C8cfs_Czekala.txt
C8cfs_Czekala.txt
—
Plain Text,
1 KB (1729 bytes)
File contents
Cycle 8 NRAO ALMA Development Study Proposal – Regularized Maximum Likelihood Techniques for ALMA Spectral Line Imaging PI: Ian Czekala, PSU ABSTRACT Regularized Maximum Likelihood (RML) imaging techniques have been demonstrated to achieve higher angular resolution with sub-mm observations of continuum sources while maintaining superior image fidelity, for example with the recent Event Horizon Telescope images of M87. There is great promise in applying RML imaging techniques to ALMA spectral line measurement sets, since the probabilistic framework can naturally treat diverse array configurations while regularizing away artefacts originating from incomplete u-v sampling. However, it remains to be seen whether prior formulations, like entropy and sparsity, that have worked well in other domains will be successful for the varied morphologies of spectral line emission. We propose to develop and implement RML algorithms for ALMA spectral line imaging, focusing on achieving high image fidelity when utilizing multi-configuration aggregate datasets. As the ALMA archive continues to mature, techniques that can accurately image large and diverse quantities of data will drive science forward in key areas that require sensitivity and angular resolution, such as the kinematic detection of planets in protoplanetary disks and astrochemical domains. We will use our GPU-accelerated open source code MPoL to implement and rigorously test promising new prior choices like Gaussian process velocity-space regularization, and power spectrum regularization. Best-practices learned throughout this study will be disseminated to the community through publications in leading astronomical journals and open source software.
