2013MNRAS.430....2D

This is the first paper in a series that presents a multiwavelength analysis of the archetype ultraluminous infrared galaxy IRAS F10214+4724, a gravitationally lensed, starburst/active galactic nucleus at z = 2.3. Here we present a new lens model and spatially resolved radio data, as well as a deep Hubble Space Telescope (HST) F160W map. The lens modelling employs a Bayesian Markov chain Monte Carlo algorithm with extended source, forward ray tracing. Using these high-resolution HST, Multi-Element Radio Linked Interferometer Network (MERLIN) and Very Large Array (VLA) maps, the algorithm allows us to constrain the level of distortion to the continuum spectral energy distribution resulting from emission components with differing magnification factors, due to their size and proximity to the caustic. Our lens model finds that the narrow-line region, and by proxy the active nucleus, is preferentially magnified. This supports previous claims that preferential magnification could mask the expected polycyclic aromatic hydrocarbon spectral features in the Spitzer mid-infrared spectrum which roughly trace the star-forming regions. Furthermore, we show that the arc-to-counter-image flux ratio is not a good estimate of the magnification in this system, despite its common use in the IRAS F10214+4724 literature. Our lens modelling suggests magnifications of µ ∼ 15-20±2 for the HST F814W, MERLIN 1.7GHz and VLA 8GHz maps, significantly lower than the canonical values of µ = 50-100 often used for this system. Systematic errors such as the dark matter density slope and colocation of stellar and dark matter centroids dominate the uncertainties in the lens model at the 40 per cent level.