I present an analysis of techniques used to determine the mass power spectrum from
observations of the intergalactic medium.
Simple Monte Carlo simulations are presented which clarify some of the problems
which must be overcome when attempting any inversion process. The 1 dimensional
flux power spectrum is dependent on both the clustering of the absorption features and
on the scaling relations existing between the dark matter and baryon distributions.
More realistic simulated spectra are used to illustrate differences in the mean flux
power spectrum for a range of cosmologies and the ratio of the 1 dimensional linear mass
and flux power spectra. The mass distribution is much more dependent on cosmological
parameters than the flux power spectrum, highlighting the very accurate measurements
of the latter quantity required for an accurate recovery of the mass density power
These spectra are further analysed by deconvolution into Voigt profiles. This technique is shown to be an excellent approximation, in spite of the current cosmic web
paradigm of the Ly-a forest. However the power spectrum of the positions of these
fitted lines is shown to be a very poor indicator of the underlying mass density field.
Inferring the 3 dimensional forms of power spectra from 1 dimensional data is
shown to be problematic. This, coupled with analysis of the correlation matrix of the
flux power spectrum, casts doubts on the reliability of the error analysis presented in