Modernisation of Software for Epoch of Reionisation Science

CI: – Dr. N. Barry

The most successful open-source Epoch of Reionisation pipeline is built in ageing, proprietary software. This is a significant limitation; only those in the community that can afford the licensing are contributing, essentially disenfranchising students from smaller universities or countries. Through ADACS, we hope to provide an easy-to-implement non-proprietary pipeline for Epoch of Reionisation graduate students and early career researchers, fully built in python. Not only would this engage a wider user base, but it would also provide vital experience to students in a modern language for both industry and academic career tracks.
This would be the first open-source python code base for reducing Epoch of Reionisation data. In addition, the pipeline to be translated already has an active development community. It has been used in over 17 publications, where 11 were student-led. The transition into python will be non-trivial given the delicate intersection of large quantities of data and high levels of precision required in Epoch of Reionisation science, and thus the expertise of ADACS is essential.
This is the second in a series of on-going proposals to realise this aim. Allocated time in 2021A determined the feasibility and required resources of the project, including the best way to approach systematic porting. A continuation of this project into 2021B would translate our three most delicate, intricate modules — the core of calibration and Fourier-space histogramming. This would allow an immediate integration into the current code base and the biggest impact possible.
Testing of the ported code will be done with small quantities of simulated data, thus reducing the computational requirements required to a modern laptop. An IDL license will be required that is version 8.5 or above in order to directly call python in situ, otherwise IDL licensing available on OzStar would suffice. In all, the computational requirements for the porting of the code are minimal.

Categories