A Fast Direct Solver for Mutual Coupling Analysis of Large Arrays of Reflector Antennas

Mutual coupling is a dominant systematic effect in dense reflector arrays, imprinting direction-dependent and frequency-dependent structure on embedded element patterns (EEPs) and currently limiting sensitivity in precision radio measurements. Accurate modelling of these effects requires full-wave simulations of structures that are electrically large at both the array and element levels, making conventional approaches computationally prohibitive. We present a Method-of-Moments (MoM) framework accelerated by a fast direct solver (FDS). The rotational symmetry of reflector dishes is exploited to efficiently compress self-interaction blocks of the impedance matrix. Mutual interactions are treated using a broadband multipole decomposition that remains efficient and accurate for closely spaced elements. We demonstrate the method on arrays of tens of reflectors from the Hydrogen Epoch of Reionization Array (HERA) telescope. To scale to larger arrays, the FDS is used to construct macro-basis functions (MBFs) from a smaller representative array and embed them within a conventional MBF scheme. This allows the first computation of EEPs for the 320-element HERA core on a 128-core workstation.

Removing the fat from your posterior samples with margarine

Bayesian workflows often require the introduction of nuisance parameters, yet for core science modelling one needs access to a marginal posterior density. In this work we use masked autoregressive flows and kernel density estimators to encapsulate the marginal posterior, allowing us to compute marginal Kullback-Leibler divergences and marginal Bayesian model dimensionalities in addition to generating samples and computing marginal log probabilities. We demonstrate this in application to topical cosmological examples of the Dark Energy Survey, and global 21cm signal experiments. In addition to the computation of marginal Bayesian statistics, this work is important for further applications in Bayesian experimental design, complex prior modelling and likelihood emulation. This technique is made publicly available in the pip-installable code margarine.