Electrical Engineering
10.17.50.243//handle/2074/7899
2022-05-28T23:58:57ZGeneration of vertex and edge cutsets
10.17.50.243//handle/12345678/2524
Generation of vertex and edge cutsets
Prasad, V C; Sankar, V; Rao, K S Prakasa
An algorithm is proposed to obtain basic minimal paths from basic minimal path tree for a network having a single source vertex and a single sink vertex. This does not require generation of all minimal paths. Using the paths thus obtained. A simple method is suggested to obtain all minimal vertex cutsets of any graph using OR and AND logic expressions. These paths can also be used to obtain all minimal edge cutsets which is advantageous for sparse graphs.
1992-01-01T00:00:00ZParallel techniques for solving large scale travelling salesperson problems
10.17.50.243//handle/12345678/2520
Parallel techniques for solving large scale travelling salesperson problems
Ravikumar, C P
As a hard combinatorial optimization problem, the travelling salesperson problem (TSP) has been of pedagogical interest for more than 50 years. More recently, the problem has generated a great deal of practical interest due to its applications in electronic circuit assembly and the drilling of printed circuit boards. In the simplest terms, the TSP is to find a minimum cost Hamiltonian tour of n cities. Since there is no known polynomial time algorithm to solve the TSP, and since n is quite large for practical problems, it is customary to use heuristic techniques and generate suboptimal tours. Even heuristic algorithms are expensive in CPU time when hundreds (or even thousands) of cities are involved. In this paper, we consider four well known heuristics for the TSP and their parallel implementations. Two constructive algorithms are considered: the farthest insertion heuristic and Christofides' approximation algorithm. Two iterative improvement algorithms are considered: the two-opt and three-opt techniques due to Lin and Kernighan. The results of applying parallel randomized search techniques to large instances of the problem are described. We demonstrate the usefulness of parallel processing in solving hard optimization problems by providing experimental evidence for both speedup improvement and an improvement in the quality of the final solutions. The target machines used for these parallel implementations are the Intel iPSC/2 hypercube and the Alliant FX/80.
1992-01-01T00:00:00ZProbabilistic analysis of subsynchronous resonance in series-compensated EHV transmission lines
10.17.50.243//handle/12345678/2496
Probabilistic analysis of subsynchronous resonance in series-compensated EHV transmission lines
Indulkar, C S; Viswanathan, B; Ramalingam, K
The stability limit curves in the R-Xc parameteric plane have been determined for EHV lines with series-compensation raking into consideration the nondeterministic nature of the generating system, mechanical system, and transmission system parameters. Studies have shown that the effect of random system parameters is to increase the area of the unstable zone in the parametric plane. First, the border line between the stable and unstable zones has been determined considering a deterministic model of the system. The value of the critical frequency at which the self-excited oscillations including the effect of hunting are set up and the system critical resistance necessary to damp out the oscillations are obtained using a modified D-partition technique. The modification involves the determination of the critical frequency by the Newton method. The nature of the random system parameters is then considered using the results of the deterministic study to calculate the variances of the critical frequency and resistance, and the stability limit curves.
1991-01-01T00:00:00ZInterval partition with bounded overlap
10.17.50.243//handle/12345678/2487
Interval partition with bounded overlap
Ravikumar, C P
The paper considers an optimization technique with applications to some resource-allocation problems that arise in high-level synthesis of digital systems. In an abstract sense, the optimization problem is that of the partitioning of a set of line intervals into a minimum number of subsets such that intervals assigned to a subset satisfy certain properties. It is shown that the problem can be solved optically in O(n) time, where n is the number of line intervals. Applications of the problem to memory allocation and task scheduling are discussed. Specifically, a program called is described for the allocation of variables to multiport memories. has been implemented in on a Sun/3 workstation.
1992-01-01T00:00:00Z