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: Deep forest: Neural network reconstruction of the Lyman-α forest
ABSTRACT We explore the use of Deep Learning to infer physical quantities from the observable transmitted flux in the Ly α forest. We train a Neural Network using redshift z = 3 outputs from cosmological hydrodynamic simulations and mock data sets constructed from them. We evaluate how well the trained network is able to reconstruct the optical depth for Ly α forest absorption from noisy and often saturated transmitted flux data. The Neural Network outperforms an alternative reconstruction method involving log inversion and spline interpolation by approximately a factor of 2 in the optical depth root mean square error. We find no significant dependence in the improvement on input data signal to noise, although the gain is greatest in high optical depth regions. The Ly α forest optical depth studied here serves as a simple, one dimensional, example but the use of Deep Learning and simulations to approach the inverse problem in cosmology could be extended to other physical quantities and higher dimensional data.  more » « less
Award ID(s):
1909193 2020295
PAR ID:
10290272
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
Monthly Notices of the Royal Astronomical Society
Volume:
506
Issue:
4
ISSN:
0035-8711
Page Range / eLocation ID:
5212 to 5222
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; (, Monthly Notices of the Royal Astronomical Society)
    ABSTRACT We explore the use of deep learning to infer the temperature of the intergalactic medium from the transmitted flux in the high-redshift Ly α forest. We train neural networks on sets of simulated spectra from redshift z = 2–3 outputs of cosmological hydrodynamic simulations, including high-temperature regions added in post-processing to approximate bubbles heated by He ii reionization. We evaluate how well the trained networks are able to reconstruct the temperature from the effect of Doppler broadening in the simulated input Ly α forest absorption spectra. We find that for spectra with high resolution (10 $$\, {\rm km}\, {\rm s}^{-1}$$ pixel) and moderate signal-to-noise ratio (20–50), the neural network is able to reconstruct the intergalactic medium temperature smoothed on scales of $$\sim 6 \, h^{-1}\, {\rm Mpc}$$ quite well. Concentrating on discontinuities, we find that high-temperature regions of width $$25 \, h^{-1}\, {\rm Mpc}$$ and temperature $$20\, 000$$ K can be fairly easily detected and characterized. We show an example where multiple sightlines are combined to yield tomographic images of hot bubbles. Deep learning techniques may be useful in this way to help us understand the complex temperature structure of the intergalactic medium around the time of helium reionization. 
    more » « less
  2. ; ; ; ; ; ; ; ; (, The Astrophysical Journal)
    Abstract We investigate the metal species associated with the Ly α forest in eBOSS quasar spectra. Metal absorption lines are revealed in stacked spectra from cross-correlating the selected Ly α absorbers in the forest and the flux fluctuation field. Up to 13 metal species are identified as being associated with relatively strong Ly α absorbers (those with flux fluctuations − 1.0 < δ Ly α < − 0.6 and with a neutral hydrogen column density of ∼ 10 15−16 cm −2 ) over the absorber redshift range of 2 < z abs < 4. The column densities of these species decrease toward higher redshift and for weaker Ly α absorbers. From modeling the column densities of various species, we find that the column density pattern suggests contributions from multiple gas components, both in the circumgalactic medium (CGM) and the intergalactic medium (IGM). While the low-ionization species (e.g., C ii , Si ii , and Mg ii ) can be explained by high-density, cool gas ( T ∼ 10 4 K) from the CGM, the high-ionization species may reside in low-density or high-temperature gas in the IGM. The measurements provide inputs for modeling the metal contamination in the Ly α forest baryon acoustic oscillation measurements. Comparisons with metal absorptions in high-resolution quasar spectra and hydrodynamic galaxy formation simulations can further elucidate the physical conditions of these Ly α absorbers. 
    more » « less
  3. ; ; ; ; ; ; ; ; ; ; et al (, The Astrophysical Journal)
    Abstract We present a new investigation of the intergalactic medium near reionization using dark gaps in the Lyβforest. With its lower optical depth, Lyβoffers a potentially more sensitive probe to any remaining neutral gas compared to the commonly used Lyαline. We identify dark gaps in the Lyβforest using spectra of 42 QSOs atzem> 5.5, including new data from the XQR-30 VLT Large Programme. Approximately 40% of these QSO spectra exhibit dark gaps longer than 10h−1Mpc atz≃ 5.8. By comparing the results to predictions from simulations, we find that the data are broadly consistent both with models where fluctuations in the Lyαforest are caused solely by ionizing ultraviolet background fluctuations and with models that include large neutral hydrogen patches atz< 6 due to a late end to reionization. Of particular interest is a very long (L= 28h−1Mpc) and dark (τeff≳ 6) gap persisting down toz≃ 5.5 in the Lyβforest of thez= 5.85 QSO PSO J025−11. This gap may support late reionization models with a volume-weighted average neutral hydrogen fraction of 〈xH I〉 ≳ 5% byz= 5.6. Finally, we infer constraints on 〈xH I〉 over 5.5 ≲z≲ 6.0 based on the observed Lyβdark gap length distribution and a conservative relationship between gap length and neutral fraction derived from simulations. We find 〈xH I〉 ≤ 0.05, 0.17, and 0.29 atz≃ 5.55, 5.75, and 5.95, respectively. These constraints are consistent with models where reionization ends significantly later thanz= 6. 
    more » « less
  4. ; ; ; (, The Astrophysical Journal)
    Abstract The density and temperature properties of the intergalactic medium (IGM) reflect the heating and ionization history during cosmological structure formation, and are primarily probed by the Ly α forest of neutral hydrogen absorption features in the observed spectra of background sources. We present the methodology and initial results from the Cholla IGM Photoheating Simulation (CHIPS) suite performed with the graphics process unit–accelerated Cholla code to study the IGM at high, uniform spatial resolution maintained over large volumes. In this first paper, we examine the IGM structure in CHIPS cosmological simulations that include IGM uniform photoheating and photoionization models where hydrogen reionization is completed early or by redshift z ∼ 6. Comparing with observations of the large- and small-scale Ly α transmitted flux power spectra P ( k ) at redshifts 2 ≲ z ≲ 5.5, the relative agreement of the models depends on scale, with the self-consistent Puchwein et al. IGM photoheating and photoionization model in good agreement with the flux P ( k ) at k ≳ 0.01 s km −1 at redshifts 2 ≲ z ≲ 3.5. On larger scales, the P ( k ) measurements increase in amplitude from z ∼ 4.6 to z ∼ 2.2, faster than the models, and lie in between the model predictions at 2.2 ≲ z ≲ 4.6 for k ≈ 0.002–0.01 s km −1 . We argue that the models could improve by changing the He ii photoheating rate associated with active galactic nuclei to reduce the IGM temperature at z ∼ 3. At higher redshifts, z ≳ 4.5, the observed flux P ( k ) amplitude increases at a rate intermediate between the models, and we argue that for models where hydrogen reionization is completed late ( z ∼ 5.5–6), resolving this disagreement will require inhomogeneous or “patchy” reionization. We then use an additional set of simulations to demonstrate that our results have numerically converged and are not strongly affected by varying cosmological parameters. 
    more » « less
  5. ; ; ; ; ; ; ; ; ; ; et al (, Monthly Notices of the Royal Astronomical Society)
    ABSTRACT We present the first observational measurements of the Lyman-α (Ly α) forest flux autocorrelation functions in ten redshift bins from 5.1 ≤ z ≤ 6.0. We use a sample of 35 quasar sightlines at z > 5.7 from the extended XQR-30 data set; these data have signal-to-noise ratios of >20 per spectral pixel. We carefully account for systematic errors in continuum reconstruction, instrumentation, and contamination by damped Ly α systems. With these measurements, we introduce software tools to generate autocorrelation function measurements from any simulation. Our measurements of the smallest bin of the autocorrelation function increase with redshift when normalizing by the mean flux, 〈F〉. This increase may come from decreasing 〈F〉 or increasing mean free path of hydrogen-ionizing photons, λmfp. Recent work has shown that the autocorrelation function from simulations at z > 5 is sensitive to λmfp, a quantity that contains vital information on the ending of reionization. For an initial comparison, we show our autocorrelation measurements with simulation models for recently measured λmfp values and find good agreements. Further work in modelling and understanding the covariance matrices of the data is necessary to get robust measurements of λmfp from this data. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email