Bibliographic citations
Tupia, M., (2005). Un Algoritmo GRASP para resolver el problema de la programación de tareas dependientes en máquinas diferentes (task scheduling) [Tesis, Universidad Nacional Mayor de San Marcos]. https://hdl.handle.net/20.500.12672/3241
Tupia, M., Un Algoritmo GRASP para resolver el problema de la programación de tareas dependientes en máquinas diferentes (task scheduling) [Tesis]. PE: Universidad Nacional Mayor de San Marcos; 2005. https://hdl.handle.net/20.500.12672/3241
@mastersthesis{renati/476885,
title = "Un Algoritmo GRASP para resolver el problema de la programación de tareas dependientes en máquinas diferentes (task scheduling)",
author = "Tupia Anticona, Manuel Francisco",
publisher = "Universidad Nacional Mayor de San Marcos",
year = "2005"
}
--- Operation’s planning for Software Development has been a complicated by-solve problem experienced since the golden age of the use methodologies whose rule those process. In which it is used, as well as in the efficiency and sophistication of the algorithms that try to solve the problems that appear in a software project, since its origin in the middle of the 20th century. The previously mentioned problem is known within algorithmic as task scheduling and it is defined as follows: given a group of tasks (operations) to be scheduled within a group of machines (or human resources, or human-machine resources), find an appropriate execution order. It is a NP-difficult complexity problem, so it justifies the usage of heuristic methods to obtain approximate solutions. This thesis presents a GRASP heuristic goal to solve the variant in which tasks are dependent and executing entities are different one from the other: now it could be possible the planning of the operation s from the inception’s RUP phase. We are remarking in particular, two disciplines of RUP methodology: business modeling and requirement. Both a greedy algorithm and a GRASP heuristic goal with two relaxation parameters have been developed. Innovative proposition because until now nobody has tried to solve the problem this way. Likewise a mathematical model for the specific variant of the problem to be considered is shown. To show efficiency of the GRASP algorithm, we developed a prototype program that executes and compares the results obtained by greedy and GRASP algorithms. The GRASP algorithm improves by 6% the results of the greedy algorithm, for instances with up to 12500 variables involved. Finally we measured the quality of these results with those of the mathematical model which would obtain the exact solution for smaller instances, taking advantage of software that solves linear programming problems: the GRASP algorithm got close to the exact result within a range of 95 to 99%, and even equaled it in some tests. Keywords: Task scheduling, GRASP algorithm, Software development, resource planning in software projects, RUP.
This item is licensed under a Creative Commons License