A new generation DIFMAP for Modelfitting Interferometric Data and Estimating Variances, Biases and Correlations

TL;DR

This paper introduces ngDIFMAP, an advanced interferometric data modeling tool that improves parameter estimation accuracy and error analysis for radio astronomy observations, especially under noisy and incomplete data conditions.

Contribution

The paper presents ngDIFMAP, a new version of DIFMAP that employs simulated annealing for better global optimization in interferometric model fitting, enhancing accuracy and error estimation.

Findings

ngDIFMAP achieves more accurate parameter estimates in noisy data.

The tool effectively predicts variances, biases, and correlations in interferometric measurements.

It demonstrates the impact of data errors and coverage loss on parameter recovery.

Abstract

We present the program `Catalogue of proper motions in extragalactic jets from Active galactic Nuclei with Very large Array Studies' or CAgNVAS, with the objective of using archival and new VLA observations to measure proper motions of jet components beyond hundred parsecs. This objective requires extremely high accuracy in component localization. Interferometric datasets are noisy and often lack optimal coverage of the visibility plane, making interpretation of subtleties in deconvolved imaging inaccurate. Fitting models to complex visibilities, rather than working in the imaging plane, is generally preferred as a solution when one needs the most accurate description of the true source structure. In this paper, we present a new generation version of $\texttt{DIFMAP}$ (\texttt{ngDIFMAP}) to model and fit interferometric closure quantities developed for the CAgNVAS program. \texttt{ngDIFMAP} uses a global optimization algorithm based on simulated annealing, which results in more accurate parameter estimation especially when the number of parameters is high. Using this package we demonstrate the ramifications of amplitude and phase errors, as well as loss of $u-v$ coverage, on parameters estimated from visibility data. The package can be used to accurately predict variance, bias, and correlations between parameters. Our results demonstrate the limits on information recovery from noisy interferometric data, with a particular focus on the accurate reporting of errors on measured quantities.