Classifying non-redundancy in the HERA array

Abstract

HERA is a highly redundant radio interferometer array, where pairs of receivers with the same position vector between them should see exactly the same signal from the sky. We can use this fact to do a really good job of calibrating them. Unfortunately, the receivers are not perfectly identical, and so they don’t see exactly the same signal. This is called "non-redundancy". This project classifies the level of redundancy using a clustering machine learning technique. The aim is to see if any particular clustering algorithm can group different segments of the array into very similar blocks, so we can at least do a good job of redundantly calibrating within those blocks. We call this new calibration method, logi_cal, while the standard calibration method used in HERA is called redcal.