We use hydrodynamic cosmological simulations to explore the evolution of the intergalactic medium ( IGM ) transmissivity from z=2 through the epoch of reionization .
We simulate a concordance \Lambda CDM model in a 9.6 Mpc box with a comoving spatial resolution of 37.5 kpc .
Reionization is treated in the optically thin approximation using an ultraviolet background ( UVB ) that includes evolving stellar and QSO source populations .
In this approximation , ionization bubble overlap is treated by ramping up the UVB over a finite redshift interval \Delta z \approx 0.5 , consistent with the inhomogeneous reionization simulations of Razoumov et al .
( 2002 ) for several reionization redshifts .
We construct noiseless synthetic HI Ly \alpha absorption spectra by casting lines of sight ( LOS ) through our continuously evolving box and analyze their properties using standard techniques .
Different spectral resolutions are also studied by convolving full resolution data down to R=36,000 and R=5,300 depending on redshift .
Parametric fits to the data are provided based on either analytic approximations or straightforward regressions .

We find a smooth evolution of the effective optical depth under a power law with a slope of 4.16 \pm 0.02 up to the epoch of reionization .
The smooth profile can also be fitted to the Songaila and Cowie ( 2002 ) parametrization F ( g , z ) .
The normalized ionization rate g is then recovered through our spectra which agrees , within the error of the fit , with the input ionization rate we used in the simulation .
As we cross into the epoch of reionization , the mean transmitted flux ( MTF ) and variance to the mean transmitted flux sharply deviate from a smooth evolution .
However , the simultaneous sharp increase of the variance and sharp decrease of the mean transmitted flux introduces large margins of error which place a high degree of uncertainty to the computed optical depth evolution profile .
A distinction between high and low transmission lines of sight shows that the two subsamples skew the results towards two different directions that may infer a continuation of a smooth profile or underestimate the global transmissivity properties during reionization .
However , despite the statistical uncertainty in inferring the reionization profile from spectra , the end of an opacity phase transition of the IGM correlates well with the redshift when both the mean and variance of the transmitted flux rapidly deviate from a smooth profile .
Furthermore , we quantify the relation between the line of sight and the cosmic flux variance , which is computed from the statistical average of the flux variance along all lines of sight and conclude that because the latter is a lower bound estimate due to limitations imposed by our box size , it is a more sensitive tool than the MTF in mapping reionization .
It nonetheless causes the distribution of mean fluxes along lines of sight to have an increasing flatness as the redshift increases .
We estimate that in our cosmic realization of reionization and regardless of spectral resolution , an unobtainable number of lines of sight is needed to yield a normal distribution of LOS-mean fluxes that would allow an estimate of the MTF with less than 10 % relative margin of error .

In addition to optical transmission , we compare the predicted dark gap length distribution with observations .
We show that this statistic is sensitive to spectral resolution at reionization redshifts , but overall in agreement with results by Songaila and Cowie ( 2002 ) .
Finally , we derive a positive correlation between the mean optical depth within a gap and the size of the gap , which relates ” transmission statistics ” to ” dark gap statistics ” in high redshift studies of the IGM .