2024/01/24 by Alejandro Mus, Hendrik Müller, Mus, Alejandro +5 · 1 citation
Engineering · Environmental Science · Physics and Astronomy · #Astrophysics of Galaxies (astro-ph.GA) #FOS: Mathematics #FOS: Physical sciences #GNSS positioning and interference #Instrumentation and Methods for Astrophysics (astro-ph.IM) #Optimization and Control (math.OC) #Radio Astronomy Observations and Technology #Soil Moisture and Remote Sensing
paper · pdf · doi:10.48550/arxiv.2401.13349
openalex publication_date 2024/01/24 · openalex created_date 2024/01/26 · openalex updated_date 2026/07/28
In Very Long Baseline Interferometry (VLBI), signals from multiple antennas combine to create a sparsely sampled virtual aperture, its effective diameter determined by the largest antenna separation. The inherent sparsity makes VLBI imaging an ill-posed inverse problem, prompting the use of algorithms like the Multiobjective Evolutionary Algorithm by Decomposition (MOEA/D), as proposed in the first paper of this series. This study focuses on extending MOEA/D to polarimetric and time dynamic reconstructions, particularly relevant for the VLBI community and the Event Horizon Telescope Collaboration (EHTC). MOEA/D's success in providing a unique, fast, and largely unsupervised representation of image structure serves as the basis for exploring these extensions. The extension involves incorporating penalty terms specific to total intensity imaging, time-variable, and polarimetric variants within MOEA/D's multiobjective, evolutionary framework. The Pareto front, representing non-dominated solutions, is computed, revealing clusters of proximities. Testing MOEA/D with synthetic datasets representative of EHTC's main targets demonstrates successful recovery of polarimetric and time-dynamic signatures despite sparsity and realistic data corruptions. MOEA/D's extension proves effective in the anticipated EHTC setting, offering an alternative and independent claim to existing methods. It not only explores the problem globally but also eliminates the need for parameter surveys, distinguishing it from Regularized Maximum Likelihood (RML) methods. MOEA/D emerges as a novel and useful tool for robustly characterizing polarimetric and dynamic signatures in VLBI datasets with minimal user-based choices. Future work aims to address the last remaining limitation of MOEA/D, specifically regarding the number of pixels and numerical performance, to establish it within the VLBI data reduction pipeline.