SERVAL

SERVAL (Spherical Expansions for Radio Visibility Analysis at Large-N) is a visibility synthesis code designed for wide-field driftscan radio interferometers with a large number of baselines and non-redundant beams. It utilizes spherical harmonic methods to partially collapse the visibility integral, enabling efficient computation through flexible caching of terms. This architecture facilitates fast visibility synthesis in two flexible scenarios: handling varying beam models after caching a sky model, or handling varying sky models after caching the beam models. The goal of these approaches is to facilitate efficient forward modeling for cosmological 21cm interferometers similar in scale to HIRAX.

The code is designed to be highly performant on CPU-based HPC systems, featuring a C++ backend and a Python frontend. It includes useful tools for efficiently computing and sum-producting over families of Gaunt coefficients and Wigner 3j symbols, which are the core numerical aspects of the code.