: ..,
..
:
: 118
:
: 2025
: .., .. // . - 2025. - . 118. - .325-361.
: , ,
(.): resilience of electrical networks, search for critical nodes, graph models of networks
: --- , . . . , , , . . , ( ) . , , , . , . , .
(.): One of the current research directions in network robustness is the class of critical node identification problems sets of nodes whose removal would cause the maximum damage to the network's functionality. To solve problems of this class, the network is modeled as an unweighted, undirected graph. Such models are used in studies of network connectivity, and a widely adopted metric for assessing damage is the number of connected pairs of vertices. A set of vertices is considered critical if, in the graph from which this set is removed, the number of connected vertex pairs is minimized among all sets of the same size. In addition to exhaustive search methods, this problem is often approached by reducing it to an equivalent linear programming problem. The current research examines a method for reducing the dimensionality of the linear programming problem by selecting two vertex characteristics such that several vertices with the highest values of the first characteristic are almost always critical (appear in most solutions), and several vertices with the lowest values of the second characteristic are almost never critical. This allows for the partial exclusion of variables corresponding to the identified critical and non-critical vertices from the linear programming problem, thereby reducing its size. To correctly formulate the problem, a number of preparatory sub-tasks, described in previous work, needed to be addressed. This paper is devoted to the second, final part of the study.
PDF
: 52, : 9, : 7.