Accuracy vs. Complexity: Calibrating radio interferometer arrays with non-homogeneous element patterns

Abstract: Radio interferometer arrays with non-homogeneous element patterns are more difficult to calibrate compared to the more common homogeneous array. In particular, the non-homogeneity of the patterns has significant implications on the computational tractability of evaluating the calibration solutions. We apply the A-stacking technique to this problem and explore the trade-off to be made between the calibration accuracy and computational complexity. Through simulations, we show that this technique can be favourably applied in the context of an SKA-Low station. We show that the minimum accuracy requirements can be met at a significantly reduced computational cost, and this cost can be reduced even further if the station calibration timescale is relaxed from 10 minutes to several hours. We demonstrate the impact antenna designs with differing levels of non-homogeneity have on the overall computational complexity in addition to some cases where calibration performs poorly.