are very difficult to be segmented. An example of this type of objects are "comet" from Single Cell Gel Electrophoresis images (also called comet assay images). In comet assay images the comet region is formed by unconnected fragments of DNA (Fig. 1). Due to unsatisfying results of comet segmentation by stan-dard methods (Figs. 2and 3) a new, two-stage method for segmentation of such images has been developed. The first stage is image segmentation whose result is a set of comet elements ei representing DNA fragments. In the second stage the minimum spanning trees Tp are created - graph vertexes vi represent elements ei, while length dij of edge eij between vertexes vi and vj is equal to the minimum distance between pixels of elements ei and ei. Then for each connected tree Tp its convex hull defining the region of comet Kp (Fig. 4) is created. In case of defects appearing in comet images (Fig. 5) the incorrect region can be rejected e.g. by use of geometrical features describing regions.