| Because of the complexity presented by different science and engineering problems, there methods for the computation of the Fast Fourier Transform (FFT) have been developed. Some of these problems show symmetries in the input succession for the FFT computation. Taking into consideration the size of this succession and the kind of symmetry, it is possible to design symmetric algorithms for optimizing the FFT.; Currently, the usual method for the FFT computation does not take into consideration the symmetries of the input data, or the symmetry in the intermediate steps of the process. This thesis presents a method that takes into consideration these symmetries based on the mathematic theory exposed before in several articles. The problems presented when the algorithm is developed are: flow data alteration, generation of additional computations, which add computational cost to the algorithm, and the inability of expressing segments codes in a habitual form.; This work presents an implementation of the symmetric method, and the incorporation of other operations that makes the automatic way possible. This takes into consideration the problems that this process creates. This thesis gives evidence of an important reduction in execution times by using symmetric algorithm compared with a non-symmetric algorithm, for two types of symmetries commonly used in different applications in science and engineering. |
