Exploiting the diversity of modeling methods to probe systematic biases in strong lensing analyses

In the past decade, the diversity of strong lens modeling methods has exploded, from being purely analytical to pixelated and non-parametric, or based on deep learning. We embrace this diversity by selecting different software packages and use them to blindly model independently simulated Hubble Space Telescope imaging data. To overcome the difficulties arising from using different codes and conventions, we use the COde-independent Organized LEns STandard (COOLEST) to store, compare and release all models in a self-consistent and human-readable manner. From an ensemble of six modeling methods, we study the recovery of the lens potential parameters and properties of the reconstructed source. In particular, we simulate and infer parameters of an elliptical power-law mass distribution with external shear for the lens while each modeling method reconstructs the source differently. We find that overall, both lens and source properties are recovered reasonably well, but systematic biases arise in all methods. Interestingly, we do not observe that a single method is significantly more accurate than others, and the amount of bias largely depends on the specific lens or source property of interest. By combining posterior distributions from individual methods using equal weights, the maximal systematic biases on lens model parameters inferred from individual models are reduced by a factor of 5.4 on average. We investigate a selection of modeling effects that partly explain the observed biases, such as the cuspy nature of the background source and the accuracy of the point spread function. This work introduces, for the first time, a generic framework to compare and ease the combination of models obtained from different codes and methods, which will be key to retain accuracy in future strong lensing analyses.
Comment: 24 pages, 16 figures. Submitted to A&A