Detecting damped Ly α absorbers with Gaussian processes

Garnett, Roman; Ho, Shirley; Bird, Simeon; Schneider, Jeff

doi:10.1093/mnras/stx1958

Cited by 39 publications

(87 citation statements)

References 45 publications

(53 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also show the metal-DLA composite from Mas-Ribas et al (2017), that corresponds to a sub-sample of ∼12,000 DLA candidates in which prominent metal lines are detected. Finally, we created a composite spectrum of extremely strong DLAs (ESDLAs, with log N(H i)(cm −2 ) ≥ 21.7) visually inspected and selected from DLAs automatically discovered in the SDSS by Garnett et al (2016) and Parks et al (2018) (DR12) as well as by the procedure of Noterdaeme et al (2012) applied to DR14 (as in Noterdaeme et al 2014). Because of the small number of systems -only 51 ESDLAs contribute to the wavelength region of H 2 lines -the composite was carefully built after visually normalising the quasar continuum using B-splines for each spectrum and calculating the composite at each wavelength using a median average.…”

Section: Detection Of the H Signal In Composite Dla Spectramentioning

confidence: 99%

Constraining the H2 column density distribution at z ∼ 3 from composite DLA spectra

Balashev

Noterdaeme

2018

Monthly Notices of the Royal Astronomical Society: Letters

View full text Add to dashboard Cite

We present the detection of the average H 2 absorption signal in the overall population of neutral gas absorption systems at z ∼ 3 using composite absorption spectra built from the Sloan Digital Sky Survey-III damped Lyman-α catalogue. We present a new technique to directly measure the H 2 column density distribution function f H 2 (N) from the average H 2 absorption signal. Assuming a power-law column density distribution, we obtain a slope β = −1.29±0.06(stat)±0.10(sys) and an incidence rate of strong H 2 absorptions (with N(H 2 ) 10 18 cm −2 ) to be 4.0±0.5(stat)±1.0(sys) % in H i absorption systems with N(H i)≥ 10 20 cm −2 . Assuming the same inflexion point where f H 2 (N) steepens as at z = 0, we estimate that the cosmological density of H 2 in the column density range log N(H 2 )(cm −2 ) = 18 − 22 is ∼ 15% of the total. We find one order of magnitude higher H 2 incident rate in a sub-sample of extremely strong DLAs (log N(H i)(cm −2 ) ≥ 21.7), which, together with the the derived shape of f H 2 (N), suggests that the typical H i-H 2 transition column density in DLAs is log N(H)(cm −2 ) 22.3 in agreement with theoretical expectations for the average (low) metallicity of DLAs at high-z.

show abstract

Section: Detection Of the H Signal In Composite Dla Spectramentioning

confidence: 99%

Constraining the H2 column density distribution at z ∼ 3 from composite DLA spectra

Balashev

Noterdaeme

2018

Monthly Notices of the Royal Astronomical Society: Letters

View full text Add to dashboard Cite

show abstract

“…In future works, we intend to implement machine-learning techniques to perform the analysis on future, large datasets (e.g. Garnett et al 2017;Parks et al 2018).…”

Section: Lls Identificationmentioning

confidence: 99%

Imprints of the first billion years: Lyman limit systems atz∼ 5

Crighton

Prochaska

Murphy

et al. 2018

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

Lyman Limit systems (LLSs) trace the low-density circumgalactic medium and the most dense regions of the intergalactic medium, so their number density and evolution at high redshift, just after reionisation, are important to constrain. We present a survey for LLSs at high redshifts, z LLS = 3.5-5.4, in the homogeneous dataset of 153 optical quasar spectra at z ∼ 5 from the Giant Gemini GMOS survey. Our analysis includes detailed investigation of survey biases using mock spectra which provide important corrections to the raw measurements. We estimate the incidence of LLSs per unit redshift at z ≈ 4.4 to be (z) = 2.6 ± 0.4. Combining our results with previous surveys at z LLS < 4, the best-fit power-law evolution is (z) = * [(1 + z)/4] α with * = 1.46 ± 0.11 and α = 1.70 ± 0.22 (68% confidence intervals). Despite hints in previous z LLS < 4 results, there is no indication for a deviation from this single power-law soon after reionization. Finally, we integrate our new results with previous surveys of the intergalactic and circumgalactic media to constrain the hydrogen column density distribution function, f (N HI , X), over 10 orders of magnitude. The data at z ∼ 5 are not well described by the f (N HI , X) model previously reported for z ∼ 2-3 (after re-scaling) and a 7-pivot model fitting the full z ∼ 2-5 dataset is statistically unacceptable. We conclude that there is significant evolution in the shape of f (N HI , X) over this ∼2 billion year period.

show abstract

“…These issues inspired our group, and another team (Garnett et al 2016), to employ machine learning techniques to generate fully automated analysis of quasar spectra for the survey of DLAs. The Garnett et al (2016) approach used Gaussian processes to train on data from the 9th data release of the BOSS survey (Noterdaeme et al 2012) to learn a model of the quasar emission spectrum without DLAs. Their algorithm then estimates the probability of a DLA occurring within a given spectrum and generates the probability distribution function for its redshift and column density.…”

Section: Introductionmentioning

confidence: 99%

Deep learning of quasar spectra to discover and characterize damped Lyα systems

Parks

Prochaska

Dong

et al. 2018

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

We have designed, developed, and applied a convolutional neural network (CNN) architecture using multi-task learning to search for and characterize strong HI Lyα absorption in quasar spectra. Without any explicit modeling of the quasar continuum nor application of the predicted line-profile for Lyα from quantum mechanics, our algorithm predicts the presence of strong HI absorption and estimates the corresponding redshift z abs and HI column density N HI , with emphasis on damped Lyα systems (DLAs, absorbers with N HI ≥ 2 × 10 20 cm −2 ). We tuned the CNN model using a custom training set of DLAs injected into DLA-free quasar spectra from the Sloan Digital Sky Survey (SDSS), data release 5 (DR5). Testing on a held-back validation set demonstrates a high incidence of DLAs recovered by the algorithm (97.4% as DLAs and 99% as an HI absorber with N HI > 10 19.5 cm −2 ) and excellent estimates for z abs and N HI . Similar results are obtained against a human-generated survey of the SDSS DR5 dataset. The algorithm yields a low incidence of false positives and negatives but is challenged by overlapping DLAs and/or very high N HI systems. We have applied this CNN model to the quasar spectra of SDSS-DR7 and the Baryonic Oscillation Spectroscopic Survey (BOSS, data release 12) and provide catalogs of 4,913 and 50,969 DLAs respectively (including 1,659 and 9,230 high-confidence DLAs that were previously unpublished). This work validates the application of deep learning techniques to astronomical spectra for both classification and quantitative measurements.

show abstract

Detecting damped Ly α absorbers with Gaussian processes

Cited by 39 publications

References 45 publications

Constraining the H2 column density distribution at z ∼ 3 from composite DLA spectra

Constraining the H2 column density distribution at z ∼ 3 from composite DLA spectra

Imprints of the first billion years: Lyman limit systems atz∼ 5

Deep learning of quasar spectra to discover and characterize damped Lyα systems

Contact Info

Product

Resources

About