Optimization Techniques for Solving Complex Problems (Wiley Series on Parallel and Distributed Computing)
Format: PDF / Kindle (mobi) / ePub
Real-world problems and modern optimization techniques to solve them
Here, a team of international experts brings together core ideas for solving complex problems in optimization across a wide variety of real-world settings, including computer science, engineering, transportation, telecommunications, and bioinformatics.
Part One—covers methodologies for complex problem solving including genetic programming, neural networks, genetic algorithms, hybrid evolutionary algorithms, and more.
Part Two—delves into applications including DNA sequencing and reconstruction, location of antennae in telecommunication networks, metaheuristics, FPGAs, problems arising in telecommunication networks, image processing, time series prediction, and more.
All chapters contain examples that illustrate the applications themselves as well as the actual performance of the algorithms.?Optimization Techniques for Solving Complex Problems is a valuable resource for practitioners and researchers who work with optimization in real-world settings.
Description of the Problem To determine the function of specific genes, scientists have learned to read the sequence of nucleotides comprising a DNA sequence in a process called DNA sequencing. To do that, multiple exact copies of the original DNA sequence are made. Each copy is then cut into short fragments at random positions. These are the first three steps depicted in Figure 16.1, and they take place in the laboratory. After the fragment set is obtained, a traditional assemble approach
algorithm, population size was fixed at 100 individuals, initialized with random feasible solutions. Mutation probability was set at 2 bits per string, recombination probability at 0.9, a binary tournament selection method was used, and a standard uniform crossover operator was chosen. The results are shown in Figure 21.2, where the relative distances to the best solution found by any of the algorithms are shown. As can be seen, the hybrid algorithm outperforms the original algorithms in
is explored by a complete algorithm to find the best possible offspring solution. We refer the interested reader to the book by Blum et al.  for a comprehensive introduction to hybrid metaheuristics. 23.1.1 Background The k-cardinality tree (KCT) problem was defined by Hamacher et al. . Subsequently [17,26], the NP-hardness of the problem was shown. The problem has several applications in practice (see Table 23.1). Optimization Techniques for Solving Complex Problems, Edited by
and desirable feature for the solutions which allow the MHs to lower the adaptation cost. CANONICAL MHs FOR DOPs 91 To generate solutions as flexible as possible, the following approaches can be considered: • Multiobjective. The MH is designed to optimize two objectives, one rep- resented by the original fitness function and the other by a metric value indicating solution flexibility . • Dynamic fitness. When the time to the next change is known or can be estimated, the fitness
problem into smaller problems with the same structure as that of the original problem. The class Solution has to provide an algorithm to put together partial solutions in order to obtain the solution to a larger problem through the combine() function. MALLBA::DNC provides a sequential resolution pattern and a message-passing master– slave resolution pattern for distributed memory machines [14,15]. This work presents a new fully distributed parallel skeleton that provides the same user