Abstract
The data reduction procedure for radio interferometers can be viewed as a combined calibration and imaging problem. We present an algorithm that unifies cross-calibration, self-calibration, and imaging. Because it is a Bayesian method, this algorithm not only calculates an estimate of the sky brightness distribution, but also provides an estimate of the joint uncertainty which entails both the uncertainty of the calibration and that of the actual observation. The algorithm is formulated in the language of information field theory and uses Metric Gaussian Variational Inference (MGVI) as the underlying statistical method. So far only direction-independent antenna-based calibration is considered. This restriction may be released in future work. An implementation of the algorithm is contributed as well.
Highlights
The algorithm is formulated in the language of information field theory and uses Metric Gaussian Variational Inference (MGVI) as the underlying statistical method
The expensive part of the evaluation of the sky model is a fast Fourier transform (FFT), which is in O(n log n) where n is the total number of pixels of the sky model
For realworld data sets the cost for thegridding exceeds the FFTs by far such that one likelihood evaluation is in O(N), where N
Summary
Resolve is reasonably fast but cannot compete in pure speed with algorithms like the Cotton-Schwab algorithm (Schwab 1984) as implemented in CASA This is rooted in the fact that resolve provides a single sky brightness distribution but needs to update the sky prior probability distribution according to the raw data in order to properly state how much the data has constrained the probability distribution and how much uncertainty is left in the final result. The current approach in the IFT community is to define a generative model that turns the degrees of freedom, which are learned by the algorithm into synthetic data that can be compared to the actual data in a squared-norm fashion (in the case of additive Gaussian noise) This approach is similar to the so-called radio interferometric measurement equation (RIME; Hamaker et al 1996; Perkins et al 2015; Smirnov 2011).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
