Probing the nature of dark matter by forward modelling flux ratios in strong gravitational lenses

The free streaming length of dark matter particles determines the abundance of structure on sub-galactic scales. We present a statistical technique, amendable to any parametrization of subhalo density profile and mass function, to probe dark matter on these scales with quadrupole image lenses. We consider a warm dark matter particle with a mass function characterized by a normalization and free streaming scale m(hm). We forecast bounds on dark matter warmth for 120-180 lenses, attainable with future surveys, at typical lens (source) redshifts of 0.5 (1.5) in early-type galaxies with velocity dispersions of 220-270 km s(-1). We demonstrate that limits on mhm deteriorate rapidly with increasing uncertainty in image fluxes, underscoring the importance of precise measurements and accurate lens models. For our forecasts, we assume the deflectors in the lens sample do not exhibit complex morphologies, so we neglect systematic errors in their modelling. Omitting the additional signal from line-of-sight haloes, our constraints underestimate the true power of the method. Assuming cold dark matter, for a low normalization, corresponding to the destruction of all subhaloes within the host scale radius, we forecast 2 sigma bounds on m(hm) (thermal relic mass) of 10(7.5) (5.0), 10(8) (3.6), and 10(8.5) (2.7) M-circle dot (keV) for flux errors of 2 per cent, 4 per cent, and 8 per cent. With a higher normalization, these constraints improve to 10(7.2) (6.6), 10(7.5) (5.3), and 10(7.8) (4.3) M-circle dot (keV) with 120 systems. We are also able to measure the normalization of the mass function, which has implications for baryonic feedback models and tidal stripping.