|
|
The Parallel Genetic Algorithm for Multicore Systems
Vrábel, Lukáš ; Šimek, Václav (oponent) ; Jaroš, Jiří (vedoucí práce)
Genetic algorithm is a powerful optimization and search method successfully used in practice to solve many different problems. Underlying concept is based on the evolutionary mechanics observed in nature. As the GAs are computationaly intense applications, it is natural that there are many efficient methods for parallelization of these algorithms. However, most of these methods deal with supercomputers or large computer clusters with specialized hardware, as these were the most common parallel architectures in the past. With modern-day computers the trend in personal computer design is also moving towards parallel architectures bringing small and cheap parallel multicore processors. That's why it is imperative to have efficient methods to exploit capabilities of this system. This document presents prototypes of new methods of parallel genetic algorithms designed especially for these multiprocessor computers with shared memory.
|
|
|
Verification Of Transmission Parameters Of Multicore Optical Fiber
Látal, Michal
With the ever-increasing demands on the transmission parameters of telecommunicationoptical systems, such as transmission capacity, new challenges arise for scientists to meet these requirements.One of the possible solutions to increase the transmission capacity of telecommunicationnetworks is the multicore optical fiber, which is the current trend in modern space division multiplex.First and second chapters of the article deal with the modern concept of space division multiplex.A selected seven-core optical fiber was tested for insertion loss, attenuation, length for individualcores, and fiber continuity using the direct and backscattering method. The achieved results of theperformed measurements are summarized and commented in the third chapter of the article.
|
|
|
The Parallel Genetic Algorithm for Multicore Systems
Vrábel, Lukáš ; Šimek, Václav (oponent) ; Jaroš, Jiří (vedoucí práce)
Genetic algorithm is a powerful optimization and search method successfully used in practice to solve many different problems. Underlying concept is based on the evolutionary mechanics observed in nature. As the GAs are computationaly intense applications, it is natural that there are many efficient methods for parallelization of these algorithms. However, most of these methods deal with supercomputers or large computer clusters with specialized hardware, as these were the most common parallel architectures in the past. With modern-day computers the trend in personal computer design is also moving towards parallel architectures bringing small and cheap parallel multicore processors. That's why it is imperative to have efficient methods to exploit capabilities of this system. This document presents prototypes of new methods of parallel genetic algorithms designed especially for these multiprocessor computers with shared memory.
|
host ::
RSS.