Showing 1–50 of 197 results for author: Scaife, A

E(2)-Equivariant Features in Machine Learning for Morphological Classification of Radio Galaxies

Authors: Natalie E. P. Lines, Joan Font-Quer Roset, Anna M. M. Scaife

Abstract: With the growth of data from new radio telescope facilities, machine-learning approaches to the morphological classification of radio galaxies are increasingly being utilised. However, while widely employed deep-learning models using convolutional neural networks (CNNs) are equivariant to translations within images, neither CNNs nor most other machine-learning approaches are equivariant to additio… ▽ More With the growth of data from new radio telescope facilities, machine-learning approaches to the morphological classification of radio galaxies are increasingly being utilised. However, while widely employed deep-learning models using convolutional neural networks (CNNs) are equivariant to translations within images, neither CNNs nor most other machine-learning approaches are equivariant to additional isometries of the Euclidean plane, such as rotations and reflections. Recent work has attempted to address this by using G-steerable CNNs, designed to be equivariant to a specified subset of 2-dimensional Euclidean, E(2), transformations. Although this approach improved model performance, the computational costs were a recognised drawback. Here we consider the use of directly extracted E(2)-equivariant features for the classification of radio galaxies. Specifically, we investigate the use of Minkowski functionals (MFs), Haralick features (HFs) and elliptical Fourier descriptors (EFDs). We show that, while these features do not perform equivalently well to CNNs in terms of accuracy, they are able to inform the classification of radio galaxies, requiring ~50 times less computational runtime. We demonstrate that MFs are the most informative, EFDs the least informative, and show that combinations of all three result in only incrementally improved performance, which we suggest is due to information overlap between feature sets. △ Less

Submitted 13 June, 2024; originally announced June 2024.

Comments: Submitted RASTI

Evaluating Bayesian deep learning for radio galaxy classification

Authors: Devina Mohan, Anna M. M. Scaife

Abstract: The radio astronomy community is rapidly adopting deep learning techniques to deal with the huge data volumes expected from the next generation of radio observatories. Bayesian neural networks (BNNs) provide a principled way to model uncertainty in the predictions made by such deep learning models and will play an important role in extracting well-calibrated uncertainty estimates on their outputs.… ▽ More The radio astronomy community is rapidly adopting deep learning techniques to deal with the huge data volumes expected from the next generation of radio observatories. Bayesian neural networks (BNNs) provide a principled way to model uncertainty in the predictions made by such deep learning models and will play an important role in extracting well-calibrated uncertainty estimates on their outputs. In this work, we evaluate the performance of different BNNs against the following criteria: predictive performance, uncertainty calibration and distribution-shift detection for the radio galaxy classification problem. △ Less

Submitted 28 May, 2024; originally announced May 2024.

Comments: Accepted to the 40th Conference on Uncertainty in Artificial Intelligence (UAI 2024)

Scaling Laws for Galaxy Images

Authors: Mike Walmsley, Micah Bowles, Anna M. M. Scaife, Jason Shingirai Makechemu, Alexander J. Gordon, Annette M. N. Ferguson, Robert G. Mann, James Pearson, Jürgen J. Popp, Jo Bovy, Josh Speagle, Hugh Dickinson, Lucy Fortson, Tobias Géron, Sandor Kruk, Chris J. Lintott, Kameswara Mantha, Devina Mohan, David O'Ryan, Inigo V. Slijepevic

Abstract: We present the first systematic investigation of supervised scaling laws outside of an ImageNet-like context - on images of galaxies. We use 840k galaxy images and over 100M annotations by Galaxy Zoo volunteers, comparable in scale to Imagenet-1K. We find that adding annotated galaxy images provides a power law improvement in performance across all architectures and all tasks, while adding trainab… ▽ More We present the first systematic investigation of supervised scaling laws outside of an ImageNet-like context - on images of galaxies. We use 840k galaxy images and over 100M annotations by Galaxy Zoo volunteers, comparable in scale to Imagenet-1K. We find that adding annotated galaxy images provides a power law improvement in performance across all architectures and all tasks, while adding trainable parameters is effective only for some (typically more subjectively challenging) tasks. We then compare the downstream performance of finetuned models pretrained on either ImageNet-12k alone vs. additionally pretrained on our galaxy images. We achieve an average relative error rate reduction of 31% across 5 downstream tasks of scientific interest. Our finetuned models are more label-efficient and, unlike their ImageNet-12k-pretrained equivalents, often achieve linear transfer performance equal to that of end-to-end finetuning. We find relatively modest additional downstream benefits from scaling model size, implying that scaling alone is not sufficient to address our domain gap, and suggest that practitioners with qualitatively different images might benefit more from in-domain adaption followed by targeted downstream labelling. △ Less

Submitted 3 April, 2024; originally announced April 2024.

Comments: 10+6 pages, 12 figures. Appendix C2 based on arxiv:2206.11927. Code, demos, documentation at https://github.com/mwalmsley/zoobot

MIGHTEE Polarization Early Science Fields: The Deep Polarized Sky

Authors: A. R. Taylor, S. Sekhar, L. Heino, A. M. M. Scaife, J. Stil, M. Bowles, M. Jarvis, I. Heywood, J. D. Collier

Abstract: The MeerKAT International GigaHertz Tiered Extragalactic Exploration (MIGHTEE) is one of the MeerKAT large survey projects, designed to pathfind SKA key science. MIGHTEE is undertaking deep radio imaging of four well observed fields (COSMOS, XMM-LSS, ELAIS S1 and CDFS) totaling 20 square degrees to $μ$Jy sensitivities. Broadband imaging observations between 880--1690 MHz yield total intensity cont… ▽ More The MeerKAT International GigaHertz Tiered Extragalactic Exploration (MIGHTEE) is one of the MeerKAT large survey projects, designed to pathfind SKA key science. MIGHTEE is undertaking deep radio imaging of four well observed fields (COSMOS, XMM-LSS, ELAIS S1 and CDFS) totaling 20 square degrees to $μ$Jy sensitivities. Broadband imaging observations between 880--1690 MHz yield total intensity continuum, spectro-polarimetry, and atomic hydrogen spectral imaging. Early science data from MIGHTEE are being released from initial observations of COSMOS and XMM-LSS. This paper describes the spectro-polarimetric observations, the polarization data processing of the MIGHTEE early science fields, and presents polarization data images and catalogues. The catalogues include radio spectral index, redshift information and Faraday rotation measure synthesis results for 13,271 total intensity radio sources down to a polarized intensity detection limit of $\sim$20 $μ$Jy\,bm$^{-1}$. Polarized signals were detected from 324 sources. For the polarized detections we include a catalogue of Faraday Depth from both Faraday Synthesis and $Q$, $U$ fitting, as well as total intensity and polarization spectral indices. The distribution of redshift of the total radio sources and detected polarized sources are the same, with median redshifts of 0.86 and 0.82 respectively. Depolarization of the emission at longer-wavelengths is seen to increase with decreasing total-intensity spectral index, implying that depolarisation is intrinsic to the radio sources. No evidence is seen for a redshift dependence of the variance of Faraday Depth. △ Less

Submitted 20 December, 2023; originally announced December 2023.

Comments: 12 pages, 16 figures

Rare Galaxy Classes Identified In Foundation Model Representations

Authors: Mike Walmsley, Anna M. M. Scaife

Abstract: We identify rare and visually distinctive galaxy populations by searching for structure within the learned representations of pretrained models. We show that these representations arrange galaxies by appearance in patterns beyond those needed to predict the pretraining labels. We design a clustering approach to isolate specific local patterns, revealing groups of galaxies with rare and scientifica… ▽ More We identify rare and visually distinctive galaxy populations by searching for structure within the learned representations of pretrained models. We show that these representations arrange galaxies by appearance in patterns beyond those needed to predict the pretraining labels. We design a clustering approach to isolate specific local patterns, revealing groups of galaxies with rare and scientifically-interesting morphologies. △ Less

Submitted 5 December, 2023; originally announced December 2023.

Comments: Accepted at Machine Learning and the Physical Sciences Workshop, NeurIPS 2023

Bayesian Imaging for Radio Interferometry with Score-Based Priors

Authors: Noe Dia, M. J. Yantovski-Barth, Alexandre Adam, Micah Bowles, Pablo Lemos, Anna M. M. Scaife, Yashar Hezaveh, Laurence Perreault-Levasseur

Abstract: The inverse imaging task in radio interferometry is a key limiting factor to retrieving Bayesian uncertainties in radio astronomy in a computationally effective manner. We use a score-based prior derived from optical images of galaxies to recover images of protoplanetary disks from the DSHARP survey. We demonstrate that our method produces plausible posterior samples despite the misspecified galax… ▽ More The inverse imaging task in radio interferometry is a key limiting factor to retrieving Bayesian uncertainties in radio astronomy in a computationally effective manner. We use a score-based prior derived from optical images of galaxies to recover images of protoplanetary disks from the DSHARP survey. We demonstrate that our method produces plausible posterior samples despite the misspecified galaxy prior. We show that our approach produces results which are competitive with existing radio interferometry imaging algorithms. △ Less

Submitted 29 November, 2023; originally announced November 2023.

Comments: 10+4 pages, 6 figures, Machine Learning and the Physical Sciences Workshop, NeurIPS 2023

Combining astrophysical datasets with CRUMB

Authors: Fiona A. M. Porter, Anna M. M. Scaife

Abstract: At present, the field of astronomical machine learning lacks widely-used benchmarking datasets; most research employs custom-made datasets which are often not publicly released, making comparisons between models difficult. In this paper we present CRUMB, a publicly-available image dataset of Fanaroff-Riley galaxies constructed from four "parent" datasets extant in the literature. In addition to pr… ▽ More At present, the field of astronomical machine learning lacks widely-used benchmarking datasets; most research employs custom-made datasets which are often not publicly released, making comparisons between models difficult. In this paper we present CRUMB, a publicly-available image dataset of Fanaroff-Riley galaxies constructed from four "parent" datasets extant in the literature. In addition to providing the largest image dataset of these galaxies, CRUMB uses a two-tier labelling system: a "basic" label for classification and a "complete" label which provides the original class labels used in the four parent datasets, allowing for disagreements in an image's class between different datasets to be preserved and selective access to sources from any desired combination of the parent datasets. △ Less

Submitted 17 November, 2023; originally announced November 2023.

Comments: Accepted in Machine Learning and the Physical Sciences Workshop at NeurIPS 2023; 6 pages, 1 figure, 1 table

MCMC to address model misspecification in Deep Learning classification of Radio Galaxies

Authors: Devina Mohan, Anna Scaife

Abstract: The radio astronomy community is adopting deep learning techniques to deal with the huge data volumes expected from the next-generation of radio observatories. Bayesian neural networks (BNNs) provide a principled way to model uncertainty in the predictions made by deep learning models and will play an important role in extracting well-calibrated uncertainty estimates from the outputs of these mode… ▽ More The radio astronomy community is adopting deep learning techniques to deal with the huge data volumes expected from the next-generation of radio observatories. Bayesian neural networks (BNNs) provide a principled way to model uncertainty in the predictions made by deep learning models and will play an important role in extracting well-calibrated uncertainty estimates from the outputs of these models. However, most commonly used approximate Bayesian inference techniques such as variational inference and MCMC-based algorithms experience a "cold posterior effect (CPE)", according to which the posterior must be down-weighted in order to get good predictive performance. The CPE has been linked to several factors such as data augmentation or dataset curation leading to a misspecified likelihood and prior misspecification. In this work we use MCMC sampling to show that a Gaussian parametric family is a poor variational approximation to the true posterior and gives rise to the CPE previously observed in morphological classification of radio galaxies using variational inference based BNNs. △ Less

Submitted 14 November, 2023; originally announced November 2023.

Comments: Accepted in Machine Learning and the Physical Sciences Workshop at NeurIPS 2023; 6 pages, 1 figure, 1 table

Galaxy Zoo DESI: Detailed Morphology Measurements for 8.7M Galaxies in the DESI Legacy Imaging Surveys

Authors: Mike Walmsley, Tobias Géron, Sandor Kruk, Anna M. M. Scaife, Chris Lintott, Karen L. Masters, James M. Dawson, Hugh Dickinson, Lucy Fortson, Izzy L. Garland, Kameswara Mantha, David O'Ryan, Jürgen Popp, Brooke Simmons, Elisabeth M. Baeten, Christine Macmillan

Abstract: We present detailed morphology measurements for 8.67 million galaxies in the DESI Legacy Imaging Surveys (DECaLS, MzLS, and BASS, plus DES). These are automated measurements made by deep learning models trained on Galaxy Zoo volunteer votes. Our models typically predict the fraction of volunteers selecting each answer to within 5-10\% for every answer to every GZ question. The models are trained o… ▽ More We present detailed morphology measurements for 8.67 million galaxies in the DESI Legacy Imaging Surveys (DECaLS, MzLS, and BASS, plus DES). These are automated measurements made by deep learning models trained on Galaxy Zoo volunteer votes. Our models typically predict the fraction of volunteers selecting each answer to within 5-10\% for every answer to every GZ question. The models are trained on newly-collected votes for DESI-LS DR8 images as well as historical votes from GZ DECaLS. We also release the newly-collected votes. Extending our morphology measurements outside of the previously-released DECaLS/SDSS intersection increases our sky coverage by a factor of 4 (5,000 to 19,000 deg$^2$) and allows for full overlap with complementary surveys including ALFALFA and MaNGA. △ Less

Submitted 20 September, 2023; originally announced September 2023.

Comments: 20 pages. Accepted at MNRAS. Catalog available via https://zenodo.org/record/7786416. Pretrained models available via https://github.com/mwalmsley/zoobot. Vizier and Astro Data Lab access not yet available. With thanks to the Galaxy Zoo volunteers

Probing magnetic fields in the circumgalactic medium using polarization data from MIGHTEE

Authors: K. Böckmann, M. Brüggen, V. Heesen, A. Basu, S. P. O'Sullivan, I. Heywood, M. Jarvis, A. Scaife, J. Stil, R. Taylor, N. J. Adams, R. A. A. Bowler, M. N. Tudorache

Abstract: The detection and study of magnetic fields surrounding galaxies is important to understand galaxy evolution since magnetic fields are tracers for dynamical processes in the circumgalactic medium (CGM) and can have a significant impact on the evolution of the CGM. The Faraday rotation measure (RM) of the polarized light of background radio sources passing through the magnetized CGM of intervening g… ▽ More The detection and study of magnetic fields surrounding galaxies is important to understand galaxy evolution since magnetic fields are tracers for dynamical processes in the circumgalactic medium (CGM) and can have a significant impact on the evolution of the CGM. The Faraday rotation measure (RM) of the polarized light of background radio sources passing through the magnetized CGM of intervening galaxies can be used as a tracer for the strength and extent of magnetic fields around galaxies. We use rotation measures observed by the MIGHTEE-POL (MeerKAT International GHz Tiered Extragalactic Exploration POLarisation) survey by MeerKAT in the XMM-LSS and COSMOS fields to investigate the RM around foreground star-forming galaxies. We use spectroscopic catalogs of star-forming and blue cloud galaxies to measure the RM of MIGHTEE-POL sources as a function of the impact parameter from the intervening galaxy. We then repeat this procedure using a deeper galaxy catalog with photometric redshifts. For the spectroscopic star-forming sample we find a redshift-corrected |RM| excess of 5.6 +/- 2.3 rad m-2 which corresponds to a 2.5 sigma significance around galaxies with a median redshift of z = 0.46 for impact parameters below 130 kpc only selecting the intervenor with the smallest impact parameter. Making use of a photometric galaxy catalog and taking into account all intervenors with Mg < -13.6 mag, the signal disappears. We find no indication for a correlation between redshift and RM, nor do we find a connection between the total number of intervenors to the total |RM| . We have presented tentative evidence that the CGM of star-forming galaxies is permeated by coherent magnetic fields within the virial radius. We conclude that mostly bright, star-forming galaxies with impact parameters less than 130 kpc significantly contribute to the RM of the background radio source. △ Less

Submitted 22 August, 2023; originally announced August 2023.

Comments: 11 pages, 8 figures, accepted for publication in A&A

Journal ref: A&A 678, A56 (2023)

Radio Galaxy Zoo: Towards building the first multi-purpose foundation model for radio astronomy with self-supervised learning

Authors: Inigo V. Slijepcevic, Anna M. M. Scaife, Mike Walmsley, Micah Bowles, O. Ivy Wong, Stanislav S. Shabala, Sarah V. White

Abstract: In this work, we apply self-supervised learning with instance differentiation to learn a robust, multi-purpose representation for image analysis of resolved extragalactic continuum images. We train a multi-use model which compresses our unlabelled data into a structured, low dimensional representation which can be used for a variety of downstream tasks (e.g. classification, similarity search). We… ▽ More In this work, we apply self-supervised learning with instance differentiation to learn a robust, multi-purpose representation for image analysis of resolved extragalactic continuum images. We train a multi-use model which compresses our unlabelled data into a structured, low dimensional representation which can be used for a variety of downstream tasks (e.g. classification, similarity search). We exceed baseline supervised Fanaroff-Riley classification performance by a statistically significant margin, with our model reducing the test set error by up to half. Our model is also able to maintain high classification accuracy with very few labels, with only 7.79% error when only using 145 labels. We further demonstrate that by using our foundation model, users can efficiently trade off compute, human labelling cost and test set accuracy according to their respective budgets, allowing for efficient classification in a wide variety of scenarios. We highlight the generalizability of our model by showing that it enables accurate classification in a label scarce regime with data from the new MIGHTEE survey without any hyper-parameter tuning, where it improves upon the baseline by ~8%. Visualizations of our labelled and un-labelled data show that our model's representation space is structured with respect to physical properties of the sources, such as angular source extent. We show that the learned representation is scientifically useful even if no labels are available by performing a similarity search, finding hybrid sources in the RGZ DR1 data-set without any labels. We show that good augmentation design and hyper-parameter choice can help achieve peak performance, while emphasising that optimal hyper-parameters are not required to obtain benefits from self-supervised pre-training. △ Less

Submitted 18 October, 2023; v1 submitted 25 May, 2023; originally announced May 2023.

MiraBest: A Dataset of Morphologically Classified Radio Galaxies for Machine Learning

Authors: Fiona A. M. Porter, Anna M. M. Scaife

Abstract: The volume of data from current and future observatories has motivated the increased development and application of automated machine learning methodologies for astronomy. However, less attention has been given to the production of standardised datasets for assessing the performance of different machine learning algorithms within astronomy and astrophysics. Here we describe in detail the MiraBest… ▽ More The volume of data from current and future observatories has motivated the increased development and application of automated machine learning methodologies for astronomy. However, less attention has been given to the production of standardised datasets for assessing the performance of different machine learning algorithms within astronomy and astrophysics. Here we describe in detail the MiraBest dataset, a publicly available batched dataset of 1256 radio-loud AGN from NVSS and FIRST, filtered to $0.03 < z < 0.1$, manually labelled by Miraghaei and Best (2017) according to the Fanaroff-Riley morphological classification, created for machine learning applications and compatible for use with standard deep learning libraries. We outline the principles underlying the construction of the dataset, the sample selection and pre-processing methodology, dataset structure and composition, as well as a comparison of MiraBest to other datasets used in the literature. Existing applications that utilise the MiraBest dataset are reviewed, and an extended dataset of 2100 sources is created by cross-matching MiraBest with other catalogues of radio-loud AGN that have been used more widely in the literature for machine learning applications. △ Less

Submitted 18 May, 2023; originally announced May 2023.

Comments: 14 pages, 2 figures, accepted by RASTI

Radio Galaxy Zoo EMU: Towards a Semantic Radio Galaxy Morphology Taxonomy

Authors: Micah Bowles, Hongming Tang, Eleni Vardoulaki, Emma L. Alexander, Yan Luo, Lawrence Rudnick, Mike Walmsley, Fiona Porter, Anna M. M. Scaife, Inigo Val Slijepcevic, Elizabeth A. K. Adams, Alexander Drabent, Thomas Dugdale, Gülay Gürkan, Andrew M. Hopkins, Eric F. Jimenez-Andrade, Denis A. Leahy, Ray P. Norris, Syed Faisal ur Rahman, Xichang Ouyang, Gary Segal, Stanislav S. Shabala, O. Ivy Wong

Abstract: We present a novel natural language processing (NLP) approach to deriving plain English descriptors for science cases otherwise restricted by obfuscating technical terminology. We address the limitations of common radio galaxy morphology classifications by applying this approach. We experimentally derive a set of semantic tags for the Radio Galaxy Zoo EMU (Evolutionary Map of the Universe) project… ▽ More We present a novel natural language processing (NLP) approach to deriving plain English descriptors for science cases otherwise restricted by obfuscating technical terminology. We address the limitations of common radio galaxy morphology classifications by applying this approach. We experimentally derive a set of semantic tags for the Radio Galaxy Zoo EMU (Evolutionary Map of the Universe) project and the wider astronomical community. We collect 8,486 plain English annotations of radio galaxy morphology, from which we derive a taxonomy of tags. The tags are plain English. The result is an extensible framework which is more flexible, more easily communicated, and more sensitive to rare feature combinations which are indescribable using the current framework of radio astronomy classifications. △ Less

Submitted 14 April, 2023; originally announced April 2023.

Comments: 17 pages, 11 Figures, Accepted at MNRAS

The Faraday Rotation Measure Grid of the LOFAR Two-metre Sky Survey: Data Release 2

Authors: S. P. O'Sullivan, T. W. Shimwell, M. J. Hardcastle, C. Tasse, G. Heald, E. Carretti, M. Brüggen, V. Vacca, C. Sobey, C. L. Van Eck, C. Horellou, R. Beck, M. Bilicki, S. Bourke, A. Botteon, J. H. Croston, A. Drabent, K. Duncan, V. Heesen, S. Ideguchi, M. Kirwan, L. Lawlor, B. Mingo, B. Nikiel-Wroczyński, J. Piotrowska , et al. (2 additional authors not shown)

Abstract: A Faraday rotation measure (RM) catalogue, or RM Grid, is a valuable resource for the study of cosmic magnetism. Using the second data release (DR2) from the LOFAR Two-metre Sky Survey (LoTSS), we have produced a catalogue of 2461 extragalactic high-precision RM values across 5720 deg$^{2}$ of sky (corresponding to a polarized source areal number density of $\sim$0.43 deg$^{-2}$). The linear polar… ▽ More A Faraday rotation measure (RM) catalogue, or RM Grid, is a valuable resource for the study of cosmic magnetism. Using the second data release (DR2) from the LOFAR Two-metre Sky Survey (LoTSS), we have produced a catalogue of 2461 extragalactic high-precision RM values across 5720 deg$^{2}$ of sky (corresponding to a polarized source areal number density of $\sim$0.43 deg$^{-2}$). The linear polarization and RM properties were derived using RM synthesis from the Stokes $Q$ and $U$ channel images at an angular resolution of 20'' across a frequency range of 120 to 168 MHz with a channel bandwidth of 97.6 kHz. The fraction of total intensity sources ($>1$ mJy beam$^{-1}$) found to be polarized was $\sim$0.2%. The median detection threshold was 0.6 mJy beam$^{-1}$ ($8σ_{QU}$), with a median RM uncertainty of 0.06 rad m$^{-2}$ (although a systematic uncertainty of up to 0.3 rad m$^{-2}$ is possible, after the ionosphere RM correction). The median degree of polarization of the detected sources is 1.8%, with a range of 0.05% to 31%. Comparisons with cm-wavelength RMs indicate minimal amounts of Faraday complexity in the LoTSS detections, making them ideal sources for RM Grid studies. Host galaxy identifications were obtained for 88% of the sources, along with redshifts for 79% (both photometric and spectroscopic), with the median redshift being 0.6. The focus of the current catalogue was on reliability rather than completeness, and we expect future versions of the LoTSS RM Grid to have a higher areal number density. In addition, 25 pulsars were identified, mainly through their high degrees of linear polarization. △ Less

Submitted 18 January, 2023; originally announced January 2023.

Comments: 20 pages, 23 figures, 3 tables. Accepted for publication in MNRAS. Go to https://lofar-mksp.org/data/ to download the catalogue and associated data products

JCMT BISTRO Observations: Magnetic Field Morphology of Bubbles Associated with NGC 6334

Authors: Mehrnoosh Tahani, Pierre Bastien, Ray S. Furuya, Kate Pattle, Doug Johnstone, Doris Arzoumanian, Yasuo Doi, Tetsuo Hasegawa, Shu-ichiro Inutsuka, Simon Coudé, Laura Fissel, Michael Chun-Yuan Chen, Frédérick Poidevin, Sarah Sadavoy, Rachel Friesen, Patrick M. Koch, James Di Francesco, Gerald H. Moriarty-Schieven, Zhiwei Chen, Eun Jung Chung, Chakali Eswaraiah, Lapo Fanciullo, Tim Gledhill, Valentin J. M. Le Gouellec, Thiem Hoang , et al. (120 additional authors not shown)

Abstract: We study the HII regions associated with the NGC 6334 molecular cloud observed in the sub-millimeter and taken as part of the B-fields In STar-forming Region Observations (BISTRO) Survey. In particular, we investigate the polarization patterns and magnetic field morphologies associated with these HII regions. Through polarization pattern and pressure calculation analyses, several of these bubbles… ▽ More We study the HII regions associated with the NGC 6334 molecular cloud observed in the sub-millimeter and taken as part of the B-fields In STar-forming Region Observations (BISTRO) Survey. In particular, we investigate the polarization patterns and magnetic field morphologies associated with these HII regions. Through polarization pattern and pressure calculation analyses, several of these bubbles indicate that the gas and magnetic field lines have been pushed away from the bubble, toward an almost tangential (to the bubble) magnetic field morphology. In the densest part of NGC 6334, where the magnetic field morphology is similar to an hourglass, the polarization observations do not exhibit observable impact from HII regions. We detect two nested radial polarization patterns in a bubble to the south of NGC 6334 that correspond to the previously observed bipolar structure in this bubble. Finally, using the results of this study, we present steps (incorporating computer vision; circular Hough Transform) that can be used in future studies to identify bubbles that have physically impacted magnetic field lines. △ Less

Submitted 21 December, 2022; originally announced December 2022.

Comments: Accepted for publication in Astrophysical Journal (ApJ)

A New Task: Deriving Semantic Class Targets for the Physical Sciences

Authors: Micah Bowles, Hongming Tang, Eleni Vardoulaki, Emma L. Alexander, Yan Luo, Lawrence Rudnick, Mike Walmsley, Fiona Porter, Anna M. M. Scaife, Inigo Val Slijepcevic, Gary Segal

Abstract: We define deriving semantic class targets as a novel multi-modal task. By doing so, we aim to improve classification schemes in the physical sciences which can be severely abstracted and obfuscating. We address this task for upcoming radio astronomy surveys and present the derived semantic radio galaxy morphology class targets. We define deriving semantic class targets as a novel multi-modal task. By doing so, we aim to improve classification schemes in the physical sciences which can be severely abstracted and obfuscating. We address this task for upcoming radio astronomy surveys and present the derived semantic radio galaxy morphology class targets. △ Less

Submitted 27 October, 2022; v1 submitted 26 October, 2022; originally announced October 2022.

Comments: 6 pages, 1 figure, Accepted at Fifth Workshop on Machine Learning and the Physical Sciences (NeurIPS 2022), Neural Information Processing Systems 2022

The JCMT BISTRO Survey: A Spiral Magnetic Field in a Hub-filament Structure, Monoceros R2

Authors: Jihye Hwang, Jongsoo Kim, Kate Pattle, Chang Won Lee, Patrick M. Koch, Doug Johnstone, Kohji Tomisaka, Anthony Whitworth, Ray S. Furuya, Ji-hyun Kang, A-Ran Lyo, Eun Jung Chung, Doris Arzoumanian, Geumsook Park, Woo** Kwon, Shinyoung Kim, Motohide Tamura, Jungmi Kwon, Archana Soam, Ilseung Han, Thiem Hoang, Kyoung Hee Kim, Takashi Onaka, Eswaraiah Chakali, Derek Ward-Thompson , et al. (135 additional authors not shown)

Abstract: We present and analyze observations of polarized dust emission at 850 $μ$m towards the central 1 pc $\times$ 1 pc hub-filament structure of Monoceros R2 (Mon R2). The data are obtained with SCUBA-2/POL-2 on the James Clerk Maxwell Telescope (JCMT) as part of the BISTRO (B-fields in Star-forming Region Observations) survey. The orientations of the magnetic field follow the spiral structure of Mon R… ▽ More We present and analyze observations of polarized dust emission at 850 $μ$m towards the central 1 pc $\times$ 1 pc hub-filament structure of Monoceros R2 (Mon R2). The data are obtained with SCUBA-2/POL-2 on the James Clerk Maxwell Telescope (JCMT) as part of the BISTRO (B-fields in Star-forming Region Observations) survey. The orientations of the magnetic field follow the spiral structure of Mon R2, which are well-described by an axisymmetric magnetic field model. We estimate the turbulent component of the magnetic field using the angle difference between our observations and the best-fit model of the underlying large-scale mean magnetic field. This estimate is used to calculate the magnetic field strength using the Davis-Chandrasekhar-Fermi method, for which we also obtain the distribution of volume density and velocity dispersion using a column density map derived from $Herschel$ data and the C$^{18}$O ($J$ = 3-2) data taken with HARP on the JCMT, respectively. We make maps of magnetic field strengths and mass-to-flux ratios, finding that magnetic field strengths vary from 0.02 to 3.64 mG with a mean value of 1.0 $\pm$ 0.06 mG, and the mean critical mass-to-flux ratio is 0.47 $\pm$ 0.02. Additionally, the mean Alfvén Mach number is 0.35 $\pm$ 0.01. This suggests that in Mon R2, magnetic fields provide resistance against large-scale gravitational collapse, and magnetic pressure exceeds turbulent pressure. We also investigate the properties of each filament in Mon R2. Most of the filaments are aligned along the magnetic field direction and are magnetically sub-critical. △ Less

Submitted 13 December, 2022; v1 submitted 12 October, 2022; originally announced October 2022.

Comments: This paper is accepted to the ApJ

Learning useful representations for radio astronomy "in the wild" with contrastive learning

Authors: Inigo Val Slijepcevic, Anna M. M. Scaife, Mike Walmsley, Micah Bowles

Abstract: Unknown class distributions in unlabelled astrophysical training data have previously been shown to detrimentally affect model performance due to dataset shift between training and validation sets. For radio galaxy classification, we demonstrate in this work that removing low angular extent sources from the unlabelled data before training produces qualitatively different training dynamics for a co… ▽ More Unknown class distributions in unlabelled astrophysical training data have previously been shown to detrimentally affect model performance due to dataset shift between training and validation sets. For radio galaxy classification, we demonstrate in this work that removing low angular extent sources from the unlabelled data before training produces qualitatively different training dynamics for a contrastive model. By applying the model on an unlabelled data-set with unknown class balance and sub-population distribution to generate a representation space of radio galaxies, we show that with an appropriate cut threshold we can find a representation with FRI/FRII class separation approaching that of a supervised baseline explicitly trained to separate radio galaxies into these two classes. Furthermore we show that an excessively conservative cut threshold blocks any increase in validation accuracy. We then use the learned representation for the downstream task of performing a similarity search on rare hybrid sources, finding that the contrastive model can reliably return semantically similar samples, with the added bonus of finding duplicates which remain after pre-processing. △ Less

Submitted 18 July, 2022; originally announced July 2022.

Comments: Accepted in: Machine Learning for Astrophysics Workshop at the Thirty Ninth International Conference on Machine Learning (ICML 2022); final version

A GPU-based Imager with Polarised Primary-beam Correction

Authors: Chris J. Skipper, Anna M. M. Scaife, J. Patrick Leahy

Abstract: The next generation of radio telescopes will strive for unprecedented dynamic range across wide fields of view, and direction-dependent gains such as the gain from the primary-beam pattern, or leakage of one Stokes product into another, must be removed from the cleaned images if dynamic range is to reach its full potential. Unfortunately, such processing is extremely computationally intensive, and… ▽ More The next generation of radio telescopes will strive for unprecedented dynamic range across wide fields of view, and direction-dependent gains such as the gain from the primary-beam pattern, or leakage of one Stokes product into another, must be removed from the cleaned images if dynamic range is to reach its full potential. Unfortunately, such processing is extremely computationally intensive, and is made even more challenging by the very large volumes of data that these instruments will generate. Here we describe a new GPU-based imager, aimed primarily at use with the ASKAP telescope, that is capable of generating cleaned, full-polarisation images that include wide-field, primary-beam, and polarisation leakage corrections. △ Less

Submitted 30 June, 2022; originally announced June 2022.

Comments: 4 pages, 1 figure, conference paper accepted for 3rd URSI AT-AP-RASC, 29 May - 3 June 2022

Towards Galaxy Foundation Models with Hybrid Contrastive Learning

Authors: Mike Walmsley, Inigo Val Slijepcevic, Micah Bowles, Anna M. M. Scaife

Abstract: New astronomical tasks are often related to earlier tasks for which labels have already been collected. We adapt the contrastive framework BYOL to leverage those labels as a pretraining task while also enforcing augmentation invariance. For large-scale pretraining, we introduce GZ-Evo v0.1, a set of 96.5M volunteer responses for 552k galaxy images plus a further 1.34M comparable unlabelled galaxie… ▽ More New astronomical tasks are often related to earlier tasks for which labels have already been collected. We adapt the contrastive framework BYOL to leverage those labels as a pretraining task while also enforcing augmentation invariance. For large-scale pretraining, we introduce GZ-Evo v0.1, a set of 96.5M volunteer responses for 552k galaxy images plus a further 1.34M comparable unlabelled galaxies. Most of the 206 GZ-Evo answers are unknown for any given galaxy, and so our pretraining task uses a Dirichlet loss that naturally handles unknown answers. GZ-Evo pretraining, with or without hybrid learning, improves on direct training even with plentiful downstream labels (+4% accuracy with 44k labels). Our hybrid pretraining/contrastive method further improves downstream accuracy vs. pretraining or contrastive learning, especially in the low-label transfer regime (+6% accuracy with 750 labels). △ Less

Submitted 23 June, 2022; originally announced June 2022.

Comments: Accepted at the ICML 2022 Workshop on Machine Learning for Astrophysics. Data: www.github.com/mwalmsley/pytorch-galaxy-datasets. Please reach out to share your labelled data - all contributions will be credited in future work

A Compressed Sensing Faraday Depth Reconstruction Framework for the MeerKAT MIGHTEE-POL Survey

Authors: Miguel Cárcamo, Anna Scaife, Russ Taylor, Matt Jarvis, Micah Bowles, Srikrishna Sekhar, Lennart Heino, Jeroen Stil

Abstract: In this work we present a novel compute framework for reconstructing Faraday depth signals from noisy and incomplete spectro-polarimetric radio datasets. This framework is based on a compressed-sensing approach that addresses a number of outstanding issues in Faraday depth reconstruction in a systematic and scaleable manner. We apply this framework to early-release data from the MeerKAT MIGHTEE po… ▽ More In this work we present a novel compute framework for reconstructing Faraday depth signals from noisy and incomplete spectro-polarimetric radio datasets. This framework is based on a compressed-sensing approach that addresses a number of outstanding issues in Faraday depth reconstruction in a systematic and scaleable manner. We apply this framework to early-release data from the MeerKAT MIGHTEE polarisation survey. △ Less

Submitted 7 June, 2022; originally announced June 2022.

Comments: URSI AT-RASC 2022 Young Scientist Award paper. arXiv admin note: substantial text overlap with arXiv:2205.01413

CS-ROMER: A novel compressed sensing framework for Faraday depth reconstruction

Authors: Miguel Cárcamo, Anna M. M. Scaife, Emma L. Alexander, J. Patrick Leahy

Abstract: The reconstruction of Faraday depth structure from incomplete spectral polarization radio measurements using the RM Synthesis technique is an under-constrained problem requiring additional regularisation. In this paper we present cs-romer: a novel object-oriented compressed sensing framework to reconstruct Faraday depth signals from spectro-polarization radio data. Unlike previous compressed sensi… ▽ More The reconstruction of Faraday depth structure from incomplete spectral polarization radio measurements using the RM Synthesis technique is an under-constrained problem requiring additional regularisation. In this paper we present cs-romer: a novel object-oriented compressed sensing framework to reconstruct Faraday depth signals from spectro-polarization radio data. Unlike previous compressed sensing applications, this framework is designed to work directly with data that are irregularly sampled in wavelength-squared space and to incorporate multiple forms of compressed sensing regularisation. We demonstrate the framework using simulated data for the VLA telescope under a variety of observing conditions, and we introduce a methodology for identifying the optimal basis function for reconstruction of these data, using an approach that can also be applied to datasets from other telescopes and over different frequency ranges. In this work we show that the delta basis function provides optimal reconstruction for VLA L-band data and we use this basis with observations of the low-mass galaxy cluster Abell 1314 in order to reconstruct the Faraday depth of its constituent cluster galaxies. We use the cs-romer framework to de-rotate the Galactic Faraday depth contribution directly from the wavelength-squared data and to handle the spectral behaviour of different radio sources in a direction-dependent manner. The results of this analysis show that individual galaxies within Abell 1314 deviate from the behaviour expected for a Faraday-thin screen such as the intra-cluster medium and instead suggest that the Faraday rotation exhibited by these galaxies is dominated by their local environments. △ Less

Submitted 26 October, 2022; v1 submitted 3 May, 2022; originally announced May 2022.

Comments: Final version after peer reviewing

Journal ref: Miguel Cárcamo, Anna M M Scaife, Emma L Alexander, J Patrick Leahy, CS-ROMER: A novel compressed sensing framework for Faraday depth reconstruction, Monthly Notices of the Royal Astronomical Society, 2022;, stac3031

Radio Galaxy Zoo: Using semi-supervised learning to leverage large unlabelled data-sets for radio galaxy classification under data-set shift

Authors: Inigo V. Slijepcevic, Anna M. M. Scaife, Mike Walmsley, Micah Bowles, Ivy Wong, Stanislav S. Shabala, Hongming Tang

Abstract: In this work we examine the classification accuracy and robustness of a state-of-the-art semi-supervised learning (SSL) algorithm applied to the morphological classification of radio galaxies. We test if SSL with fewer labels can achieve test accuracies comparable to the supervised state-of-the-art and whether this holds when incorporating previously unseen data. We find that for the radio galaxy… ▽ More In this work we examine the classification accuracy and robustness of a state-of-the-art semi-supervised learning (SSL) algorithm applied to the morphological classification of radio galaxies. We test if SSL with fewer labels can achieve test accuracies comparable to the supervised state-of-the-art and whether this holds when incorporating previously unseen data. We find that for the radio galaxy classification problem considered, SSL provides additional regularisation and outperforms the baseline test accuracy. However, in contrast to model performance metrics reported on computer science benchmarking data-sets, we find that improvement is limited to a narrow range of label volumes, with performance falling off rapidly at low label volumes. Additionally, we show that SSL does not improve model calibration, regardless of whether classification is improved. Moreover, we find that when different underlying catalogues drawn from the same radio survey are used to provide the labelled and unlabelled data-sets required for SSL, a significant drop in classification performance is observered, highlighting the difficulty of applying SSL techniques under dataset shift. We show that a class-imbalanced unlabelled data pool negatively affects performance through prior probability shift, which we suggest may explain this performance drop, and that using the Frechet Distance between labelled and unlabelled data-sets as a measure of data-set shift can provide a prediction of model performance, but that for typical radio galaxy data-sets with labelled sample volumes of O(1000), the sample variance associated with this technique is high and the technique is in general not sufficiently robust to replace a train-test cycle. △ Less

Submitted 4 May, 2022; v1 submitted 19 April, 2022; originally announced April 2022.

Comments: Accepted to MNRAS. 14 pages

The contribution of magnetized galactic outflows to extragalactic Faraday rotation

Authors: Andres Aramburo-Garcia, Kyrylo Bondarenko, Alexey Boyarsky, Andrii Neronov, Anna Scaife, Anastasia Sokolenko

Abstract: Galactic outflows driven by star formation and active galactic nuclei blow bubbles into their local environments, causing galactic magnetic fields to be carried into intergalactic space. We explore the redshift-dependent effect of these magnetized bubbles on the Faraday Rotation Measure (RM) of extragalactic radio sources. Using the IllustrisTNG cosmological simulations, we separate the contributi… ▽ More Galactic outflows driven by star formation and active galactic nuclei blow bubbles into their local environments, causing galactic magnetic fields to be carried into intergalactic space. We explore the redshift-dependent effect of these magnetized bubbles on the Faraday Rotation Measure (RM) of extragalactic radio sources. Using the IllustrisTNG cosmological simulations, we separate the contribution from magnetic bubbles from that of the volume-filling magnetic component expected to be due to the seed field originating in the Early Universe. We use this separation to extract the redshift dependence of each component and to compare TNG model predictions with observation measurements of the NRAO VLA Sky Survey (NVSS). We find that magnetized bubbles provide a sizeable contribution to the extragalactic RM, with redshift-independent $\langle |{\rm RM}| \rangle \simeq 13$ rad/m$^2$ for sources at redshifts $z\ge 2$. This is close to the mean residual RM of $16$ rad/m$^2$ found from NVSS data in this redshift range. Using the IllustrisTNG simulations, we also evaluate a simple model for the contribution to residual RM from individual host galaxies and show that this contribution is negligible at high-redshift. While the contribution from magnetic bubbles in the IllustrisTNG model is currently compatible with observational measurements of residual RM, the next-generation RM sky surveys, which will be free from the wrap** uncertainty, have larger statistics and better sensitivity should be able to observe predicted flat contribution from magnetic bubbles at large redshifts. This should allow to experimentally probe magnetic bubbles and check models of galaxy feedback in cosmological simulations. △ Less

Submitted 13 April, 2022; originally announced April 2022.

Comments: 6 pages, 3 figures

Revision of Faraday rotation measure constraints on the primordial magnetic field using the IllustrisTNG simulation

Authors: Andres Aramburo-Garcia, Kyrylo Bondarenko, Alexey Boyarsky, Andrii Neronov, Anna Scaife, Anastasia Sokolenko

Abstract: Previously derived Faraday rotation constraints on the volume-filling intergalactic magnetic field (IGMF) have used analytic models that made a range of simplifying assumptions about magnetic field evolution in the intergalactic medium and did not consider the effect of baryonic feedback on large-scale structures. In this work we revise existing Faraday rotation constraints on the IGMF using a num… ▽ More Previously derived Faraday rotation constraints on the volume-filling intergalactic magnetic field (IGMF) have used analytic models that made a range of simplifying assumptions about magnetic field evolution in the intergalactic medium and did not consider the effect of baryonic feedback on large-scale structures. In this work we revise existing Faraday rotation constraints on the IGMF using a numerical model of the intergalactic medium from the IllustrisTNG cosmological simulation that includes a sophisticated model of the baryonic feedback. We use the IllustrisTNG model to calculate the rotation measure and compare the resulting mean and median of the absolute value of the rotation measure with data from the NRAO VLA Sky Survey (NVSS). The numerical model of the intergalactic medium includes a full magneto-hydrodynamic model of the compressed primordial magnetic field as well as a model of the regions where the magnetic field is not primordial, but is rather produced by the process of baryonic feedback. Separating these two types of regions, we are able to assess the influence of the primordial magnetic field on the Faraday rotation signal. We find that by correcting for regions of compressed primordial field and accounting for the fact that part of the intergalactic medium is occupied by magnetic fields spread by baryonic feedback processes rather than by the primordial field relaxes the Faraday rotation bound by a factor of $\simeq 3$. This results in $B_0<1.8\times10^{-9}$ G for large correlation length IGMFs. △ Less

Submitted 7 September, 2022; v1 submitted 12 April, 2022; originally announced April 2022.

Comments: Replaced with the published version

Report number: FERMILAB-PUB-22-299-V

Magnetic field strength in cosmic web filaments

Authors: E. Carretti, V. Vacca, S. P. O'Sullivan, G. H. Heald, C. Horellou, H. J. A. Rottgering, A. M. M. Scaife, T. W. Shimwell, A. Shulevski, C. Stuardi, T. Vernstrom

Abstract: We used the Rotation Measure (RM) catalogue derived from the LOFAR Two-metre Sky Survey Data Release 2 (LoTSS DR2) at 144-MHz to measure the evolution with redshift of the extragalactic RM (RRM: Residual RM) and the polarization fraction ($p$) of sources in low density environments. We also measured the same at 1.4-GHz by cross-matching with the NRAO VLA Sky Survey RM catalogue. We find that RRM v… ▽ More We used the Rotation Measure (RM) catalogue derived from the LOFAR Two-metre Sky Survey Data Release 2 (LoTSS DR2) at 144-MHz to measure the evolution with redshift of the extragalactic RM (RRM: Residual RM) and the polarization fraction ($p$) of sources in low density environments. We also measured the same at 1.4-GHz by cross-matching with the NRAO VLA Sky Survey RM catalogue. We find that RRM versus redshift is flat at 144-MHz, but, once redshift-corrected, it shows evolution at high significance. Also $p$ evolves with redshift with a decrement by a factor of $\sim$8 at $z\sim2$. Comparing the 144-MHz and 1.4-GHz data, we find that the observed RRM and $p$ are most likely to have an origin local to the source at 1.4-GHz, while a cosmic web filament origin is favoured at 144-MHz. If we attribute the entire signal to filaments, we infer a mean rest frame RRM per filament of RRM_{0,f} = 0.71 \pm 0.07 rad m^{-2} and a magnetic field per filament of B_f = 32 \pm 3 nG. This is in agreement with estimates obtained with a complementary method based on synchrotron emission stacking, and with cosmological simulations if primordial magnetic fields are amplified by astrophysical source field seeding. The measurement of an RRM_{0,f} supports the presence of diffuse baryonic gas in filaments. We also estimated a conservative upper limit of the filament magnetic turbulence of σ_{ RRM_{0,f}} =0.039 \pm 0.001 rad m^{-2}, concluding that the ordered magnetic field component dominates in filaments. △ Less

Submitted 9 February, 2022; originally announced February 2022.

Comments: Accepted for publication on MNRAS

Quantifying Uncertainty in Deep Learning Approaches to Radio Galaxy Classification

Authors: Devina Mohan, Anna M. M. Scaife, Fiona Porter, Mike Walmsley, Micah Bowles

Abstract: In this work we use variational inference to quantify the degree of uncertainty in deep learning model predictions of radio galaxy classification. We show that the level of model posterior variance for individual test samples is correlated with human uncertainty when labelling radio galaxies. We explore the model performance and uncertainty calibration for different weight priors and suggest that… ▽ More In this work we use variational inference to quantify the degree of uncertainty in deep learning model predictions of radio galaxy classification. We show that the level of model posterior variance for individual test samples is correlated with human uncertainty when labelling radio galaxies. We explore the model performance and uncertainty calibration for different weight priors and suggest that a sparse prior produces more well-calibrated uncertainty estimates. Using the posterior distributions for individual weights, we demonstrate that we can prune 30% of the fully-connected layer weights without significant loss of performance by removing the weights with the lowest signal-to-noise ratio. A larger degree of pruning can be achieved using a Fisher information based ranking, but both pruning methods affect the uncertainty calibration for Fanaroff-Riley type I and type II radio galaxies differently. Like other work in this field, we experience a cold posterior effect, whereby the posterior must be down-weighted to achieve good predictive performance. We examine whether adapting the cost function to accommodate model misspecification can compensate for this effect, but find that it does not make a significant difference. We also examine the effect of principled data augmentation and find that this improves upon the baseline but also does not compensate for the observed effect. We interpret this as the cold posterior effect being due to the overly effective curation of our training sample leading to likelihood misspecification, and raise this as a potential issue for Bayesian deep learning approaches to radio galaxy classification in future. △ Less

Submitted 24 January, 2022; v1 submitted 4 January, 2022; originally announced January 2022.

Comments: accepted by MNRAS

Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning

Authors: H. Tang, A. M. M. Scaife, O. I. Wong, S. S. Shabala

Abstract: In this work, we explore the potential of multi-domain multi-branch convolutional neural networks (CNNs) for identifying comparatively rare giant radio galaxies from large volumes of survey data, such as those expected for new-generation radio telescopes like the SKA and its precursors. The approach presented here allows models to learn jointly from multiple survey inputs, in this case NVSS and FI… ▽ More In this work, we explore the potential of multi-domain multi-branch convolutional neural networks (CNNs) for identifying comparatively rare giant radio galaxies from large volumes of survey data, such as those expected for new-generation radio telescopes like the SKA and its precursors. The approach presented here allows models to learn jointly from multiple survey inputs, in this case NVSS and FIRST, as well as incorporating numerical redshift information. We find that the inclusion of multi-resolution survey data results in correction of 39% of the misclassifications seen from equivalent single domain networks for the classification problem considered in this work. We also show that the inclusion of redshift information can moderately improve the classification of giant radio galaxies. △ Less

Submitted 7 December, 2021; originally announced December 2021.

Comments: 22 pages, 11 figures, accepted by MNRAS

Quantum Machine Learning for Radio Astronomy

Authors: Mohammad Kordzanganeh, Aydin Utting, Anna Scaife

Abstract: In this work we introduce a novel approach to the pulsar classification problem in time-domain radio astronomy using a Born machine, often referred to as a quantum neural network. Using a single-qubit architecture, we show that the pulsar classification problem maps well to the Bloch sphere and that comparable accuracies to more classical machine learning approaches are achievable. We introduce a… ▽ More In this work we introduce a novel approach to the pulsar classification problem in time-domain radio astronomy using a Born machine, often referred to as a quantum neural network. Using a single-qubit architecture, we show that the pulsar classification problem maps well to the Bloch sphere and that comparable accuracies to more classical machine learning approaches are achievable. We introduce a novel single-qubit encoding for the pulsar data used in this work and show that this performs comparably to a multi-qubit QAOA encoding. △ Less

Submitted 9 December, 2021; v1 submitted 5 December, 2021; originally announced December 2021.

Comments: Accepted in: Fourth Workshop on Machine Learning and the Physical Sciences (35th Conference on Neural Information Processing Systems; NeurIPS2021); final version

Weight Pruning and Uncertainty in Radio Galaxy Classification

Authors: Devina Mohan, Anna Scaife

Abstract: In this work we use variational inference to quantify the degree of epistemic uncertainty in model predictions of radio galaxy classification and show that the level of model posterior variance for individual test samples is correlated with human uncertainty when labelling radio galaxies. We explore the model performance and uncertainty calibration for a variety of different weight priors and sugg… ▽ More In this work we use variational inference to quantify the degree of epistemic uncertainty in model predictions of radio galaxy classification and show that the level of model posterior variance for individual test samples is correlated with human uncertainty when labelling radio galaxies. We explore the model performance and uncertainty calibration for a variety of different weight priors and suggest that a sparse prior produces more well-calibrated uncertainty estimates. Using the posterior distributions for individual weights, we show that signal-to-noise ratio (SNR) ranking allows pruning of the fully-connected layers to the level of 30% without significant loss of performance, and that this pruning increases the predictive uncertainty in the model. Finally we show that, like other work in this field, we experience a cold posterior effect. We examine whether adapting the cost function in our model to accommodate model misspecification can compensate for this effect, but find that it does not make a significant difference. We also examine the effect of principled data augmentation and find that it improves upon the baseline but does not compensate for the observed effect fully. We interpret this as the cold posterior effect being due to the overly effective curation of our training sample leading to likelihood misspecification, and raise this as a potential issue for Bayesian deep learning approaches to radio galaxy classification in future. △ Less

Submitted 29 November, 2021; v1 submitted 23 November, 2021; originally announced November 2021.

Comments: Accepted in: Fourth Workshop on Machine Learning and the Physical Sciences (35th Conference on Neural Information Processing Systems; NeurIPS2021); final version (corrected typo)

E(2) Equivariant Self-Attention for Radio Astronomy

Authors: Micah Bowles, Matthew Bromley, Max Allen, Anna Scaife

Abstract: In this work we introduce group-equivariant self-attention models to address the problem of explainable radio galaxy classification in astronomy. We evaluate various orders of both cyclic and dihedral equivariance, and show that including equivariance as a prior both reduces the number of epochs required to fit the data and results in improved performance. We highlight the benefits of equivariance… ▽ More In this work we introduce group-equivariant self-attention models to address the problem of explainable radio galaxy classification in astronomy. We evaluate various orders of both cyclic and dihedral equivariance, and show that including equivariance as a prior both reduces the number of epochs required to fit the data and results in improved performance. We highlight the benefits of equivariance when using self-attention as an explainable model and illustrate how equivariant models statistically attend the same features in their classifications as human astronomers. △ Less

Submitted 30 November, 2021; v1 submitted 8 November, 2021; originally announced November 2021.

Comments: Accepted in: Fourth Workshop on Machine Learning and the Physical Sciences (35th Conference on Neural Information Processing Systems; NeurIPS2021); final version; 7 pages, 3 figures

Can semi-supervised learning reduce the amount of manual labelling required for effective radio galaxy morphology classification?

Authors: Inigo V. Slijepcevic, Anna M. M. Scaife

Abstract: In this work, we examine the robustness of state-of-the-art semi-supervised learning (SSL) algorithms when applied to morphological classification in modern radio astronomy. We test whether SSL can achieve performance comparable to the current supervised state of the art when using many fewer labelled data points and if these results generalise to using truly unlabelled data. We find that although… ▽ More In this work, we examine the robustness of state-of-the-art semi-supervised learning (SSL) algorithms when applied to morphological classification in modern radio astronomy. We test whether SSL can achieve performance comparable to the current supervised state of the art when using many fewer labelled data points and if these results generalise to using truly unlabelled data. We find that although SSL provides additional regularisation, its performance degrades rapidly when using very few labels, and that using truly unlabelled data leads to a significant drop in performance. △ Less

Submitted 1 February, 2022; v1 submitted 8 November, 2021; originally announced November 2021.

Comments: Accepted in: Fourth Workshop on Machine Learning and the Physical Sciences (35th Conference on Neural Information Processing Systems; NeurIPS2021); final version

Practical Galaxy Morphology Tools from Deep Supervised Representation Learning

Authors: Mike Walmsley, Anna M. M. Scaife, Chris Lintott, Michelle Lochner, Verlon Etsebeth, Tobias Géron, Hugh Dickinson, Lucy Fortson, Sandor Kruk, Karen L. Masters, Kameswara Bharadwaj Mantha, Brooke D. Simmons

Abstract: Astronomers have typically set out to solve supervised machine learning problems by creating their own representations from scratch. We show that deep learning models trained to answer every Galaxy Zoo DECaLS question learn meaningful semantic representations of galaxies that are useful for new tasks on which the models were never trained. We exploit these representations to outperform several rec… ▽ More Astronomers have typically set out to solve supervised machine learning problems by creating their own representations from scratch. We show that deep learning models trained to answer every Galaxy Zoo DECaLS question learn meaningful semantic representations of galaxies that are useful for new tasks on which the models were never trained. We exploit these representations to outperform several recent approaches at practical tasks crucial for investigating large galaxy samples. The first task is identifying galaxies of similar morphology to a query galaxy. Given a single galaxy assigned a free text tag by humans (e.g. "#diffuse"), we can find galaxies matching that tag for most tags. The second task is identifying the most interesting anomalies to a particular researcher. Our approach is 100% accurate at identifying the most interesting 100 anomalies (as judged by Galaxy Zoo 2 volunteers). The third task is adapting a model to solve a new task using only a small number of newly-labelled galaxies. Models fine-tuned from our representation are better able to identify ring galaxies than models fine-tuned from terrestrial images (ImageNet) or trained from scratch. We solve each task with very few new labels; either one (for the similarity search) or several hundred (for anomaly detection or fine-tuning). This challenges the longstanding view that deep supervised methods require new large labelled datasets for practical use in astronomy. To help the community benefit from our pretrained models, we release our fine-tuning code Zoobot. Zoobot is accessible to researchers with no prior experience in deep learning. △ Less

Submitted 8 June, 2022; v1 submitted 25 October, 2021; originally announced October 2021.

Comments: 20 pages plus appendix. Accepted to MNRAS (open-access DOI below). Code, documentation, pretrained models: https://github.com/mwalmsley/zoobot (PyTorch and TensorFlow)

Journal ref: MNRAS Volume 513, Issue 2, June 2022, Pages 1581-1599

The JCMT BISTRO Survey: An 850/450$μ$m Polarization Study of NGC 2071IR in OrionB

Authors: A-Ran Lyo, Jongsoo Kim, Sarah Sadavoy, Doug Johnstone, David Berry, Kate Pattle, Woo** Kwon, Pierre Bastien, Takashi Onaka, James Di Francesco, Ji-Hyun Kang, Ray Furuya, Charles L. H. Hull, Motohide Tamura, Patrick M. Koch, Derek Ward-Thompson, Tetsuo Hasegawa, Thiem Hoang, Doris Arzoumanian, Chang Won Lee, Chin-Fei Lee, Do-Young Byun, Florian Kirchschlager, Yasuo Doi, Kee-Tae Kim , et al. (121 additional authors not shown)

Abstract: We present the results of simultaneous 450 $μ$m and 850 $μ$m polarization observations toward the massive star forming region NGC 2071IR, a target of the BISTRO (B-fields in Star-Forming Region Observations) Survey, using the POL-2 polarimeter and SCUBA-2 camera mounted on the James Clerk Maxwell Telescope. We find a pinched magnetic field morphology in the central dense core region, which could b… ▽ More We present the results of simultaneous 450 $μ$m and 850 $μ$m polarization observations toward the massive star forming region NGC 2071IR, a target of the BISTRO (B-fields in Star-Forming Region Observations) Survey, using the POL-2 polarimeter and SCUBA-2 camera mounted on the James Clerk Maxwell Telescope. We find a pinched magnetic field morphology in the central dense core region, which could be due to a rotating toroidal disk-like structure and a bipolar outflow originating from the central young stellar object, IRS 3. Using the modified Davis-Chandrasekhar-Fermi method, we obtain a plane-of-sky magnetic field strength of 563$\pm$421 $μ$G in the central $\sim$0.12 pc region from 850 $μ$m polarization data. The corresponding magnetic energy density of 2.04$\times$10$^{-8}$ erg cm$^{-3}$ is comparable to the turbulent and gravitational energy densities in the region. We find that the magnetic field direction is very well aligned with the whole of the IRS 3 bipolar outflow structure. We find that the median value of polarization fractions, 3.0 \%, at 450 $μ$m in the central 3 arcminute region, which is larger than the median value of 1.2 \% at 850 $μ$m. The trend could be due to the better alignment of warmer dust in the strong radiation environment. We also find that polarization fractions decrease with intensity at both wavelengths, with slopes, determined by fitting a Rician noise model, of $0.59 \pm 0.03$ at 450 $μ$m and $0.36 \pm 0.04$ at 850 $μ$m, respectively. We think that the shallow slope at 850 $μ$m is due to grain alignment at the center being assisted by strong radiation from the central young stellar objects. △ Less

Submitted 28 September, 2021; originally announced September 2021.

Comments: 17 pages, 11 figures, 4 tables

Journal ref: The Astrophysical Journal, 2021, 918, 85

Square Kilometre Array : Processing Voluminous MeerKAT Data on IRIS

Authors: Priyaa Thavasimani, Anna Scaife

Abstract: Processing astronomical data often comes with huge challenges with regards to data management as well as data processing. MeerKAT telescope is one of the precursor telescopes of the World's largest observatory - Square Kilometre Array. So far, MeerKAT data was processed using the South African computing facility i.e. IDIA, and exploited to make ground-breaking discoveries. However, to process Meer… ▽ More Processing astronomical data often comes with huge challenges with regards to data management as well as data processing. MeerKAT telescope is one of the precursor telescopes of the World's largest observatory - Square Kilometre Array. So far, MeerKAT data was processed using the South African computing facility i.e. IDIA, and exploited to make ground-breaking discoveries. However, to process MeerKAT data on UK's IRIS computing facility requires new implementation of the MeerKAT pipeline. This paper focuses on how to transfer MeerKAT data from the South African site to UK's IRIS systems for processing. We discuss about our RapifXfer Data transfer framework for transferring the MeerKAT data from South Africa to the UK, and the MeerKAT job processing framework pertaining to the UK's IRIS resources. △ Less

Submitted 15 April, 2023; v1 submitted 30 May, 2021; originally announced May 2021.

Comments: arXiv admin note: This article has been withdrawn due to an unresolvable internal author dispute

ACM Class: H.3.2; H.2.7; B.3.4; C.1.4

LOFAR imaging of Cygnus A -- Direct detection of a turnover in the hotspot radio spectra

Authors: J. P. McKean, L. E. H. Godfrey, S. Vegetti, M. W. Wise, R. Morganti, M. J. Hardcastle, D. Rafferty, J. Anderson, I. M. Avruch, R. Beck, M. E. Bell, I. van Bemmel, M. J. Bentum, G. Bernardi, P. Best, R. Blaauw, A. Bonafede, F. Breitling, J. W. Broderick, M. Bruggen, L. Cerrigone, B. Ciardi, F. de Gasperin, A. Deller, S. Duscha , et al. (53 additional authors not shown)

Abstract: The low-frequency radio spectra of the hotspots within powerful radio galaxies can provide valuable information about the physical processes operating at the site of the jet termination. These processes are responsible for the dissipation of jet kinetic energy, particle acceleration, and magnetic-field generation. Here we report new observations of the powerful radio galaxy Cygnus A using the Low… ▽ More The low-frequency radio spectra of the hotspots within powerful radio galaxies can provide valuable information about the physical processes operating at the site of the jet termination. These processes are responsible for the dissipation of jet kinetic energy, particle acceleration, and magnetic-field generation. Here we report new observations of the powerful radio galaxy Cygnus A using the Low Frequency Array (LOFAR) between 109 and 183 MHz, at an angular resolution of ~3.5 arcsec. The radio emission of the lobes is found to have a complex spectral index distribution, with a spectral steepening found towards the centre of the source. For the first time, a turnover in the radio spectrum of the two main hotspots of Cygnus A has been directly observed. By combining our LOFAR imaging with data from the Very Large Array at higher frequencies, we show that the very rapid turnover in the hotspot spectra cannot be explained by a low-energy cut-off in the electron energy distribution, as has been previously suggested. Thermal (free-free) absorption or synchrotron self absorption models are able to describe the low-frequency spectral shape of the hotspots, however, as with previous studies, we find that the implied model parameters are unlikely, and interpreting the spectra of the hotspots remains problematic. △ Less

Submitted 31 March, 2021; originally announced March 2021.

Comments: 9 pages, 3 figures, accepted for publication in MNRAS

Journal ref: MNRAS, 463, 3143 (2016)

Revealing the diverse magnetic field morphologies in Taurus dense cores with sensitive sub-millimeter polarimetry

Authors: Chakali Eswaraiah, Di Li, Ray S. Furuya, Tetsuo Hasegawa, Derek Ward-Thompson, Ke** Qiu, Nagayoshi Ohashi, Kate Pattle, Sarah Sadavoy, Charles L. H. Hull, David Berry, Yasuo Doi, Tao-Chung Ching, Shih-** Lai, Jia-Wei Wang, Patrick M. Koch, Jungmi Kwon, Woo** Kwon, Pierre Bastien, Doris Arzoumanian, Simon Coudé, Archana Soam, Lapo Fanciullo, Hsi-Wei Yen, Junhao Liu , et al. (120 additional authors not shown)

Abstract: We have obtained sensitive dust continuum polarization observations at 850 $μ$m in the B213 region of Taurus using POL-2 on SCUBA-2 at the James Clerk Maxwell Telescope (JCMT), as part of the BISTRO (B-fields in STar-forming Region Observations) survey. These observations allow us to probe magnetic field (B-field) at high spatial resolution ($\sim$2000 au or $\sim$0.01 pc at 140 pc) in two protost… ▽ More We have obtained sensitive dust continuum polarization observations at 850 $μ$m in the B213 region of Taurus using POL-2 on SCUBA-2 at the James Clerk Maxwell Telescope (JCMT), as part of the BISTRO (B-fields in STar-forming Region Observations) survey. These observations allow us to probe magnetic field (B-field) at high spatial resolution ($\sim$2000 au or $\sim$0.01 pc at 140 pc) in two protostellar cores (K04166 and K04169) and one prestellar core (Miz-8b) that lie within the B213 filament. Using the Davis-Chandrasekhar-Fermi method, we estimate the B-field strengths in K04166, K04169, and Miz-8b to be 38$\pm$14 $μ$G, 44$\pm$16 $μ$G, and 12$\pm$5 $μ$G, respectively. These cores show distinct mean B-field orientations. B-field in K04166 is well ordered and aligned parallel to the orientations of the core minor axis, outflows, core rotation axis, and large-scale uniform B-field, in accordance with magnetically regulated star formation via ambipolar diffusion taking place in K04166. B-field in K04169 is found to be ordered but oriented nearly perpendicular to the core minor axis and large-scale B-field, and not well-correlated with other axes. In contrast, Miz-8b exhibits disordered B-field which show no preferred alignment with the core minor axis or large-scale field. We found that only one core, K04166, retains a memory of the large-scale uniform B-field. The other two cores, K04169 and Miz-8b, are decoupled from the large-scale field. Such a complex B-field configuration could be caused by gas inflow onto the filament, even in the presence of a substantial magnetic flux. △ Less

Submitted 3 March, 2021; originally announced March 2021.

Comments: 20 pages (10 are main), 5 figures (3 are main), and 3 tables (2 are main); Accepted for its publication in ApJL

Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks

Authors: Anna M. M. Scaife, Fiona Porter

Abstract: Weight sharing in convolutional neural networks (CNNs) ensures that their feature maps will be translation-equivariant. However, although conventional convolutions are equivariant to translation, they are not equivariant to other isometries of the input image data, such as rotation and reflection. For the classification of astronomical objects such as radio galaxies, which are expected statistical… ▽ More Weight sharing in convolutional neural networks (CNNs) ensures that their feature maps will be translation-equivariant. However, although conventional convolutions are equivariant to translation, they are not equivariant to other isometries of the input image data, such as rotation and reflection. For the classification of astronomical objects such as radio galaxies, which are expected statistically to be globally orientation invariant, this lack of dihedral equivariance means that a conventional CNN must learn explicitly to classify all rotated versions of a particular type of object individually. In this work we present the first application of group-equivariant convolutional neural networks to radio galaxy classification and explore their potential for reducing intra-class variability by preserving equivariance for the Euclidean group E(2), containing translations, rotations and reflections. For the radio galaxy classification problem considered here, we find that classification performance is modestly improved by the use of both cyclic and dihedral models without additional hyper-parameter tuning, and that a D16 equivariant model provides the best test performance. We use the Monte Carlo Dropout method as a Bayesian approximation to recover epistemic uncertainty as a function of image orientation and show that E(2)-equivariant models are able to reduce variations in model confidence as a function of rotation. △ Less

Submitted 18 February, 2021; v1 submitted 16 February, 2021; originally announced February 2021.

Comments: 11 pages, 6 figures, accepted MNRAS

Structured Variational Inference for Simulating Populations of Radio Galaxies

Authors: David J. Bastien, Anna M. M. Scaife, Hongming Tang, Micah Bowles, Fiona Porter

Abstract: We present a model for generating postage stamp images of synthetic Fanaroff-Riley Class I and Class II radio galaxies suitable for use in simulations of future radio surveys such as those being developed for the Square Kilometre Array. This model uses a fully-connected neural network to implement structured variational inference through a variational auto-encoder and decoder architecture. In orde… ▽ More We present a model for generating postage stamp images of synthetic Fanaroff-Riley Class I and Class II radio galaxies suitable for use in simulations of future radio surveys such as those being developed for the Square Kilometre Array. This model uses a fully-connected neural network to implement structured variational inference through a variational auto-encoder and decoder architecture. In order to optimise the dimensionality of the latent space for the auto-encoder we introduce the radio morphology inception score (RAMIS), a quantitative method for assessing the quality of generated images, and discuss in detail how data pre-processing choices can affect the value of this measure. We examine the 2-dimensional latent space of the VAEs and discuss how this can be used to control the generation of synthetic populations, whilst also cautioning how it may lead to biases when used for data augmentation. △ Less

Submitted 24 February, 2021; v1 submitted 1 February, 2021; originally announced February 2021.

Comments: 20 pages, 20 figures, accepted MNRAS

Gaussian Process Modelling for Improved Resolution in Faraday Depth Reconstruction

Authors: S. W. Ndiritu, A. M. M. Scaife, D. L. Tabb, M. Carcamo, J. Hanson

Abstract: The incomplete sampling of data in complex polarization measurements from radio telescopes negatively affects both the rotation measure (RM) transfer function and the Faraday depth spectra derived from these data. Such gaps in polarization data are mostly caused by flagging of radio frequency interference and their effects worsen as the percentage of missing data increases. In this paper we presen… ▽ More The incomplete sampling of data in complex polarization measurements from radio telescopes negatively affects both the rotation measure (RM) transfer function and the Faraday depth spectra derived from these data. Such gaps in polarization data are mostly caused by flagging of radio frequency interference and their effects worsen as the percentage of missing data increases. In this paper we present a novel method for inferring missing polarization data based on Gaussian processes (GPs). Gaussian processes are stochastic processes that enable us to encode prior knowledge in our models. They also provide a comprehensive way of incorporating and quantifying uncertainties in regression modelling. In addition to providing non-parametric model estimates for missing values, we also demonstrate that Gaussian process modelling can be used for recovering rotation measure values directly from complex polarization data, and that inferring missing polarization data using this probabilistic method improves the resolution of reconstructed Faraday depth spectra. △ Less

Submitted 18 January, 2021; originally announced January 2021.

Comments: 16 pages, 10 figures, submitted to MNRAS

Dust polarized emission observations of NGC 6334; BISTRO reveals the details of the complex but organized magnetic field structure of the high-mass star-forming hub-filament network

Authors: D. Arzoumanian, R. Furuya, T. Hasegawa, M. Tahani, S. Sadavoy, C. L. H. Hull, D. Johnstone, P. M. Koch, S. -i. Inutsuka, Y. Doi, T. Hoang, T. Onaka, K. Iwasaki, Y. Shimajiri, T. Inoue, N. Peretto, P. André, P. Bastien, D. Berry, H. -R. V. Chen, J. Di Francesco, C. Eswaraiah, L. Fanciullo, L. M. Fissel, J. Hwang , et al. (123 additional authors not shown)

Abstract: [Abridged] Filaments and hubs have received special attention recently thanks to studies showing their role in star formation. While the column density and velocity structures of both filaments and hubs have been studied, their magnetic fields (B-field) are not yet characterized. We aim to understand the role of the B-field in the dynamical evolution of the NGC 6334 hub-filament network. We presen… ▽ More [Abridged] Filaments and hubs have received special attention recently thanks to studies showing their role in star formation. While the column density and velocity structures of both filaments and hubs have been studied, their magnetic fields (B-field) are not yet characterized. We aim to understand the role of the B-field in the dynamical evolution of the NGC 6334 hub-filament network. We present new observations of the dust polarized emission at 850$μ$m towards NGC 6334 obtained with the JCMT/POL-2. We study the distribution and dispersion of the polarized intensity ($PI$), the polarization fraction ($PF$), and the B-field angle ($θ_{B}$). We derive the power spectrum of the intensity and $θ_{B}$ along the ridge crest. Our analyses show a complex B-field structure when observed over the whole region ($\sim10$ pc), however, at smaller scales ($\sim1$ pc), $θ_{B}$ varies coherently along the filaments. The observed power spectrum of $θ_{B}$ can be well represented with a power law function with a slope $-1.33\pm0.23$, which is $\sim20\%$ shallower than that of $I$. This result is compatible with the properties of simulated filaments and may indicate the processes at play in the formation of filaments. $θ_{B}$ rotates from being mostly perpendicular to the filament crests to mostly parallel as they merge with the hubs. This variation of $θ_{B}$ may be tracing local velocity flows of matter in-falling onto the hubs. Our analysis suggests a variation of the energy balance along the crests of these filaments, from magnetically critical/supercritical at their far ends to magnetically subcritical near the hubs. We detect an increase of $PF$ towards the high-column density star cluster-forming hubs that may result from the increase of grain alignment efficiency due to stellar radiation from the newborn stars. △ Less

Submitted 23 December, 2020; originally announced December 2020.

Comments: Accepted for publication in Astronomy & Astrophysics

Journal ref: A&A 647, A78 (2021)

Observations of magnetic fields surrounding LkH$α$ 101 taken by the BISTRO survey with JCMT-POL-2

Authors: Nguyen Bich Ngoc, Pham Ngoc Diep, Harriet Parsons, Kate Pattle, Thiem Hoang, Derek Ward-Thompson, Le Ngoc Tram, Charles L. H. Hull, Mehrnoosh Tahani, Ray Furuya, Pierre Bastien, Ke** Qiu, Tetsuo Hasegawa, Woo** Kwon, Yasuo Doi, Shih-** Lai, Simon Coude, David Berry, Tao-Chung Ching, Jihye Hwang, Archana Soam, Jia-Wei Wang, Doris Arzoumanian, Tyler L. Bourke, Do-Young Byun , et al. (124 additional authors not shown)

Abstract: We report the first high spatial resolution measurement of magnetic fields surrounding LkH$α$ 101, a part of the Auriga-California molecular cloud. The observations were taken with the POL-2 polarimeter on the James Clerk Maxwell Telescope within the framework of the B-fields In Star-forming Region Observations (BISTRO) survey. Observed polarization of thermal dust emission at 850 $μ$m is found to… ▽ More We report the first high spatial resolution measurement of magnetic fields surrounding LkH$α$ 101, a part of the Auriga-California molecular cloud. The observations were taken with the POL-2 polarimeter on the James Clerk Maxwell Telescope within the framework of the B-fields In Star-forming Region Observations (BISTRO) survey. Observed polarization of thermal dust emission at 850 $μ$m is found to be mostly associated with the red-shifted gas component of the cloud. The magnetic field displays a relatively complex morphology. Two variants of the Davis-Chandrasekhar-Fermi method, unsharp masking and structure function, are used to calculate the strength of magnetic fields in the plane of the sky, yielding a similar result of $B_{\rm POS}\sim 115$ $\mathrmμ$G. The mass-to-magnetic-flux ratio in critical value units, $λ\sim0.3$, is the smallest among the values obtained for other regions surveyed by POL-2. This implies that the LkH$α$ 101 region is sub-critical and the magnetic field is strong enough to prevent gravitational collapse. The inferred $δB/B_0\sim 0.3$ implies that the large scale component of the magnetic field dominates the turbulent one. The variation of the polarization fraction with total emission intensity can be fitted by a power-law with an index of $α=0.82\pm0.03$, which lies in the range previously reported for molecular clouds. We find that the polarization fraction decreases rapidly with proximity to the only early B star (LkH$α$ 101) in the region. The magnetic field tangling and the joint effect of grain alignment and rotational disruption by radiative torques are potential of explaining such a decreasing trend. △ Less

Submitted 8 December, 2020; originally announced December 2020.

Comments: 25 pages, 19 figures, Accepted for publication in The Astrophysical Journal

Attention-gating for improved radio galaxy classification

Authors: Micah Bowles, Anna M. M. Scaife, Fiona Porter, Hongming Tang, David J. Bastien

Abstract: In this work we introduce attention as a state of the art mechanism for classification of radio galaxies using convolutional neural networks. We present an attention-based model that performs on par with previous classifiers while using more than 50% fewer parameters than the next smallest classic CNN application in this field. We demonstrate quantitatively how the selection of normalisation and a… ▽ More In this work we introduce attention as a state of the art mechanism for classification of radio galaxies using convolutional neural networks. We present an attention-based model that performs on par with previous classifiers while using more than 50% fewer parameters than the next smallest classic CNN application in this field. We demonstrate quantitatively how the selection of normalisation and aggregation methods used in attention-gating can affect the output of individual models, and show that the resulting attention maps can be used to interpret the classification choices made by the model. We observe that the salient regions identified by the our model align well with the regions an expert human classifier would attend to make equivalent classifications. We show that while the selection of normalisation and aggregation may only minimally affect the performance of individual models, it can significantly affect the interpretability of the respective attention maps and by selecting a model which aligns well with how astronomers classify radio sources by eye, a user can employ the model in a more effective manner. △ Less

Submitted 1 February, 2021; v1 submitted 2 December, 2020; originally announced December 2020.

Comments: 18 pages, 16 figures, Published in MNRAS

Journal ref: MNRAS 501 (2021) 4579-4595

The great Kite in the sky: a LOFAR observation of the radio source in Abell 2626

Authors: A. Ignesti, T. Shimwell, G. Brunetti, M. Gitti, H. Intema, R. J. van Weeren, M. J. Hardcastle, A. O. Clarke, A. Botteon, G. Di Gennaro, M. Brüggen, I. Browne, S. Mandal, H. J. A. Röttgering, V. Cuciti, F. de Gasperin, R. Cassano, A. M. M. Scaife

Abstract: The radio source at the center of the galaxy cluster Abell 2626, also known as the Kite, stands out for its unique morphology composed of four, symmetric arcs. Previous studies have probed the properties of this source at different frequencies and its interplay with the surrounding thermal plasma, but the puzzle of its origin is still unsolved. We use new LOw Frequency ARray (LOFAR) observation fr… ▽ More The radio source at the center of the galaxy cluster Abell 2626, also known as the Kite, stands out for its unique morphology composed of four, symmetric arcs. Previous studies have probed the properties of this source at different frequencies and its interplay with the surrounding thermal plasma, but the puzzle of its origin is still unsolved. We use new LOw Frequency ARray (LOFAR) observation from the LOFAR Two-meter Sky Survey at 144 MHz to investigate the origin of the Kite.} We present a detailed analysis of the new radio data which we combined with archival radio and X-ray observations. We have produced a new, resolved spectral index map of the source with a resolution of 7$''$ and we studied the spatial correlation of radio and X-ray emission to investigate the interplay between thermal and non-thermal plasma. The new LOFAR data have changed our view of the Kite by discovering two steep-spectrum ($α<-1.5$) plumes of emission connected to the arcs. The spectral analysis shows, for the first time, a spatial trend of the spectrum along the arcs with evidence of curved synchrotron spectra and a spatial correlation with the X-ray surface brightness. On the basis of our results, we propose that the Kite was originally an X-shaped radio galaxy whose fossil radio plasma, after the end of the activity of the central active galactic nucleus, has been compressed due to motions of the thermal plasma in which it is encompassed. The interplay between the compression and advection of the fossil plasma, with the restarting of the nuclear activity of the central galaxy, could have enhanced the radio emission of the fossil plasma producing the arcs of the Kite. We present also the first, low-frequency observation of a jellyfish galaxy in the same field, in which we detect extended, low-frequency emission without a counterpart at higher frequencies. △ Less

Submitted 23 September, 2020; originally announced September 2020.

Comments: 10 pages, 7 figures. Accepted for publication on Astronomy & Astrophysics on 23/09/2020

Journal ref: A&A 643, A172 (2020)

Radio Galaxy Zoo: New Giant Radio Galaxies in the RGZ DR1catalogue

Authors: H. Tang, A. M. M. Scaife, O. I. Wong, A. D. Kapinska, L. Rudnick, S. S. Shabala, N. Seymour, R. P. Norris

Abstract: In this paper, we present the identification of five previously unknown giant radio galaxies (GRGs) using Data Release 1 of the Radio Galaxy Zoo citizen science project and a selection method appropriate to the training and validation of deep learning algorithms for new radio surveys. We associate one of these new GRGs with the brightest cluster galaxy (BCG) in the galaxy cluster GMBCG J251.67741+… ▽ More In this paper, we present the identification of five previously unknown giant radio galaxies (GRGs) using Data Release 1 of the Radio Galaxy Zoo citizen science project and a selection method appropriate to the training and validation of deep learning algorithms for new radio surveys. We associate one of these new GRGs with the brightest cluster galaxy (BCG) in the galaxy cluster GMBCG J251.67741+36.45295 and use literature data to identify a further 13 previously known GRGs as BCG candidates, increasing the number of known BCG GRGs by >60%. By examining local galaxy number densities for the number of all known BCG GRGs, we suggest that the existence of this growing number implies that GRGs are able to reside in the centers of rich ($\sim 10^{14}$ M$_{\odot}$) galaxy clusters and challenges the hypothesis that GRGs grow to such sizes only in locally under-dense environments. △ Less

Submitted 8 September, 2020; originally announced September 2020.

The JCMT BISTRO Survey: Magnetic Fields Associated with a Network of Filaments in NGC 1333

Authors: Yasuo Doi, Tetsuo Hasegawa, Ray S. Furuya, Simon Coudé, Charles L. H. Hull, Doris Arzoumanian, Pierre Bastien, Michael Chun-Yuan Chen, James di Francesco, Rachel Friesen, Martin Houde, Shu-ichiro Inutsuka, Steve Mairs, Masafumi Matsumura, Takashi Onaka, Sarah Sadavoy, Yoshito Shimajiri, Mehrnoosh Tahani, Kohji Tomisaka, Chakali Eswaraiah, Patrick M. Koch, Kate Pattle, Chang Won Lee, Motohide Tamura, David Berry , et al. (113 additional authors not shown)

Abstract: We present new observations of the active star-formation region NGC 1333 in the Perseus molecular cloud complex from the James Clerk Maxwell Telescope B-Fields In Star-forming Region Observations (BISTRO) survey with the POL-2 instrument. The BISTRO data cover the entire NGC 1333 complex (~1.5 pc x 2 pc) at 0.02 pc resolution and spatially resolve the polarized emission from individual filamentary… ▽ More We present new observations of the active star-formation region NGC 1333 in the Perseus molecular cloud complex from the James Clerk Maxwell Telescope B-Fields In Star-forming Region Observations (BISTRO) survey with the POL-2 instrument. The BISTRO data cover the entire NGC 1333 complex (~1.5 pc x 2 pc) at 0.02 pc resolution and spatially resolve the polarized emission from individual filamentary structures for the first time. The inferred magnetic field structure is complex as a whole, with each individual filament aligned at different position angles relative to the local field orientation. We combine the BISTRO data with low- and high- resolution data derived from Planck and interferometers to study the multiscale magnetic field structure in this region. The magnetic field morphology drastically changes below a scale of ~1 pc and remains continuous from the scales of filaments (~0.1 pc) to that of protostellar envelopes (~0.005 pc or ~1000 au). Finally, we construct simple models in which we assume that the magnetic field is always perpendicular to the long axis of the filaments. We demonstrate that the observed variation of the relative orientation between the filament axes and the magnetic field angles are well reproduced by this model, taking into account the projection effects of the magnetic field and filaments relative to the plane of the sky. These projection effects may explain the apparent complexity of the magnetic field structure observed at the resolution of BISTRO data toward the filament network. △ Less

Submitted 21 July, 2020; v1 submitted 30 June, 2020; originally announced July 2020.

Comments: 33 pages, 21 figures, accepted for publication in ApJ

An updated estimate of the cosmic radio background and implications for ultra-high-energy photon propagation

Authors: I. C. Niţu, H. T. J. Bevins, J. D. Bray, A. M. M. Scaife

Abstract: We present an updated estimate of the cosmic radio background (CRB) and the corresponding attenuation lengths for ultra-high energy photons. This new estimate provides associated uncertainties as a function of frequency derived from observational constraints on key physical parameters. We also present the expected variation in the spectrum of the CRB as a function of these parameters, as well as a… ▽ More We present an updated estimate of the cosmic radio background (CRB) and the corresponding attenuation lengths for ultra-high energy photons. This new estimate provides associated uncertainties as a function of frequency derived from observational constraints on key physical parameters. We also present the expected variation in the spectrum of the CRB as a function of these parameters, as well as accounting for the expected variation in spectral index among the population of radio galaxies. The new estimate presented in this work shows better agreement with observational constraints from radio source-count measurements than previous calculations. In the energy regime where we expect cosmogenic photons dominantly attenuated by the CRB, our calculation of the attenuation length differs from previous estimates by a factor of up to 3, depending on energy and the specific model for comparison. These results imply a decrease in the expected number of cosmogenic photons with energies $\sim 10^{19}-10^{20}$ eV. △ Less

Submitted 28 April, 2020; originally announced April 2020.

Low-frequency observations of the Giant Radio Galaxy NGC 6251

Authors: T. M. Cantwell, J. D. Bray, J. H. Croston, A. M. M. Scaife, D. D. Mulcahy, P. N. Best, M. Bruggen, G. Brunetti, J. R. Callingham, A. O. Clarke, M. J. Hardcastle, J. J. Harwood, G. Heald, V. Heesen, M. Iacobelli, M. Jamrozy, R. Morganti, E. Orru, S. P. O'Sullivan, C. J. Riseley, H. J. A. Rottgering, A. Shulevski, S. S. Sridhar, C. Tasse, C. L. Van Eck

Abstract: We present LOFAR observations at 150 MHz of the borderline FRI/FRII giant radio galaxy NGC 6251. This paper presents the most sensitive and highest-resolution images of NGC 6251 at these frequencies to date, revealing for the first time a low-surface-brightness extension to the northern lobe, and a possible backflow associated with the southern lobe. The integrated spectra of components of NGC 625… ▽ More We present LOFAR observations at 150 MHz of the borderline FRI/FRII giant radio galaxy NGC 6251. This paper presents the most sensitive and highest-resolution images of NGC 6251 at these frequencies to date, revealing for the first time a low-surface-brightness extension to the northern lobe, and a possible backflow associated with the southern lobe. The integrated spectra of components of NGC 6251 are consistent with previous measurements at higher frequencies, similar to results from other LOFAR studies of nearby radio galaxies. We find the outer structures of NGC 6251 to be either at equipartition or slightly electron dominated, similar to those of FRII sources rather than FRIs; but this conclusion remains tentative because of uncertainties associated with the geometry and the extrapolation of X-ray measurements to determine the external pressure distribution on the scale of the outer lobes. We place lower limits on the ages of the extension of the northern lobe and the backflow of the southern lobe of $t \gtrsim 250$ Myr and $t \gtrsim 210$ Myr respectively. We present the first detection of polarisation at 150 MHz in NGC 6251. Taking advantage of the high Faraday resolution of LOFAR, we place an upper limit on the magnetic field in the group of $B < 0.2 (Λ_B / 10 {\rm kpc})^{-0.5} μ$G for a coherence scale of $Λ_B < 60 {\rm kpc}$ and $B < 13 μ$G for $Λ_B = 240$ kpc. △ Less

Submitted 23 April, 2020; originally announced April 2020.

Comments: 19 pages, 17 figures; accepted by MNRAS

Identifying galaxies, quasars, and stars with machine learning: A new catalogue of classifications for 111 million SDSS sources without spectra

Authors: A. O. Clarke, A. M. M. Scaife, R. Greenhalgh, V. Griguta

Abstract: We used 3.1 million spectroscopically labelled sources from the Sloan Digital Sky Survey (SDSS) to train an optimised random forest classifier using photometry from the SDSS and the Widefield Infrared Survey Explorer (WISE). We applied this machine learning model to 111 million previously unlabelled sources from the SDSS photometric catalogue which did not have existing spectroscopic observations.… ▽ More We used 3.1 million spectroscopically labelled sources from the Sloan Digital Sky Survey (SDSS) to train an optimised random forest classifier using photometry from the SDSS and the Widefield Infrared Survey Explorer (WISE). We applied this machine learning model to 111 million previously unlabelled sources from the SDSS photometric catalogue which did not have existing spectroscopic observations. Our new catalogue contains 50.4 million galaxies, 2.1 million quasars, and 58.8 million stars. We provide individual classification probabilities for each source, with 6.7 million galaxies (13%), 0.33 million quasars (15%), and 41.3 million stars (70%) having classification probabilities greater than 0.99; and 35.1 million galaxies (70%), 0.72 million quasars (34%), and 54.7 million stars (93%) having classification probabilities greater than 0.9. Precision, Recall, and F1 score were determined as a function of selected features and magnitude error. We investigate the effect of class imbalance on our machine learning model and discuss the implications of transfer learning for populations of sources at fainter magnitudes than the training set. We used a non-linear dimension reduction technique, Uniform Manifold Approximation and Projection (UMAP), in unsupervised, semi-supervised, and fully-supervised schemes to visualise the separation of galaxies, quasars, and stars in a two-dimensional space. When applying this algorithm to the 111 million sources without spectra, it is in strong agreement with the class labels applied by our random forest model. △ Less

Submitted 21 May, 2020; v1 submitted 24 September, 2019; originally announced September 2019.

Comments: Accepted for publication in A&A

Journal ref: A&A 639, A84 (2020)

Cleaning radio interferometric images using a spherical wavelet decomposition

Authors: Chris J. Skipper, Anna M. M. Scaife, Jason D. McEwen

Abstract: The deconvolution, or cleaning, of radio interferometric images often involves computing model visibilities from a list of clean components, in order that the contribution from the model can be subtracted from the observed visibilities. This step is normally performed using a forward fast Fourier transform (FFT), followed by a 'degridding' step that interpolates over the uv plane to construct the… ▽ More The deconvolution, or cleaning, of radio interferometric images often involves computing model visibilities from a list of clean components, in order that the contribution from the model can be subtracted from the observed visibilities. This step is normally performed using a forward fast Fourier transform (FFT), followed by a 'degridding' step that interpolates over the uv plane to construct the model visibilities. An alternative approach is to calculate the model visibilities directly by summing over all the members of the clean component list, which is a more accurate method that can also be much slower. However, if the clean components are used to construct a model image on the surface of the celestial sphere then the model visibilities can be generated directly from the wavelet coefficients, and the sparsity of the model means that most of these coefficients are zero, and can be ignored. We have constructed a prototype imager that uses a spherical-wavelet representation of the model image to generate model visibilities during each major cycle, and find empirically that the execution time scales with the wavelet resolution level, J, as O(1.07 J), and with the number of distinct clean components, N_C, as O(N_C). The prototype organises the wavelet coefficients into a tree structure, and does not store or process the zero wavelet coefficients. △ Less

Submitted 9 September, 2019; originally announced September 2019.

Comments: 12 pages, 12 figures