Documentation

FFT function. This function generates an fft of a real or complex data set. The output can be either a power spectrum or the raw fft. The input data can be either a single array of floats for real input data or an array of pairs of floats for complex input data array.

Returns:

s -1 if error, 0 otherwise.

Parameters:

p_mode - For an explanation of possible values see above Flags.
pset - A structure of type fftparam (see header fftmodule.h for definition). pset->windowed_data and pset->window_array should be allocated memory by the calling function. pset->windowed_data should have 2*timeseries_length*sizeof(float) length if the data is complex, or timeseries_length*sizeof(float) if real. pset->window_array should be of size timeseries_length*sizeof(float). This structure can be kept between multiple function calls if psGen will be called for the same sized data set multiple times for the same data_type.
timeseries_length - The number of real or complex data points in the time series. Should always be a power of two. Function will return an error (-1) if it is not.
data_type - Either DATA_REAL (0) or DATA_COMPLEX (1). Refers to the timeseries data input.
data - An array of real or complex input data.
timestep - The time interval (s) between successive data points.
output_format - One of 5 choices OUTPUT_AMPPERRTBIN (=0) a power spectrum with normalization such that a sine wave input of amplitude 1 yields a peak of height 1/sqrt2. OUTPUT_PSD (=1) a power spectral density with units rms amplitude per rtHz. OUTPUT_COMPMINUSFTOF (=2) an fft output with the most negative negative frequency { (-Fn*(n/2-1 / n/2)) where Fn is the nyquist frequency and n is the number of timeseries bins } first and the nyquist frequency last. Each bin has the real part first and the imaginary second. The normalization is such that the sum of contributions from -F and F yields a power spectral density in amplitude rms per rtHz. OUTPUT_COMPNATIVE (=3) returns data in the same format as used by fftw. The relevant manual pages are in: theory.lcs.mit.edu/~fftw/doc/fftw_3.htmlSEC17.For complex input data, the format is the same as numerical recipes fig 12.2.2 OUTPUT_GDSFORMAT (=4) .For real input data the format is r0 i0 r1 i1 ... r(n/2-1) i(n/2-1). The first bin is DC, so i0=0. The last bin is the frequency just below the nyquist frequency. For complex input data, the format is the same as numerical recipes figure 12.2.2 (DC first, then positive frequency components in ascending order, then the nyquist frequency, then from the most negative frequency above (-) the nyquist frequency up to the bin just below DC. Each bin is 2 elements, real then imag.
window - Determines the window type. WINDOW_UNIFORM (=0) no windowing WINDOW_HANNING (=1) from numerical recipes WINDOW_FLATTOP (=2) from Stanford SR785 manual WINDOW_WELCH (=3) from Numerical Recipes WINDOW_BARTLETT (=4) from Numerical Recipes WINDOW_BMH (=5) from Stanford SR785 manual WINDOW_HAMMING (=6) from Blackman and Tukey (see top of page)
result - Pointer to the result data. Always allocate timeseries_length*2*sizeof(float) bytes to this pointer, even if the output data set will be shorter than this.

Author:

Edward Daw, June 1999