31 pages, 16 figures, accepted in the ApJ
We consider the flare prediction problem that distinguishes flare-imminent active regions that produce an M- or X-class flare in the future 24 hours, from quiet active regions that do not produce any flare within $\pm 24$ hours. Using line-of-sight magnetograms and parameters of active regions in two data products covering Solar Cycle 23 and 24, we train and evaluate two deep learning algorithms -- CNN and LSTM -- and their stacking ensembles. The decisions of CNN are explained using visual attribution methods. We have the following three main findings. (1) LSTM trained on data from two solar cycles achieves significantly higher True Skill Scores (TSS) than that trained on data from a single solar cycle with a confidence level of at least 0.95. (2) On data from Solar Cycle 23, a stacking ensemble that combines predictions from LSTM and CNN using the TSS criterion achieves significantly higher TSS than the "select-best" strategy with a confidence level of at least 0.95. (3) A visual attribution method called Integrated Gradients is able to attribute the CNN's predictions of flares to the emerging magnetic flux in the active region. It also reveals a limitation of CNN as a flare prediction method using line-of-sight magnetograms: it treats the polarity artifact of line-of-sight magnetograms as positive evidence of flares.
17 pages,10 figures
We assess the dark matter halo masses of luminous AGNs over the redshift range 0.2 to 1.2 using galaxy-galaxy lensing based on imaging data from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). We measure the weak lensing signal of a sample of 8882 AGNs constructed using HSC and WISE photometry. The lensing detection around AGNs has a signal-to-noise ratio of 15. As expected, we find that the lensing mass profile is consistent with that of massive galaxies ($M_{*}\sim 10.8~M_\odot$). Surprisingly, the lensing signal remains unchanged when the AGN sample is split into low and high stellar mass hosts. Specifically, we find that the excess surface density (ESD) of AGNs, residing in galaxies with high stellar masses, significantly differs from that of the control sample. We further fit a halo occupation distribution model to the data to infer the posterior distribution of parameters including the average halo mass. We find that the characteristic halo mass of the full AGN population lies near the knee ($\rm log(M_h/h^{-1}M_{\odot})=12.0$) of the stellar-to-halo mass relation (SHMR). Illustrative of the results given above, the halo masses of AGNs residing in host galaxies with high stellar masses (i.e., above the knee of the SHMR) falls below the calibrated SHMR while the halo mass of the low stellar mass sample is more consistent with the established SHMR. These results indicate that massive halos with higher clustering bias tend to suppress AGN activity, probably due to the lack of available gas.
submitted to JATIS
Starshades are a leading technology to enable the direct detection and spectroscopic characterization of Earth-like exoplanets. To keep the starshade and telescope aligned over large separations, reliable sensing of the peak of the diffracted light of the occluded star is required. Current techniques rely on image matching or model fitting, both of which put substantial computational burdens on resource-limited spacecraft computers. We present a lightweight image processing method based on a convolutional neural network paired with a simulation-based inference technique to estimate the position of the spot of Arago and its uncertainty. The method achieves an accuracy of a few centimeters across the entire pupil plane, while only requiring 1.6 MB in stored data structures and 5.3 MFLOPs (million floating point operations) per image at test time. By deploying our method at the Princeton Starshade Testbed, we demonstrate that the neural network can be trained on simulated images and used on real images, and that it can successfully be integrated in the control system for closed-loop formation flying.
20 pages, 13 figures; Accepted for publications in MNRAS
Submitted to ApJ
7 pages, 7 Figures, accepted as a Letter by Astronomy Astrophysics, March 28 2022
36 pages, 12 figures
Accepted for publication in Astronomy and astrophysics
24 pages, 15 figures, accepted for publication in ApJ
22 pages, 6 figures, 4 tables, submitted to the AAS Journals
28 pages, 13 figures. Accepted for publication in ApJS. Comments welcomed
accepted for publication in Astronomy & Astrophysics
25 pages, 14 figures, submitted to AAS Journals
16 pages, 10 figures, 9 tables. Accepted for publication in Monthly Notices of the Royal Astronomical Society (MNRAS)
16 pages, 4 figures, Accepted for publication in ApJ
11 pages, 6 figures
17 pages, 7 figures, 6 tables. Code available at this https URL
19 pages Accepted for publication on A&A
9 pages, 7 figures, accepted for publication in PoP
8 pages
13 pages, 10 figures, 6 tables. Accepted in MNRAS
10 pages, 8 figures
Accepted by ApJ, 9 pages, 7 figures
8 pages, 9 figures
23 pages, 8 figures
Submit to apj. Welcome for comments
14 pages, 8 figures, submitted to JKAS
13 pages, 7 figures, published in MNRAS
submitted to MNRAS, 1st report received: under revision Have partially addressed referee's concerns, namely that the model predicts high dust masses and redder bright galaxies than expected, by discussing this aspect around the relevant results. Work is being carried out to present a clearer parameter exploration of the dust model
20 pages, 11 figures. To be published in PSJ. See this http URL for an executable version of this paper
17 pages, 22 figures, accepted in MNRAS
25 pages, 11 figures, 3 tables, submitted
14 pages, 11 figures, MNRAS accepted
15 pages, 10 figures. Submitted to MNRAS
Accepted for publication in ApJ, 28 Pages, 11 figures, 2 Tables (Appendix: 5 Figures, 1 Table)
Accepted to PASA. 27 figures. Data to be made available under the DOI 10.5281/zenodo.5347875 at the time of publication
To appear in PNAS
An updated rendering of arXiv:1810.03514 with an extra author
26 pages, 14 figures, accepted for ApJ
Accepted for publication in MNRAS, 20 pages, 17 figures
Submitted to ApJ; 27 pages, 13 figures, 4 tables
15 pages, 4 figures, 2 tables
9 pages, 5 figures, accepted for publication in A&A
15 pages, 3 captioned figures. Comments are welcome
4 pages
8 pages, 5 figures
To appear in the IEEE Proceedings of the Conference on Computer Vision and Pattern Recognition (CVPR), 2022. Supplemental material including accompanying pdf, code, and video highlight can be found in the project page: this http URL
43 pages, 40 figure multipanels, 2 tables
32 pages, 3 figures
38 pages, 4 figures
4 pages + 3 figures. Comments are welcome
14 pages, 3 figures, comments welcome