Date: June 13th, 2001

Attendees: M. Landry, Bill Butler (LHO), Erika, Virginio, Hiro, Mark Barton, Andrea, Biplab(Caltech)

Andrea Vicere talked about Burg algorithm for a linear predictor ( transparencies.) and how he applied that for studying two E2E problems (i) Problem of fast simulation of feedback loops with high frequency poles and zeros (we faced this for our model of the LIGO PSL system) (ii) Getting a good PSD (power-spectral-density) estimator which, for example, would be able to correctly compute the PSD for cases with large dynamical ranges, like in seismic noise.

The application of the linear predictor to the first case didn't improve the situation, but in the case of PSD estimator, it was successfully applied to predict a good whitening filter directly from the time-domain data stream and thus was very useful for computing steep PSDs of the seismic noise.

Currently, E2E has a module "psd_out" (written by Hiro) which calculates seismic noise PSD quite well but the whitening filter is known a priori. With the application of this linear predictor, we donot need to guess the characteristics of the whitening filter beforehand.

Any comments? write to bbhawal@ligo.caltech.edu