Deep radio-interferometric imaging with POLISH: DSA-2000 and weak lensing
Abstract
Radio interferometry allows astronomers to probe small spatial scales that are often inaccessible with single-dish instruments. However, recovering the radio sky from an interferometer is an ill-posed deconvolution problem that astronomers have worked on for half a century. More challenging still is achieving resolution below the array's diffraction limit, known as superresolution imaging. To this end, we have developed a new learning-based approach for radio interferometric imaging, leveraging recent advances in the classical computer vision problems of single-image superresolution and deconvolution. We have developed and trained a high-dynamic range residual neural network to learn the mapping between the dirty image and the true radio sky. We call this procedure POLISH, in contrast to the traditional CLEAN algorithm. The feed-forward nature of learning-based approaches like POLISH is critical for analysing data from the upcoming Deep Synoptic Array (DSA-2000). We show that POLISH achieves superresolution, and we demonstrate its ability to deconvolve real observations from the Very Large Array. Superresolution on DSA-2000 will allow us to measure the shapes and orientations of several hundred million star-forming radio galaxies (SFGs), making it a powerful cosmological weak lensing survey and probe of dark energy. We forecast its ability to constrain the lensing power spectrum, finding that it will be complementary to next-generation optical surveys such as Euclid.
Additional Information
© 2022 The Author(s). Published by Oxford University Press on behalf of Royal Astronomical Society. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model). Accepted 2022 April 15. Received 2022 April 14; in original form 2021 November 15. Published: 13 May 2022. We are grateful to Schmidt Futures for supporting the Radio Camera Initiative, under which this work was carried out. We thank Martin Krasser for his Tensorflow 2.x implementation of several superresolution algorithms and his advice on this paper's application. We thank Joe Zuntz for help with his COSMOSIS code and Ian Harrison for valuable conversations. We also thank Nitika Yadlapalli, Yuping Huang, and Jamie Bock for helpful discussion. Data Availability: The code for POLISH can be found at https://github.com/liamconnor/polish-pub/. This includes tools for simulating the microJansky radio sky as well as code to train and use the POLISH neural network.Attached Files
Published - stac1329.pdf
Submitted - 2111.03249.pdf
Files
Name | Size | Download all |
---|---|---|
md5:f90f62d08a0ba473b23cf1a251f1b0b4
|
3.6 MB | Preview Download |
md5:4dc95e233d21588264e3604afd6d6663
|
27.7 MB | Preview Download |
Additional details
- Eprint ID
- 113105
- Resolver ID
- CaltechAUTHORS:20220125-215138493
- Schmidt Futures Program
- Created
-
2022-01-26Created from EPrint's datestamp field
- Updated
-
2022-07-14Created from EPrint's last_modified field
- Caltech groups
- Astronomy Department