Reionization is thought to have occurred in the redshift range of 6 < z < 9 , which is now being probed by both deep galaxy surveys and CMB observations .
Using halo abundance matching over the redshift range 5 < z < 8 and assuming smooth , continuous gas accretion , we develop a model for the star formation efficiency f _ { \star } of dark matter halos at z > 6 that matches the measured galaxy luminosity functions at these redshifts .
We find that f _ { \star } peaks at \sim 30 \% at halo masses M \sim 10 ^ { 11 } – 10 ^ { 12 } M _ { \odot } , in qualitative agreement with its behavior at lower redshifts .
We then investigate the cosmic star formation histories and the corresponding models of reionization for a range of extrapolations to small halo masses .
We use a variety of observations to further constrain the characteristics of the galaxy populations , including the escape fraction of UV photons .
Our approach provides an empirically-calibrated , physically-motivated model for the properties of star-forming galaxies sourcing the epoch of reionization .
In the case where star formation in low-mass halos is maximally efficient , an average escape fraction \sim 0.1 can reproduce the optical depth reported by Planck , whereas inefficient star formation in these halos requires either about twice as many UV photons to escape , or an escape fraction that increases towards higher redshifts .
Our models also predict how future observations with JWST can improve our understanding of these galaxy populations .