Filtering, smoothing etc

BIN-SPECTRUM	Average into bins of width N channels
CONVOLVE-SPECTRUM	Convolve with truncated gaussian
DIFFERENTIATE-SPECTRUM	Calculate difference spectrum
FOLD-SPECTRUM	Find symmetric part of spectrum
FOURIER-TRANSFORM	Self explanatory! DFT, so need not be $2^N$ long
FOURIER-POWER-SPECTRUM	$\log_{10}$ of squared modulus of DFT
HANN-SPECTRUM	Hanning smoothing $(0.25,0.5,0.25)$
SMOOTH-SPECTRUM	Boxcar average

The current spectrum may be filtered in a large number of ways. SMOOTH-SPECTRUM forms a running mean over an arbitrary number of points, while HANN-SPECTRUM applies the Hanning weighting function $(0.25,0.5,0.25)$ similarly. CONVOLVE-SPECTRUM allows the convolution of the spectrum with a user-defined gaussian having given half-width and cut-off. BIN-SPECTRUM averages channels into bins NCHAN channels wide starting from the first channel, or symmetrically about the centre channel if possible. For deconvolution of moon observations etc. DIFFERENTIATE-SPECTRUM is provided. FOLD-SPECTRUM extracts the symmetric part of the spectrum (about the centre channel).

All operations may be performed in the Fourier Transform domain using FOURIER-TRANSFORM for both forward and reverse transforms. Note that this is a direct discrete transform, so that spectra of an arbitrary length can be repeatedly transformed without leading to oversampling. The Fourier power spectrum can be calculated using FOURIER-POWER-SPECTRUM. This displays the squared modulus of the Fourier transform, with the magnitude plotted on arbitrary log scales. This is useful for checking the noise level before applying a Fourier or Wiener filter, for example.