Search

Byline: Yanchun Yang, Robert L. Bulfin In this paper we studies a 0-1 Knapsack Problem with Setup (KPS). One set of 0-1 variables represent a family setup and serve as an Upper Bound (UB) on another set of 0-1 variables representing production of a job in a family. We present a branch-and-bound algorithm to find an optimal solution to KPS. The algorithm uses a two-stage branching strategy and chooses the next candidate problem to explore in a non-traditional way. We verify the efficacy of the algorithm through computational testing. This is the first time that an exact algorithm given to KPS with 10,000 integer variables. Computational experiments show that this algorithm is especially effective when objective and constraint coefficients are uncorrelated.

Byline: T. Vasanthi, G. Arulmozhi This paper proposes a genetic model to study the reliability optimisation in a Linear Consecutively Connected System (LCCS) by allocating M statistically independent Multi-state Elements with different characteristics to the first N positions. The system is reliable if the first position is connected with the (N + 1)th position. The reliability of LCCS is obtained using Universal Generating Function technique. The optimal allocation is obtained by means of Genetic Algorithm. In our work, for the one-to-one allocation, the Permutation Crossover is used, which reduces the complexity of the problem, and also yield better result.

Byline: Reeti Sharma, Kamal Srivastava In this paper, we deal with a particular layout problem of graph: the Minimum Linear Arrangement Problem (MinLA). A new Evolutionary Algorithm (EA) is presented here with knowledge-based operator designed to search the solution space. The proposed technique is a hybrid approach, which incorporates Simulated Annealing (SA) into the selection process of EA. It also utilises the features of level structure, Depth First Search (DFS), Frontal Increase Minimisation (FIM) method and Spectral Sequencing (SS) of the graphs. The proposed technique produces results that are well comparable with the existing approaches known for their good results.