Fig. 26.

Download original image
Deviations in the logarithmic slope γ of the power-law model with different PSF settings. In all panels, the dashed distributions show the fiducial constraints – LENSTRONOMY in green and GLEE in purple. The LENSTRONOMY fiducial constraint is from the highest BIC value power-law model (Table 2), and the GLEE fiducial constraint is from the ‘power-law fiducial’ setup (Table 1). First panel: when only the reconstructed PSFs are interchanged to optimize the models, the constraint from one software program moves towards the other’s fiducial constraint to be consistent within 1.2 − 1.6σ. We note that GLEE model artificially creates supersampled version of the LENSTRONOMY PSF through interpolation to perform the tests here. Second panel: when both the PSFs and weighted uncertainty maps are interchanged, the resultant γ becomes slightly more consistent between the software programs within 1.0 − 1.5σ. Therefore, we conclude that the differences in the PSF itself significantly contributes to the discrepancy in the power-law γ constraint and not the particular method of weighting the uncertainty map to account for PSF uncertainty. Third panel: when models from both of the software programs are optimized only with UVIS data and interchanged PSFs, the GLEE constraint does not shift significantly; however, the LENSTRONOMY constraint shifts significantly towards the GLEE constraints. This result indicates that the difference between the GLEE and LENSTRONOMY fiducial models are largely created by the difference in the IR PSF. Fourth panel: when both software programs use a supersampled PSF with a supersampling factor of 3 reconstructed by LENSTRONOMY, the resultant γ constraint agrees very well, within 0.6σ.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.