Dynamic and Polarimetric VLBI imaging with a multiscalar approach
Hendrik Müller, A. P. Lobanov
Abstract
Context . Due to the limited number of antennas and the limited observation time, an array of antennas in very long baseline interfer-ometry (VLBI) often samples the Fourier domain only very sparsely. Powerful deconvolution algorithms are needed to compute a final image. Multiscale imaging approaches such as DoG-HiT have recently been developed to solve the VLBI imaging problem and show promising performance: they are fast, accurate, unbiased, and automatic. Aims . We extend the multiscalar imaging approach to polarimetric imaging, to reconstructions of dynamically evolving sources, and finally to dynamic polarimetric reconstructions. Methods . These extensions (mr-support imaging) utilize a multiscalar approach. The time-averaged Stokes I image was decomposed by a wavelet transform into single subbands. We used the set of statistically significant wavelet coefficients, the multiresolution support (mr-support), computed by DoG-HiT as a prior in a constrained minimization manner; we fitted the single-frame (polarimetric) observables by only varying the coefficients in the multiresolution support. Results . The Event Horizon Telescope (EHT) is a VLBI array imaging supermassive black holes. We demonstrate on synthetic data that mr-support imaging offers ample regularization and is able to recover simple geometric dynamics at the horizon scale in a typical EHT setup. The approach is relatively lightweight, fast, and largely automatic and data driven. The ngEHT is a planned extension of the EHT designed to recover movies at the event horizon scales of a supermassive black hole. We benchmark the performance of mr-support imaging for the denser ngEHT configuration demonstrating the major improvements the additional ngEHT antennas will bring to dynamic polarimetric reconstructions. Conclusions . Current and upcoming instruments offer the observational possibility to do polarimetric imaging of dynamically evolving structural patterns with the highest spatial and temporal resolution. State-of-the-art dynamic reconstruction methods can capture this motion with a range of temporal regularizers and priors. With this work, we add an additional simpler regularizer to the list: constraining the reconstruction to the multiresolution support.