Spectral analysis of the Lyα forest using wavelets

It is shown how wavelets may be used to analyse the absorption properties of the Ly alpha forest. The discrete wavelet transform of a quasi-stellar object spectrum is used to decompose the light fluctuations that comprise the forest into orthogonal wavelets. It is demonstrated that most of the signal is carried by the moderate to lower frequency wavelets in high-resolution spectra, and that a statistically acceptable description of even high signal-to-noise ratio spectra is provided by only a fraction (10-30 per cent) of the wavelets. The distributions of the wavelet coefficients provide a statistical basis for discriminating between different models of the Ly alpha forest. The method is illustrated using the measured spectrum of Q1937-1009. The procedure described is readily automated and may be used to process both measured spectra and the large number of spectra generated by numerical simulations, permitting a fair comparison between the two.