On generating power law noise.

Based on the theory of spectral estimation we propose a new algorithm that is capable to produce the whole variety of possible non deterministic linear time series which exhibit a (1/f)^β spectrum. The key point of the new algorithm is to randomize both the phase and the amplitude of the Fourier transform of the data according to its stochastic nature. One possible application is the simulation of AGN lightcurves as well to analyze measured data as to test proposed models.