Using Dark Energy Explorers and Machine Learning to Enhance the Hobby–Eberly Telescope Dark Energy Experiment
Explore the source record for details and available documents.
Lindsay R. House↗
SEARCH · Search NASA
Results for “dark energy experiments”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
DESI 2024 VI: cosmological constraints from the measurements of baryon acoustic oscillations
We present cosmological results from the measurement of baryon acoustic oscillations (BAO) in galaxy, quasar and Lyman-α forest tracers from the first year of observations from the Dark Energy Spectroscopic Instrument (DESI), to be released in the DESI Data Release 1. DESI BAO provide robust measurements of the transverse comoving distance and Hubble rate, or their combination, relative to the sound horizon, in seven redshift bins from over 6 million extragalactic objects in the redshift range 0.1 < z < 4.2. To mitigate confirmation bias, a blind analysis was implemented to measure the BAO scales. DESI BAO data alone are consistent with the standard flat ΛCDM cosmological model with a matter density Ω m =0.295±0.015. Paired with a baryon density prior from Big Bang Nucleosynthesis and the robustly measured acoustic angular scale from the cosmic microwave background (CMB), DESI requires H 0 =(68.52±0.62) km s -1 Mpc -1 . In conjunction with CMB anisotropies from Planck and CMB lensing data from Planck and ACT, we find Ω m =0.307± 0.005 and H 0 =(67.97±0.38) km s -1 Mpc -1 . Extending the baseline model with a constant dark energy equation of state parameter w, DESI BAO alone requirew=-0.99 +0.15 -0.13 . In models with a time-varying dark energy equation of state parametrised by w 0 and w a , combinations of DESI with CMB or with type Ia supernovae (SN Ia) individually prefer w 0 > -1 and w a < 0. This preference is 2.6σ for the DESI+CMB combination, and persists or grows when SN Ia are added in, giving results discrepant with the ΛCDM model at the 2.5σ, 3.5σ or 3.9σ levels for the addition of the Pantheon+, Union3, or DES-SN5YR supernova datasets respectively. For the flat ΛCDM model with the sum of neutrino mass ∑ m ν free, combining the DESI and CMB data yields an upper limit ∑ m ν < 0.072 (0.113) eV at 95% confidence for a ∑ m ν > 0 (∑ m ν > 0.059) eV prior. These neutrino-mass constraints are substantially relaxed if the background dynamics are allowed to deviate from flat ΛCDM.
79 ASTRONOMY AND ASTROPHYSICS↗
DESI 2024: reconstructing dark energy using crossing statistics with DESI DR1 BAO data
Here, we implement Crossing Statistics to reconstruct in a model-agnostic manner the expansion history of the universe and properties of dark energy, using DESI Data Release 1 (DR1) BAO data in combination with one of three different supernova compilations (PantheonPlus, Union3, and DES-SN5YR) and Planck CMB observations. Our results hint towards an evolving and emergent dark energy behaviour, with negligible presence of dark energy at z ≳ 1, at varying significance depending on data sets combined. In all these reconstructions, the cosmological constant lies outside the 95% confidence intervals for some redshift ranges. This dark energy behaviour, reconstructed using Crossing Statistics, is in agreement with results from the conventional w 0 –w a dark energy equation of state parametrization reported in the DESI Key cosmology paper. Our results add an extensive class of model-agnostic reconstructions with acceptable fits to the data, including models where cosmic acceleration slows down at low redshifts. We also report constraints on H 0 r d from our model-agnostic analysis, independent of the pre-recombination physics.
79 ASTRONOMY AND ASTROPHYSICS↗
Optimal parameterizations for observational constraints on thawing dark energy
Time-varying dark energy is often modeled in observational analyses through generic parameterizations of its equation of state w(z), which typically use two free parameters {w 0 , w a } to span a broad range of behaviors as a function of redshift. However, this broad range of behaviors can only approximately capture the dynamics of any given microphysical theory of dark energy. A complementary approach is to use targeted parameterizations designed to model specific classes of dynamical dark energy with greater precision. Focusing on the class of thawing dark energy, we quantify and compare the precision with which nineteen generic and targeted parameterizations can capture the dynamics of physically motivated thawing quintessence theories. We find that a targeted parameterization derived from a Padé expansion of w is the most reliable of these, producing accurate reconstructions of w(z), the expansion history H(z), and cosmological parameters such as H 0 and Ω m for a broad range of microphysical theories.
dark energy experiments↗
DESI Spectroscopy of HETDEX Emission-line Candidates. I. Line Discrimination Validation
The Hobby–Eberly Dark Energy Experiment (HETDEX) is an untargeted spectroscopic galaxy survey that uses Lyα-emitting galaxies (LAEs) as tracers of 1.9 < z < 3.5 large-scale structure. Most detections consist of a single emission line, whose identity is inferred via a Bayesian analysis of ancillary data. To determine the accuracy of these line identifications, HETDEX detections were observed with the Dark Energy Spectroscopic Instrument (DESI). In two DESI pointings, high-confidence spectroscopic redshifts are obtained for 1157 sources, including 982 LAEs. The DESI spectra are used to evaluate the accuracy of the HETDEX object classifications and tune the methodology to achieve the HETDEX science requirement of ≲2% contamination of the LAE sample by low-redshift emission-line galaxies, while still assigning 96% of the true Lyα emission sample with the correct spectroscopic redshift. We compare emission-line measurements between the two experiments assuming a simple Gaussian line fitting model. Fitted values for the central wavelength of the emission line, the measured line flux, and line widths are consistent between the surveys within uncertainties. Derived spectroscopic redshifts, from the two classification pipelines, when both agree as an LAE classification, are consistent to within $\langle$Δz/(1 + z)$\rangle$ = 6.9 × 10 −5 with an rms scatter of 3.3 × 10 −4 . Data are available at https://data.desi.lbl.gov/desi/public/dr1/vac/dr1/hetdex.
79 ASTRONOMY AND ASTROPHYSICS↗
SPT-3G D1: Axion early dark energy with CMB experiments and DESI observations
We present the most up-to-date constraints on axion early dark energy (AEDE) from cosmic microwave background (CMB) and baryon acoustic oscillation (BAO) measurements. In particular, we assess the impact of data from ground-based CMB experiments, the South Pole Telescope (SPT) and the Atacama Cosmology Telescope (ACT)—both with and without Planck—on constraints on AEDE. We also highlight the impact that BAO information from the Dark Energy Spectroscopic Instrument (DESI) has on these constraints. From CMB data alone, we do not find statistically significant evidence for the presence of AEDE, and we find only moderate reduction in the Hubble tension. From the latest SPT data alone, we find the maximal fractional contribution of AEDE to the cosmic energy budget is f EDE < 0.12 at 95% confidence level (CL), and the Hubble tension between the SPT and Supernovae, H0, for the Equation of State of Dark Energy (SH0ES) results is reduced to the 2.3 σ level. When combining the latest SPT, ACT, and Planck datasets, we find f EDE < 0.070 at 95% CL and the Hubble tension at the 3.6 σ level. In contrast, adding DESI data to the CMB datasets results in mild preference for AEDE and, in some cases, non-negligible reduction in the Hubble tension. From SPT + DESI , we find f EDE = 0.08 1 - 0.052 + 0.037 at 68% CL, and the Hubble tension reduces to 1.5 σ . From the combination of DESI with all three CMB experiments, we get f EDE = 0.05 5 - 0.047 + 0.024 at 68% CL and a weak preference for AEDE over Λ CDM . This data combination, in turn, reduces the Hubble tension to 2.6 σ . We highlight that this shift in parameters when adding the DESI dataset is a manifestation of the discrepancy currently present between DESI and CMB experiments in the concordance model Λ CDM .
Khalife, A. R. [Paris, Inst. Astrophys.] (ORCID:00↗
Unfolding Structure Formation in the Dark Universe via Weak Gravitational Lensing: from Pixels to Cosmology in the Dark Energy Survey and Roman Space Telescope
Weak gravitational lensing (deflection of light from distant galaxies due to the gravitational potential of intervening mass) is one of the most exciting probes in cosmology as it is sensitive both to the growth of the large-scale structure and the expansion history of the Universe. We can measure the coherent distortion of galaxy shapes (which we call cosmic shear) and infer the matter distribution. With next-generation imaging surveys such as the Vera C. Rubin Observatory Legacy Survey of Space and Time (Rubin LSST) and the Nancy G. Roman Space Telescope (Roman), there is immense new promise in understanding the fundamental nature of dark matter and dark energy. With datasets from current experiments like the Dark Energy Survey (DES), we see apparent cosmological tensions between experiments that may either be real and indicate new physics or new systematics we do not yet understand. LSST and Roman will increase the number of galaxies we have observed by an order of magnitude, leading to improved constraints on our cosmological model by up to 300%. These experiments will have the potential to prove if these cosmological tensions are real. However, our control of systematic uncertainties must also improve by similar levels to achieve the promise of what these missions can deliver. To achieve these scientific outcomes, the challenges of determining galaxy shapes and redshifts and modeling the impact of astrophysical effects must be solved. My PhD has focused on enabling weak lensing science in DES and Roman. I co-led the shear analysis teams in DES to produce the largest weak lensing galaxy sample, as well as the final DES cosmological analysis of cosmic shear. For Roman, I have led two papers characterizing and mitigating shear-related systematics with image simulations.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Systematic study of projection biases in the weak lensing analysis of cosmic shear and the combination of galaxy clustering and galaxy-galaxy lensing
This paper presents the results of a systematic study of projection biases in the weak lensing analysis of cosmic shear and the combination of galaxy clustering and galaxy-galaxy lensing using data collected during the first year of running the Dark Energy Survey experiment. The study uses Lambda cold dark matter ( Λ CDM ) as the cosmological model and two-point correlation functions for the weak lensing (WL) analysis. The results in this paper show that, independent of the WL analysis, projection biases of more than 1 σ exist and are a function of the position of the true values of the parameters h , n s , Ω b , and Ω ν h 2 with respect to their prior probabilities. For cosmic shear, and the combination of galaxy clustering and galaxy-galaxy lensing, this study shows that the coverage probability of the 68.27% credible intervals ranges from as high as 93% to as low as 16% and that these credible intervals are inflated, on average, by 29% for cosmic shear and 20% for the combination of galaxy clustering and galaxy-galaxy lensing. The results of the study also show that, in six out of nine tested cases, the reduction in error bars obtained by transforming credible intervals into confidence intervals is equivalent to an increase in the amount of data by a factor of 3.
79 ASTRONOMY AND ASTROPHYSICS↗
Probing the Properties of Superheavy Dark Matter Annihilating or Decaying into Neutrinos With Ultra-High Energy Neutrino Experiments
The evidence for dark matter particles,x, is compelling based on Galactic to cosmological scale observations. Thus far, the promising weakly interacting massive particle scenario have eluded detection, motivating alternative models of dark matter. We consider scenarios involving super-heavy dark matter (SHDM) that potentially can decay or annihilate to neutrinos and antineutrinos. In the mass range m(sub x) =10(exp 7) −10(exp 15) GeV, we evaluate the sensitivities of future observatories PO-EMMA and GRAND for indirect dark matter detection via the measurement of neutrino-induced extensive air showers (EAS), compute the Auger and ANITA limits using their last up-to-date sensitivities, and compare them with IceCube limits. We also show that the uncertainties related to the dark matter distribution in the Galactic halo have a large impact on the neutrino flux. We show that a ground-based radio detector such as GRAND can achieve high sensitivities due to its large effective area and high duty cycle. Space-based Cherenkov detectors such as POEMMA that measure the EAS optical Cherenkov signal have the advantage of full-sky coverage and rapid slewing, enabling an optimized SHDM observation strategy focusing on the Galactic Center. We show that increasing the field of view of the Cherenkov detectors can significantly enhance the sensitivity. Moreover, POEMMA’s fluorescence observation mode that measures EAS above20EeV will achieve state-of-the-art sensitivity to SHDM properties at the highest mass scales.
Claire Guepin↗
Full Modeling and parameter compression methods in configuration space for DESI 2024 and beyond
In the contemporary era of high-precision spectroscopic surveys, led by projects like DESI, there is an increasing demand for optimizing the extraction of cosmological information from clustering data. This work conducts a thorough comparison of various methodologies for modeling the full shape of the two-point statistics in configuration space. We investigate the performance of both direct fits (Full Modeling) and the parameter compression approaches (ShapeFit and Standard). We utilize the ABACUS-SUMMIT simulations, tailored to exceed DESI's precision requirements. Particularly, we fit the two-point statistics of three distinct tracers (LRG, ELG, and QSO), by employing a Gaussian Streaming Model in tandem with Convolution Lagrangian Perturbation Theory and Effective Field Theory. We explore methodological setup variations, including the range of scales, the set of galaxy bias parameters, the inclusion of the hexadecapole, as well as model extensions encompassing varying ns and allowing for w 0 w a CDM dark energy model. Throughout these varied explorations, while precision levels fluctuate and certain configurations exhibit tighter parameter constraints, our pipeline consistently recovers the parameter values of the mocks within 1σ in all cases for a 1-year DESI volume. Additionally, we compare the performance of configuration space analysis with its Fourier space counterpart using three models: PyBird, FOLPS and velocileptors, presented in companion papers. We find good agreement with the results from all these models.
79 ASTRONOMY AND ASTROPHYSICS↗
Impact of systematic redshift errors on the cross-correlation of the Lyman-α forest with quasars at small scales using DESI Early Data
The Dark Energy Spectroscopic Instrument (DESI) will measure millions of quasar spectra by the end of its 5 year survey. Quasar redshift errors impact the shape of the Lyman-α forest correlation functions, which can affect cosmological analyses and therefore cosmological interpretations. Here, using data from the DESI Early Data Release and the first two months of the main survey, we measure the systematic redshift error from an offset in the cross-correlation of the Lyman-α forest with quasars. We find evidence for a redshift dependent bias causing redshifts to be underestimated with increasing redshift, stemming from improper modeling of the Lyman-α optical depth in the templates used for redshift estimation. New templates were derived for the DESI Year 1 quasar sample at z > 1.6 and we found the redshift dependent bias, Δr ∥ , increased from -1.94 ± 0.15 h -1 Mpc to -0.08 ± 0.04 h -1 Mpc (-205 ± 15 km s -1 to -9.0 ± 4.0 km s -1 ). These new templates will be used to provide redshifts for the DESI Year 1 quasar sample.
Lyman alpha forest↗
Validation of the DESI 2024 Lyman alpha forest BAL masking strategy
Broad absorption line quasars (BALs) exhibit blueshifted absorption relative to a number of their prominent broad emission features. These absorption features can contribute to quasar redshift errors and add absorption to the Lyman-α (Lyα) forest that is unrelated to large-scale structure. We present a detailed analysis of the impact of BALs on the Baryon Acoustic Oscillation (BAO) results with the Lyα forest from the first year of data from the Dark Energy Spectroscopic Instrument (DESI). The baseline strategy for the first year analysis is to mask all pixels associated with all BAL absorption features that fall within the wavelength region used to measure the forest. We explore a range of alternate masking strategies and demonstrate that these changes have minimal impact on the BAO measurements with both DESI data and synthetic data. This includes when we mask the BAL features associated with emission lines outside of the forest region to minimize their contribution to redshift errors. We identify differences in the properties of BALs in the synthetic datasets relative to the observational data, as well as use the synthetic observations to characterize the completeness of the BAL identification algorithm, and demonstrate that incompleteness and differences in the BALs between real and synthetic data also do not impact the BAO results for the Lyα forest.
Lyman alpha forest↗
Characterization of contaminants in the Lyman-alpha forest auto-correlation with DESI
Baryon Acoustic Oscillations can be measured with sub-percent precision above redshift two with the Lyman-α (Lyα) forest auto-correlation and its cross-correlation with quasar positions. This is one of the key goals of the Dark Energy Spectroscopic Instrument (DESI) which started its main survey in May 2021. We present in this paper a study of the contaminants to the Lyα forest which are mainly caused by correlated signals introduced by the spectroscopic data processing pipeline as well as astrophysical contaminants due to foreground absorption in the intergalactic medium. Notably, an excess signal caused by the sky background subtraction noise is present in the Lyα auto-correlation in the first line-of-sight separation bin. We use synthetic data to isolate this contribution, we also characterize the effect of spectro-photometric calibration noise, and propose a simple model to account for both effects in the analysis of the Lyα forest. We then measure the auto-correlation of the quasar flux transmission fraction of low redshift quasars, where there is no Lyα forest absorption but only its contaminants. We demonstrate that we can interpret the data with a two-component model: data processing noise and triply ionized Silicon and Carbon auto-correlations. This result can be used to improve the modeling of the Lyα auto-correlation function measured with DESI.
79 ASTRONOMY AND ASTROPHYSICS↗
Validation of the DESI 2024 Lyα forest BAO analysis using synthetic datasets
The first year of data from the Dark Energy Spectroscopic Instrument (DESI) contains the largest set of Lyman-α (Lyα) forest spectra ever observed. This data, collected in the DESI Data Release 1 (DR1) sample, has been used to measure the Baryon Acoustic Oscillation (BAO) feature at redshift z = 2.33. In this work, we use a set of 150 synthetic realizations of DESI DR1 to validate the DESI 2024 Lyα forest BAO measurement presented in [1]. The synthetic data sets are based on Gaussian random fields using the log-normal approximation. We produce realistic synthetic DESI spectra that include all major contaminants affecting the Lyα forest. The synthetic data sets span a redshift range 1.8 < z < 3.8, and are analyzed using the same framework and pipeline used for the DESI 2024 Lyα forest BAO measurement. To measure BAO, we use both the Lyα auto-correlation and its cross-correlation with quasar positions. We use the mean of correlation functions from the set of DESI DR1 realizations to show that our model is able to recover unbiased measurements of the BAO position. We also fit each mock individually and study the population of BAO fits in order to validate BAO uncertainties and test our method for estimating the covariance matrix of the Lyα forest correlation functions. Finally, we discuss the implications of our results and identify the needs for the next generation of Lyα forest synthetic data sets, with the top priority being to simulate the effect of BAO broadening due to non-linear evolution.
79 ASTRONOMY AND ASTROPHYSICS↗
DESI 2024 VII: cosmological constraints from the full-shape modeling of clustering measurements
We present cosmological results from the measurement of clustering of galaxy, quasar and Lyman-α forest tracers from the first year of observations with the Dark Energy Spectroscopic Instrument (DESI Data Release 1). We adopt the full-shape (FS) modeling of the power spectrum, including the effects of redshift-space distortions, in an analysis which has been thoroughly validated in a series of supporting papers as summarised in [1]. We combine the full-shape information with DESI's DR1 constraints from the baryon acoustic oscillations (BAO) of these tracers. In the flat ΛCDM cosmological model, DESI (FS+BAO), combined with a baryon density prior from Big Bang Nucleosynthesis and a weak prior on the scalar spectral index, determines matter density to Ω m = 0.2962 ± 0.0095, and the amplitude of mass fluctuations to σ 8 = 0.842 ± 0.034. The addition of the cosmic microwave background (CMB) data tightens these constraints to Ω m = 0.3056 ± 0.0049 and σ 8 = 0.8121 ± 0.0053, while further addition of the joint clustering and lensing analysis from the Dark Energy Survey Year-3 (DESY3) data further improves these measurements, and leads to a 0.4% determination of the Hubble constant, H 0 = (68.40 ± 0.27) km s -1 Mpc -1 . In models with a time-varying dark energy equation of state parametrised by w 0 and w a , combinations of DESI (FS+BAO) with CMB and type Ia supernovae continue to show the preference, previously found in the DESI DR1 BAO analysis, for w 0 > -1 and w a < 0 with similar levels of significance. DESI data, in combination with the CMB, improve the upper limits on the sum of the neutrino masses relative to the case when only the DR1 BAO was available, giving ∑m ν < 0.071 eV at 95% confidence. We finally constrain deviations from general relativity represented by two modified gravity parameters. DESI (FS+BAO) data alone measure the parameter that controls the clustering of massive particles, μ 0 = 0.11 +0.45 -0.54 , in agreement with the zero value predicted by general relativity. The combination of DESI with the CMB and the clustering and lensing analysis from DESY3 constrains both modified-gravity parameters, giving μ 0 = 0.04 ± 0.22 and Σ 0 = 0.044 ± 0.047, again in agreement with general relativity.
79 ASTRONOMY AND ASTROPHYSICS↗
Exploring HOD-dependent systematics for the DESI 2024 Full-Shape galaxy clustering analysis
We analyze the robustness of the DESI 2024 cosmological inference from the full shape of the galaxy power spectrum to uncertainties in the Halo Occupation Distribution (HOD) model of the galaxy-halo connection and the choice of priors on nuisance parameters. We assess variations in the recovered cosmological parameters across a range of mocks populated with different HOD models and find that shifts are often greater than 20% of the expected statistical uncertainties from the DESI data. We encapsulate the effect of such shifts in terms of a systematic covariance term, C HOD , and an additional diagonal contribution quantifying the impact of our choice of nuisance parameter priors on the ability of the effective field theory (EFT) model to correctly recover the cosmological parameters of the simulations. These two covariance contributions are designed to be added to the usual covariance term, C stat , describing the statistical uncertainty in the power spectrum measurement, in order to fairly represent these sources of systematic uncertainty. This novel approach should be more general and robust to the choice of model or additional external datasets used in cosmological fits than the alternative approach of adding systematic uncertainties to the recovered marginalised parameter posteriors. We compare the approaches within the context of a fixed ΛCDM model and demonstrate that our method gives conservative estimates of the systematic uncertainty that nevertheless have little impact on the final posteriors obtained from DESI data.
79 ASTRONOMY AND ASTROPHYSICS↗
Modified gravity constraints from the full shape modeling of clustering measurements from DESI 2024
We present cosmological constraints on deviations from general relativity (GR) from the first-year of clustering observations from the Dark Energy Spectroscopic Instrument (DESI) in combination with other available datasets including the CMB data from Planck with CMB-lensing from Planck and ACT, BBN constraints on the physical baryon density, the galaxy weak lensing and clustering from DESY3 and supernova data from DESY5. We first consider the μ(a,k)–Σ(a,k) modified gravity (MG) parameterization (as well as η(a,k)) in a ΛCDM and a w 0 w a CDM cosmological backgrounds. Using a functional form for time-only evolution gives μ 0 = 0.11 +0.44 -0.54 from DESI(FS+BAO)+BBN and a wide prior on n s . Using DESI(FS+BAO)+CMB+DESY3+DESY5-SN, we obtain μ 0 = 0.05 ± 0.22 and Σ 0 = 0.008 ± 0.045 and similarly μ 0 = 0.02 +0.19 -0.24 and η 0 = 0.09 +0.36 -0.60 , in an ΛCDM background. In w 0 w a CDM we obtain μ 0 = -0.24 +0.32 -0.28 and Σ 0 = 0.006 ± 0.043, consistent with GR, and we still find a preference of the data for a dynamical dark energy with w 0 > -1 and w a < 0. Using functional dependencies in both time and scale gives μ 0 and Σ 0 with a same level of precision as above but other scale MG parameters remain hard to constrain. We then move to binned parameterizations in a ΛCDM background starting with two bins in redshift and obtain, μ 1 = 1.02 ± 0.13, μ 2 = 1.04 ± 0.11, Σ 1 = 1.021 ± 0.029 and Σ 2 = 1.022 +0.027 -0.023 , all consistent with the unity value of GR in the binning formalism. We then extend the analysis to combine two bins in redshift and two in scale giving 8 MG parameters that we find all consistent with GR. We note that we find here that the tension reported in previous studies about Σ 0 being inconsistent with GR when using Planck PR3 data goes away when we use the recent LoLLiPoP+HiLLiPoP likelihoods. As noted in previous studies, this seems to indicate that the tension is indeed related to the CMB lensing anomaly in PR3 which is also resolved when using the recent likelihoods. We then constrain the class of Horndeski theory in the effective field theory of dark energy approach. We consider both EFT-basis and α-basis in the analysis. Assuming a power law parameterization for the EFT function Ω, which controls non-minimal coupling, we obtain Ω 0 = 0.012 +0.001 -0.012 and s 0 = 0.996 +0.54 -0.20 from the combination of DESI(FS+BAO)+DESY5SN+CMB in a ΛCDM background, which are consistent with GR. Similar results are obtained when using the α-basis and assuming no-braiding (α B = 0) giving c M < 1.14 at 95% CL in a ΛCDM background, also in agreement with GR. However, we see a mild yet consistent indication for c B > 0 when α B is allowed to vary which will require further study to determine whether this is due to systematics or new physics.
79 ASTRONOMY AND ASTROPHYSICS↗
Using active learning to improve quasar identification for the DESI spectra processing pipeline
The Dark Energy Spectroscopic Instrument (DESI) survey uses an automatic spectral classification pipeline to classify spectra. QuasarNET is a convolutional neural network used as part of this pipeline originally trained using data from the Baryon Oscillation Spectroscopic Survey (BOSS). In this paper we implement an active learning algorithm to optimally select spectra to use for training a new version of the QuasarNET weights file using only DESI data, with the goal of improving classification accuracy. This active learning algorithm includes a novel outlier rejection step using a Self-Organizing Map to ensure we label spectra representative of the larger quasar sample observed in DESI. We perform two iterations of the active learning pipeline, assembling a final dataset of 5600 labeled spectra, a small subset of the approximately 1.3 million quasar targets in DESI's Data Release 1. When splitting the spectra into training and validation subsets we achieve similar performance to the previously trained weights file in completeness and purity calculated on the validation dataset but do so with less than one tenth of the amount of training data. The new weights also more consistently classify objects in the same way when used on unlabeled data compared to the old weights file. In the process of improving QuasarNET's classification accuracy we discovered a systemic error in QuasarNET's redshift estimation and used our findings to improve our understanding of QuasarNET's redshifts.
Machine learning↗