reglyman

Features

reglyman is a python library developed for the inversion of the Lyman alpha forest flux along single lines of sight at high spectral resolution. It provides tools for the fast creation of synthtetic Lyman-alpha forest flux data, tools for estimating the neutral hydrogen density along the lines of sight from the normalized flux and tools for their transverse interpolation.

The toolbox has been developed at the Institute for Astrophysics in Göttingen. It contains the source code for the publication Müller, Behrens, Marsh 2020: "An Optimized Ly-alpha Forest Inversion Tool Based on a Quantitative Comparison of Existing Reconstruction Methods" (https://academic.oup.com/mnras/article/497/4/4937/5881952). This paper provides more detailed descriptions of the inversion tools that are used. Please cite this reference, if you plan to use our software.

reglyman makes strong use of the existing regpy (regularization methods for inverse problems) and nbodykit (creating mocks of galxies) toolboxes. The Lyman-alpha forest is computed with the use of the lognormal approach which is also used within nbodykit. The inverse problem of recovering the neutral hydrogen density from the Lyman-alpha forest is implemented with the help of the regpy library.

The two repositories could be found under:

• nbodykit: https://nbodykit.readthedocs.io/en/latest/index.html . Nbodykit is presented in the paper Hand et. al. 2018: "nbodykit: An Open-source, Massively Parallel Toolkit for Large-scale Structure"; AJ, 160, 156
• regpy: https://num.math.uni-goettingen.de/regpy/

Please note that this repository is under heavy development, so do expect bugs and tiny documentation. Also it depends strongly on the regpy repository which is under heavy development too. Future regpy releases will largely extend the amount on inversion procedures available for Lyman-alpha forest analysis and can be directly incorporated in this repository. The reglyman repository will be extended in future releases.

Installation

We provide a pip install option. Download the files, navigate to the place where you stored the reglyman files and simply type: pip install . . The library depends on:

• regpy
• numpy
• scipy
• matplotlib
• nbodykit
• multiprocessing
• astropy
• joblib
• pmesh

Description

The file-structure of the repository is as follows:

Paper: Contains the arxive preprint of our parallel publication (https://academic.oup.com/mnras/article/497/4/4937/5881952). In this paper we compare the different inversion schemes and propose a new one. For more details on the inversion methods implemented within this library we refer to this manuscript and the references therein.

Examples: Consists of examples for the inversion along 25 lines of sights, i.e. routines for the creation of synthetic data, for mimicing noise and instrumental effects and for performing the inversion procedure along single lines of sight:

• rl: Richardson-Lucy algorithm
• irgn: Iterative Gauss-Newton method
• pc: Probability Conservation approach
• rpc: Regularized Probability Conservation approach

In addition we provide methods for interpolating between lines of sights:

• interpolate: An example file demonstrating the transverse interpolation between density fields along single lines of sights to a three dimensional grid by Bayesian statistics
• vel: An example file demonstrating the estimation of peculiar velcoities from measurements of the density in redshift space. This is alpha-quality software and very unstable.

The synthetic_data subordner consists of files for the creation of synthetic data (i.e. the density field, the spectrum and initial guesses for inversion procedure):

• generate_density: Generate the overdensity field
• synthetic_spectrum: Calculate the noisy and smoothed spectrum, find proper initial guess for inversion (either from the hydrogen number density or the baryonic overdensity)

Reglyman: The modules extending the options available in nbodykit and regpy:

• The operators subordner consists of the operators describing the forward computation of the Lyman-alpha forest (LymanAlpha) and supporting operators for the inversion procedures.
• In the solvers subordner we stored the solvers used for this study (see examples). More solver will be available within future releases of the regpy library.
• In the density subordner we stored routines available from nbodykit to simulate the cosmic overdensity field.
• The util ordner contains some supporting routines.
• The tomography subordner contains routines for the transverse interpolation between neighbouring lines of sight.

The repository has a very similar structure as the regpy repository to maximize compatibility with future releases. Thus, for more information on the running example files we refer to their documentation and the documentation within our example files.