Research reproduction pipelines

This page contains my reproducible projects that has already been published using Maneage (Managing data lineage, hosted at https://maneage.org), or its earlier incarnations. See Akhlaghi et al. (2021, arXiv:2006.03018) for an introduction to this concept.

All the necessary software used in the research below are free software, enabling any curious scientist to easily study and experiment on the source code of the software producing the results. All the numbers and plots of the paper where I am the principal author (or the parts that I contributed to in other papers) can be exactly reproduced through the scripts listed here. All the scripts within the pipeline (that call the separate programs) are also heavily commented to explain every step for the human reader thoroughly.

The paper or section number identification is a link to the Gitlab repository keeping the pipeline. The links following the paper's title point to the published paper.

Akhlaghi et al. (2020): Towards Long-term and Archivable Reproducibility, submitted to CiSE, arXiv:2006.03018.

This paper if fully devoted to Maneage and its founding criteria. It highlights the serious longevity-related problems of the tools used in most existing reproducible solutions, and describes how Maneage is able to address them. All the necessary reproducible supplements to this paper are published in zenodo.3408481.

One interesting by-product of the reproducibility that is implemented here is the link at the end of the caption in Figure 1. It points directly to the dataset that produced the plot, hosted on Zenodo (not a repository I control and might mistakenly delete in a few years)! With the data set having the same Git hash as the paper, as well as other metadata and copyright statements. Reading/using other papers would be so easy if this was more practiced. Through Maneage, I tried to simplify implementation of such features in derived projects.

Akhlaghi (2019): Carving out the low surface brightness universe with NoiseChisel, Invited talk at IAU Symposium 355. arXiv:1909.11230.

This is the first paper to fully implement the modern implementation of Maneage (that also installs its own software). It is primarily focused on all the improvements to the NoiseChisel software over the last four years since its initial submission. All the necessary reproducible supplements to this paper are published in zenodo.3408481.

Section 4 and Section 7.3 of Bacon et al. (2017): The MUSE Hubble Ultra Deep Field Survey I. Survey description, data reduction, and source detection. A&A 608, A1, arXiv: 1710.03002

In this project, I was in charge of the analysis behind the two sections mentioned above. Section 4 discusses the astrometry and photometry checks that we did on pseudo-broad-band images created from MUSE cubes to compare the results with broad-band HST images. The dataset necessary for this section (generally, everything necessary to exactly reproduce the results) is available in Zenodo. Section 7.3 focuses on how a broad-band segmentation map was assigned to objects that could only be detected in the MUSE cube with the ORIGIN software and the corresponding broad-band measurements were made. The reproduction pipeline and all the necessary software and data of this section are also uploaded to Zenodo.

Akhlaghi & Ichikawa (2015): Noise Based Detection and Segmentation of Nebulous Objects. ApJS 220, 1, arXiv: 1505.01664

This paper is a definition of the new noise-based detection and segmentation paradigm and is my first scientific paper. My work to make a fully reproducible scientific research was born while writing this paper. My softwares project (GNU Astronomy Utilities) was also born from this paper.