We investigate the effect of dark energy on the density profiles of dark matter haloes with a suite of cosmological N-body simulations and use our results to test analytic models .
We consider constant equation of state models , and allow both w \geq - 1 and w < -1 .
Using five simulations with w ranging from -1.5 to -0.5 , and with more than \sim 1600 well-resolved haloes each , we show that the halo concentration model of Bullock et al .
( 2001 ) accurately predicts the median concentrations of haloes over the range of w , halo masses , and redshifts that we are capable of probing .
We find that the Bullock et al .
( 2001 ) model works best when halo masses and concentrations are defined relative to an outer radius set by a cosmology-dependent virial overdensity .
For a fixed power spectrum normalization and fixed-mass haloes , larger values of w lead to higher concentrations and higher halo central densities , both because collapse occurs earlier and because haloes have higher virial densities .
While precise predictions of halo densities are quite sensitive to various uncertainties , we make broad comparisons to galaxy rotation curve data .
At fixed power spectrum normalization ( fixed \sigma _ { 8 } ) , w > -1 quintessence models seem to exacerbate the central density problem relative to the standard w = -1 
model .
For example , models with w \simeq - 0.5 seem disfavored by the data , which can be matched only by allowing extremely low normalizations , \sigma _ { 8 } \la 0.6 .
Meanwhile w < -1 models help to reduce the apparent discrepancy .
We confirm that the Jenkins et al .
( 2001 ) halo mass function provides an excellent approximation to the abundance of haloes in our simulations and extend its region of validity to include models with w < -1 .