- Before using the FFT, you should cut the data to remove any discontinuity (relevant high frequency): smooth the oxidative and reductive parts of a scan separately.
- The Fourier transform
routine
does not care about the delta-x
intervals, which are only expected to be the same (i.e. the points are
thought to be evenly spaced), but it does not matter too much if they
are not: in both cases, if the points are roughly evenly spaced, the
fact that they are not exactly evenly spaced is equivalent to
introducing an additional (hopefully small) amount of noise.

Take care however about the GPES data recorded @ high scan rate: non regular delta-x in the data can be the reason for an unexpected problem using the FFT routines. The spacing along X is checked before the data are FFTed, but the test doesn't always work. See dx to check X-spacing, and int to interpolate data in the case of unevenly spaced points. - Since the delta-x value doesn't matter, in order to try and give the derivatives meaningful units, the derivative or second derivative (returned by the Fourier transform) is divided by the average delta-X or its square, respectively. This is obviously sensible only if the X data are evenly spaced. Smoothing/deriving using the spline interpolation does NOT require that points are evenly spaced, and the derivatives are returned in meaningful units (see b).