TY - JOUR AU - A. I. Sbitnev AU - V. V. Kozlov PY - 2017/12/20 Y2 - 2024/03/28 TI - Search for a Global Extremum in the Problem of Allocation of Connected Resources JF - Scientific Bulletin of the National Academy of Statistics, Accounting and Audit JA - nasoa VL - 0 IS - 4 SE - Articles DO - 10.31767/nasoa.4.2017.11 UR - https://nasoa-journal.com.ua/index.php/journal/article/view/60 AB - Methods for solving the problems o f resource allocation occupy an important place among the mathematical methods for solving special problems in making decisions for the preparation and conduct o f the operation. With the emergence o f a new kind o f resource - information, the characteristic feature o f which is internal interconnection, distribution problems acquire a new meaning. The purpose o f the study is to offer an extension o f the algorithm for searching a local extremum to achieve a global extremum in solving the distribution problems o f connected resources. To achieve global extremum, we use the well-known approach: 1. Generate some initial distribution; 2. We improve it by searching for the local extremum using a special method. 3. After the n-fold execution o f the first two steps, the obtained values are compared and the best one is selected and accepted as the value of the global extremum. The essence o f the approach proposed in the work is to use as the initial breakdown of «natural» entities on a graph model, which is used in the physical interpretations o f the solvable problems o f the distribution o f connected resources. The description o f the graph proposed in the work gives the possibility to obtain the natural initial conditions for organizing the search for a global extremum in the problem o f the distribution o f connected resources. ER -