Numerical Approximation of the Fourier Transform by the Fast Fourier Transform (FFT) AlgorithmA numerical approximation of the CFT requires evaluating a large number of integrals, each with a different integrand, since the values of this integral for a large range of are needed. The FFT can be effectively applied to this problem as follows. Let us assume that is zero outside the interval . Let be the sample spacing in for the input values of , which are assumed to be centered at zero, where is even. The values of m and β are chosen at the beginning in this procedure so the range of the interval is changing and depends on these parameters. The abscissas for the input data are , . Then we can write We define , . This is necessary for the above expression to be in the form of the DFT, denoted here by : The sample spacing (i.e., the resolution) of the result is fixed at the value as soon as one specifies the number of sample points and their interval . The above definition of the DFT is equivalent to the command Fourier[list, FourierParameters{1,-1}] in Mathematica. Usually, comparable sample spacing intervals in and are required. Then, one must put , or . It is clear that if one wishes to obtain accurate, high-resolution results using this procedure, then it may be necessary to set very large. More details can be found in D. H. Bailey and P. N. Swarztrauber, "A Fast Method for the Numerical Evaluation of Continuous Fourier and Laplace Transform," SIAM Journal on Scientific Computing, 15(5), 1994 pp. 1105–1110. You can choose the number of data points (an integer power of two) from the radio button menu. The appropriate value of the output sample interval for a given is displayed after choosing the input sample spacing . To show that the CFT in this algorithm is obtained in the form of points, check "points". "Numerical Approximation of the Fourier Transform by the Fast Fourier Transform (FFT) Algorithm" from The Wolfram Demonstrations Project http://demonstrations.wolfram.com/NumericalApproximationOfTheFourierTransformByTheFastFourierT/ Contributed by: Blazej Radzimirski |