Residual Entropy as a Diagnostic and Stopping Metric for CLEAN

We propose the use of entropy, measured from the spatial and flux distribution of pixels in the residual image, as a potential diagnostic and stopping metric for the CLEAN algorithm. Despite its broad success as the standard deconvolution approach in radio interferometry, finding the optimum stopping point for the iterative CLEAN algorithm is still a challenge. We show that the entropy of the residual image, measured during the final stages of CLEAN, can be computed without prior knowledge of the source structure or expected noise levels, and that finding the point of maximum entropy as a measure of randomness in the residual image serves as a robust stopping criterion. We also find that, when compared to the expected thermal noise in the image, the maximum entropy of the residuals is a useful diagnostic that can reveal the presence of data editing, calibration, or deconvolution issues that may limit the fidelity of the final CLEAN map.