An optimized Ly α forest inversion tool based on a quantitative comparison of existing reconstruction methods

• DOI: 10.1093/mnras/staa2225
• Publication Date: 2020-08-04T19:28:41Z
• Authors: Hendrik Müller; Christoph Behrens; David J E Marsh
• Keywords: Regularization; Robustness

We present a same-level comparison of the most prominent inversion methods for reconstruction matter density field in quasi-linear regime from Ly$\alpha$ forest flux. Moreover, we pathway refining framework numerical optimization. apply this approach to construct novel hybrid method. The which are used so far reconstructions Richardson-Lucy algorithm, an iterative Gauss-Newton method and statistical assuming one-to-one correspondence between study these high spectral resolutions such that thermal broadening becomes relevant. compared on synthetic data (generated with lognormal approach) respect their performance, accuracy, stability against noise, robustness systematic uncertainties. conclude offers accurate reconstruction, particular at small S/N, but has also largest complexity requires strongest assumptions. other two algorithms faster, comparably precise noise-levels, and, case approach, more robust inaccurate assumptions history intergalactic medium (IGM). use results refine using regularization. Our new low makes few about IGM, is shown be even if IGM not known. code will made publicly available under https://github.com/hmuellergoe/reglyman.