WEIGHTED FIT OF OPTICAL-SPECTRA

Optical power spectra obtained from experiments oi from numerical simu lations are probabilistic in nature. When fitting such spectra to few- parameter analytical line shapes, the results from a weighted nonlinea r least squares fit are significantly better than those from an unweig hted fit. This is demonstrated by fitting a Lorentzian to directly sim ulated power spectra and to power spectra obtained from a rate-equatio n simulation of a diode laser output.