Search for: All records

Abstract Key science questions, such as galaxy distance estimation and weather forecasting, often require knowing the full predictive distribution of a target variableYgiven complex inputsX. Despite recent advances in machine learning and physics-based models, it remains challenging to assess whether an initial model is calibrated for allx, and when needed, to reshape the densities ofytoward ‘instance-wise’ calibration. This paper introduces the local amortized diagnostics and reshaping of conditional densities (LADaR) framework and proposes a new computationally efficient algorithm (Cal-PIT) that produces interpretable local diagnostics and provides a mechanism for adjusting conditional density estimates (CDEs).Cal-PITlearns a single interpretable local probability–probability map from calibration data that identifies where and how the initial model is miscalibrated across feature space, which can be used to morph CDEs such that they are well-calibrated. We illustrate the LADaR framework on synthetic examples, including probabilistic forecasting from image sequences, akin to predicting storm wind speed from satellite imagery. Our main science application involves estimating the probability density functions of galaxy distances given photometric data, whereCal-PITachieves better instance-wise calibration than all 11 other literature methods in a benchmark data challenge, demonstrating its utility for next-generation cosmological analyzes⁹9Code available as a Python package here:https://github.com/lee-group-cmu/Cal-PIT.. 

Abstract Poststarburst galaxies (PSBs) are young quiescent galaxies that have recently experienced a rapid decrease in star formation, allowing us to probe the fast-quenching period of galaxy evolution. In this work, we obtained Hubble Space Telescope (HST)/WFC3 F110W imaging to measure the sizes of 171 massive ( $\log (M_{*}/M_{\odot })\sim 11)$spectroscopically identified PSBs at 1 <z1.3 selected from the DESI Survey Validation luminous red galaxy sample. This statistical sample constitutes an order of magnitude increase from the ∼20 PSBs with space-based imaging and deep spectroscopy. We perform structural fitting of the target galaxies withpysersicand compare them to quiescent and star-forming galaxies in the 3D-HST survey. We find that these PSBs are more compact than the general population of quiescent galaxies, lying systematically ∼0.1 dex below the established size–mass relation. However, their central surface mass densities are similar to those of their quiescent counterparts ( $\log \left({\mathrm{\Sigma }}_{1\mathrm{kpc}}/\right(M_{\odot }{\mathrm{kpc}}^{-2}\left)\right)\sim 10.1$). These findings are easily reconciled by later ex situ growth via minor mergers or a slight progenitor bias. These PSBs are round in projection (b/a_median∼ 0.8), suggesting that they are primarily spheroids, not disks, in 3D. We find no correlation between the time since quenching and light-weighted PSB sizes or central densities. This disfavors apparent structural growth due to the fading of centralized starbursts in this galaxy population. Instead, we posit that the fast quenching of massive galaxies at this epoch occurs preferentially in galaxies with preexisting compact structures. 

Many astrophysical analyses depend on estimates of redshifts (a proxy for distance) determined from photometric (i.e., imaging) data alone. Inaccurate estimates of photometric redshift uncertainties can result in large systematic errors. However, probability distribution outputs from many photometric redshift methods do not follow the frequentist definition of a Probability Density Function (PDF) for redshift — i.e., the fraction of times the true redshift falls between two limits z1 and z2 should be equal to the integral of the PDF between these limits. Previous works have used the global distribution of Probability Integral Transform (PIT) values to re-calibrate PDFs, but offsetting inaccuracies in different regions of feature space can conspire to limit the efficacy of the method. We leverage a recently developed regression technique that characterizes the local PIT distribution at any location in feature space to perform a local re-calibration of photometric redshift PDFs resulting in calibrated predictive distributions. Though we focus on an example from astrophysics, our method can produce predictive distributions which are calibrated at all locations in feature space for any use case. 

ABSTRACT Studies of cosmology, galaxy evolution, and astronomical transients with current and next-generation wide-field imaging surveys like the Rubin Observatory Legacy Survey of Space and Time are all critically dependent on estimates of photometric redshifts. Capsule networks are a new type of neural network architecture that is better suited for identifying morphological features of the input images than traditional convolutional neural networks. We use a deep capsule network trained on ugriz images, spectroscopic redshifts, and Galaxy Zoo spiral/elliptical classifications of ∼400 000 Sloan Digital Sky Survey galaxies to do photometric redshift estimation. We achieve a photometric redshift prediction accuracy and a fraction of catastrophic outliers that are comparable to or better than current methods for SDSS main galaxy sample-like data sets (r ≤ 17.8 and zspec ≤ 0.4) while requiring less data and fewer trainable parameters. Furthermore, the decision-making of our capsule network is much more easily interpretable as capsules act as a low-dimensional encoding of the image. When the capsules are projected on a two-dimensional manifold, they form a single redshift sequence with the fraction of spirals in a region exhibiting a gradient roughly perpendicular to the redshift sequence. We perturb encodings of real galaxy images in this low-dimensional space to create synthetic galaxy images that demonstrate the image properties (e.g. size, orientation, and surface brightness) encoded by each dimension. We also measure correlations between galaxy properties (e.g. magnitudes, colours, and stellar mass) and each capsule dimension. We publicly release our code, estimated redshifts, and additional catalogues at https://biprateep.github.io/encapZulate-1. 

ABSTRACT The field of microbial ecology, evolution, and biodiversity (EEB) is at the leading edge of understanding how microbes shape our biosphere and influence the well-being of humankind and Earth. To that end, EEB is developing new transdisciplinary tools to analyze these ecologically critical, complex microbial communities. The American Society for Microbiology’s Council on Microbial Sciences hosted a virtual retreat in 2023 to discuss the trajectory of EEB both within the Society and microbiology writ large. The retreat emphasized the interconnectedness of microbes and their outsized global influence on environmental and host health. The maximal potential impact of EEB will not be achieved without contributions from disparate fields that unite diverse technologies and data sets. In turn, this level of transdisciplinary efforts requires actively encouraging “broad” research, spanning inclusive global collaborations that incorporate both scientists and the public. Together, the American Society for Microbiology and EEB are poised to lead a paradigm shift that will result in a new era of collaboration, innovation, and societal relevance for microbiology. 

Many astrophysical analyses depend on estimates of redshifts (a proxy for distance) determined from photometric (i.e., imaging) data alone. Inaccurate estimates of photometric redshift uncertainties can result in large systematic errors. However, probability distribution outputs from many photometric redshift methods do not follow the frequentist definition of a Probability Density Function (PDF) for redshift -- i.e., the fraction of times the true redshift falls between two limits z1 and z2 should be equal to the integral of the PDF between these limits. Previous works have used the global distribution of Probability Integral Transform (PIT) values to re-calibrate PDFs, but offsetting inaccuracies in different regions of feature space can conspire to limit the efficacy of the method. We leverage a recently developed regression technique that characterizes the local PIT distribution at any location in feature space to perform a local re-calibration of photometric redshift PDFs. Though we focus on an example from astrophysics, our method can produce PDFs which are calibrated at all locations in feature space for any use case. 

Abstract We present a high-significance cross-correlation of CMB lensing maps from the Atacama Cosmology Telescope (ACT) Data Release 6 (DR6) with luminous red galaxies (LRGs) from the Dark Energy Spectroscopic Instrument (DESI) Legacy Survey spectroscopically calibrated by DESI. We detect this cross-correlation at a significance of 38σ; combining our measurement with thePlanck Public Release 4 (PR4) lensing map, we detect the cross-correlation at 50σ. Fitting this jointly with the galaxy auto-correlation power spectrum to break the galaxy bias degeneracy withσ₈, we perform a tomographic analysis in four LRG redshift bins spanning 0.4 ≤z≤ 1.0 to constrain the amplitude of matter density fluctuations through the parameter combinationS₈^×=σ₈(Ω_m/ 0.3)^0.4. Prior to unblinding, we confirm with extragalactic simulations that foreground biases are negligible and carry out a comprehensive suite of null and consistency tests. Using a hybrid effective field theory (HEFT) model that allows scales as small ask_max= 0.6 h/ Mpc, we obtain a 3.3% constraint onS₈^×=σ₈(Ω_m/ 0.3)^0.4= 0.792^+0.024_-0.028from ACT data, as well as constraints onS₈^×(z) that probe structure formation over cosmic time.Our result is consistent with the early-universe extrapolation from primary CMB anisotropies measured byPlanck PR4 within 1.2σ. Jointly fitting ACT andPlanck lensing cross-correlations we obtain a 2.7% constraint ofS₈^×= 0.776^+0.019_-0.021, which is consistent with the Planck early-universe extrapolation within 2.1σ, with the lowest redshift bin showing the largest difference in mean. The latter may motivate further CMB lensing tomography analyses atz< 0.6 to assess the impact of potential systematics or the consistency of the ΛCDM model over cosmic time. 

Abstract We utilize ∼17,000 bright luminous red galaxies (LRGs) from the novel Dark Energy Spectroscopic Instrument Survey Validation spectroscopic sample, leveraging its deep (∼2.5 hr galaxy −1 exposure time) spectra to characterize the contribution of recently quenched galaxies to the massive galaxy population at 0.4 < z < 1.3. We use Prospector to infer nonparametric star formation histories and identify a significant population of recently quenched galaxies that have joined the quiescent population within the past ∼1 Gyr. The highest-redshift subset (277 at z > 1) of our sample of recently quenched galaxies represents the largest spectroscopic sample of post-starburst galaxies at that epoch. At 0.4 < z < 0.8, we measure the number density of quiescent LRGs, finding that recently quenched galaxies constitute a growing fraction of the massive galaxy population with increasing look-back time. Finally, we quantify the importance of this population among massive ( log ( M ⋆ / M ⊙ ) > 11.2) LRGs by measuring the fraction of stellar mass each galaxy formed in the gigayear before observation, f 1 Gyr . Although galaxies with f 1 Gyr > 0.1 are rare at z ∼ 0.4 (≲0.5% of the population), by z ∼ 0.8, they constitute ∼3% of massive galaxies. Relaxing this threshold, we find that galaxies with f 1 Gyr > 5% constitute ∼10% of the massive galaxy population at z ∼ 0.8. We also identify a small but significant sample of galaxies at z = 1.1–1.3 that formed with f 1 Gyr > 50%, implying that they may be analogs to high-redshift quiescent galaxies that formed on similar timescales. Future analysis of this unprecedented sample promises to illuminate the physical mechanisms that drive the quenching of massive galaxies after cosmic noon. 

Abstract We use time-resolved spectra from the Apache Point Observatory Galactic Evolution Experiment (APOGEE) to examine the distribution of radial velocity (RV) variations in 249 stars identified as members of the Sagittarius (Sgr) dwarf spheroidal (dSph) galaxy by Hayes et al. We select Milky Way (MW) stars that have stellar parameters ( log ( g ) , T eff , and [Fe/H] ) similar to those of the Sagittarius members by means of a k-d tree of dimension 3. We find that the shape of the distribution of RV shifts in Sgr dSph stars is similar to that measured in their MW analogs, but the total fraction of RV variable stars in the Sgr dSph is larger by a factor of ∼2. After ruling out other explanations for this difference, we conclude that the fraction of close binaries in the Sgr dSph is intrinsically higher than in the MW. We discuss the implications of this result for the physical processes leading to the formation of close binaries in dwarf spheroidal and spiral galaxies.