Lessons from the first multiply imaged supernova: Revised strong-lensing models for the galaxy cluster macs j1149.5+2223

We present two grid-based lens models for the galaxy cluster MACS J1149.5+2223, concentrating on the properties of the first multiply imaged supernova Refsdal, which first appeared in this cluster as an Einstein cross (images S1-S4) around a cluster galaxy in late 2014, and about a year later as an additional image (SX) in a different image of the supernova (SN) host galaxy. One model is our updated light-traces-mass (LTM) stronglensing model for this cluster, and the second is a parametric model (dPIEeNFW), generated using the same pipeline but with a different parameterization. Together these two models probe different possible solutions in a relatively self-consistent manner and can be used to examine systematic uncertainties and relevant differences between the two parameterizations. We obtain reasonably similar (agreeing to within ;1σ-3σ, in most cases) time delays and magnification ratios, with respect to S1, from the two different methods, although the LTM predictions seem to be systematically shorter/smaller for some of the images. Most notably, the time delay [and 95% CI] between the Einstein cross (in particular, image S1), and SX, differs substantially between the parametric method (326 [300-359] days) and the LTM method (224 [198-306] days), which seems to underestimate the true reappearance time. The cause for this systematic difference is unclear at present. We speculate on its possible origin and note that a refined measurement of SN Refsdal's properties should help to more strongly discriminate between the two solutions, and thus between the two descriptions for the intrinsic shape of the underlying matter distribution. We also discuss the implications of our results for the Hubble constant.