I would like to experiment with chaotic noise versus Gaussian
noise sources.

Previous threads in comp.dsp have recommend summing 12 uniform
RNG's to approximate Gaussian random noise, which can be done
with a reasonably small computational overhead per sample.

Does anyone know of any reasonably low overhead algorithms to
generate random noise samples from an approximation to fat-tailed,
Pareto, or power law decay distributions (which seem to more closely
model chaotic noise sources)?


Thanks,

rhn A.T nicholson d.0.t C-o-M