If this is the case, the error may be triggered during the computation of the
> Local Whittle estimate of d. Javier, do you still get the error if you
> compute the raw periodogram (ie, without smoothing)?

Duh. Hate to reply to myself, but I've got it the other way around. I
meant WITH smoothing. In this case, the computation of the 2 estimates we
give of the long-memory parameter d is skipped, so if the problem is with
the Fourier routine you should be ok.

I don't understand very well what do you want to tell me. Do you refer to calculate the estimated espectral function with the Bartlett's lag window? If I do that I don't get the error