skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Cleaning our own dust: simulating and separating galactic dust foregrounds with neural networks
ABSTRACT Separating galactic foreground emission from maps of the cosmic microwave background (CMB) and quantifying the uncertainty in the CMB maps due to errors in foreground separation are important for avoiding biases in scientific conclusions. Our ability to quantify such uncertainty is limited by our lack of a model for the statistical distribution of the foreground emission. Here, we use a deep convolutional generative adversarial network (DCGAN) to create an effective non-Gaussian statistical model for intensity of emission by interstellar dust. For training data we use a set of dust maps inferred from observations by the Planck satellite. A DCGAN is uniquely suited for such unsupervised learning tasks as it can learn to model a complex non-Gaussian distribution directly from examples. We then use these simulations to train a second neural network to estimate the underlying CMB signal from dust-contaminated maps. We discuss other potential uses for the trained DCGAN, and the generalization to polarized emission from both dust and synchrotron.  more » « less
Award ID(s):
1812199 1560482 1852581
PAR ID:
10253855
Author(s) / Creator(s):
; ; ; ;
Date Published:
Journal Name:
Monthly Notices of the Royal Astronomical Society
Volume:
500
Issue:
3
ISSN:
0035-8711
Page Range / eLocation ID:
3889 to 3897
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; (, Monthly Notices of the Royal Astronomical Society)
    null (Ed.)
    ABSTRACT Emission from the interstellar medium can be a significant contaminant of measurements of the intensity and polarization of the cosmic microwave background (CMB). For planning CMB observations, and for optimizing foreground-cleaning algorithms, a description of the statistical properties of such emission can be helpful. Here, we examine a machine learning approach to inferring the statistical properties of dust from observational data. In particular, we apply a type of neural network called a variational autoencoder (VAE) to maps of the intensity of emission from interstellar dust as inferred from Planck sky maps and demonstrate its ability to (i) simulate new samples with similar summary statistics as the training set, (ii) provide fits to emission maps withheld from the training set, and (iii) produce constrained realizations. We find VAEs are easier to train than another popular architecture: that of generative adversarial networks, and are better suited for use in Bayesian inference. 
    more » « less
  2. ; ; ; ; ; ; ; ; ; ; et al (, The Astrophysical Journal)
    Abstract We characterize Galactic dust filaments by correlating BICEP/Keck and Planck data with polarization templates based on neutral hydrogen (Hi) observations. Dust polarization is important for both our understanding of astrophysical processes in the interstellar medium (ISM) and the search for primordial gravitational waves in the cosmic microwave background (CMB). In the diffuse ISM, Hiis strongly correlated with the dust and partly organized into filaments that are aligned with the local magnetic field. We analyze the deep BICEP/Keck data at 95, 150, and 220 GHz, over the low-column-density region of sky where BICEP/Keck has set the best limits on primordial gravitational waves. We separate the Hiemission into distinct velocity components and detect dust polarization correlated with the local Galactic Hibut not with the Hiassociated with Magellanic Streami. We present a robust, multifrequency detection of polarized dust emission correlated with the filamentary Himorphology template down to 95 GHz. For assessing its utility for foreground cleaning, we report that the Himorphology template correlates inBmodes at a ∼10%–65% level over the multipole range 20 <ℓ< 200 with the BICEP/Keck maps, which contain contributions from dust, CMB, and noise components. We measure the spectral index of the filamentary dust component spectral energy distribution to beβ= 1.54 ± 0.13. We find no evidence for decorrelation in this region between the filaments and the rest of the dust field or from the inclusion of dust associated with the intermediate velocity Hi. Finally, we explore the morphological parameter space in the Hi-based filamentary model. 
    more » « less
  3. ; ; ; (, Monthly Notices of the Royal Astronomical Society)
    ABSTRACT We present a novel technique for cosmic microwave background (CMB) foreground subtraction based on the framework of blind source separation. Inspired by previous work incorporating local variation to generalized morphological component analysis (GMCA), we introduce hierarchical GMCA (HGMCA), a Bayesian hierarchical graphical model for source separation. We test our method on Nside = 256 simulated sky maps that include dust, synchrotron, free–free, and anomalous microwave emission, and show that HGMCA reduces foreground contamination by $$25{{\ \rm per\ cent}}$$ over GMCA in both the regions included and excluded by the Planck UT78 mask, decreases the error in the measurement of the CMB temperature power spectrum to the 0.02–0.03 per cent level at ℓ > 200 (and $$\lt 0.26{{\ \rm per\ cent}}$$ for all ℓ), and reduces correlation to all the foregrounds. We find equivalent or improved performance when compared to state-of-the-art internal linear combination type algorithms on these simulations, suggesting that HGMCA may be a competitive alternative to foreground separation techniques previously applied to observed CMB data. Additionally, we show that our performance does not suffer when we perturb model parameters or alter the CMB realization, which suggests that our algorithm generalizes well beyond our simplified simulations. Our results open a new avenue for constructing CMB maps through Bayesian hierarchical analysis. 
    more » « less
  4. ; ; ; ; ; ; (, Astronomy & Astrophysics)
    High-resolution 3D maps of interstellar dust are critical for probing the underlying physics shaping the structure of the interstellar medium, and for foreground correction of astrophysical observations affected by dust. We aim to construct a new 3D map of the spatial distribution of interstellar dust extinction out to a distance of kpc from the Sun. We leveraged distance and extinction estimates to 54 million nearby stars derived from the Gaia BP/RP spectra. Using the stellar distance and extinction information, we inferred the spatial distribution of dust extinction. We modeled the logarithmic dust extinction with a Gaussian process in a spherical coordinate system via iterative charted refinement and a correlation kernel inferred in previous work. In total, our posterior has over 661 million degrees of freedom. We probed the posterior distribution using the variational inference method MGVI. Our 3D dust map has an angular resolution of up to $ $ ($$N_ side =256$$), and we achieve parsec-scale distance resolution, sampling the dust in $516$ logarithmically spaced distance bins spanning pc . We generated 12 samples from the variational posterior of the 3D dust distribution and release the samples alongside the mean 3D dust map and its corresponding uncertainty. Our map resolves the internal structure of hundreds of molecular clouds in the solar neighborhood and will be broadly useful for studies of star formation, Galactic structure, and young stellar populations. It is available for download in a variety of coordinate systems online and can also be queried via the publicly available dustmaps Python package. 
    more » « less
  5. ; ; (, The Astrophysical Journal)
    Abstract Using 3D dust maps and Planck polarized dust emission data, we investigate the influence of the 3D geometry of the nearby interstellar medium (ISM) on the statistics of the dust polarization on large ( 80 ′ ) scales. We test recent models that assume that the magnetic field probed by the polarized dust emission is preferentially tangential to the Local Bubble wall, but we do not find an imprint of the Local Bubble geometry on the dust polarization fraction. We also test the hypothesis that the complexity of the 3D dust distribution drives some of the measured variation of the dust polarization fraction. We compare sight lines with similar total column densities and find that, on average, the dust polarization fraction decreases when the dust column is substantially distributed among multiple components at different distances. Conversely, the dust polarization fraction is higher for sight lines where the dust is more concentrated in 3D space. This finding is statistically significant for the dust within 1.25 kpc, but the effect disappears if we only consider dust within 270 pc. In conclusion, we find that the extended 3D dust distribution, rather than solely the dust associated with the Local Bubble, plays a role in determining the observed dust polarization fraction at 80′. This conclusion is consistent with a simple analytical prediction and remains robust under various modifications to the analysis. These results illuminate the relationship between the 3D geometry of the ISM and tracers of the interstellar magnetic field. We discuss implications for our understanding of the polarized dust foreground to the cosmic microwave background. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email