ABSTRACT
We present dynesty, a public, open-source, python package to estimate Bayesian posteriors and evidences (marginal likelihoods) using the dynamic nested sampling methods developed by Higson ...et al. By adaptively allocating samples based on posterior structure, dynamic nested sampling has the benefits of Markov chain Monte Carlo (MCMC) algorithms that focus exclusively on posterior estimation while retaining nested sampling’s ability to estimate evidences and sample from complex, multimodal distributions. We provide an overview of nested sampling, its extension to dynamic nested sampling, the algorithmic challenges involved, and the various approaches taken to solve them in this and previous work. We then examine dynesty’s performance on a variety of toy problems along with several astronomical applications. We find in particular problems dynesty can provide substantial improvements in sampling efficiency compared to popular MCMC approaches in the astronomical literature. More detailed statistical results related to nested sampling are also included in the appendix.
Stellar Population Inference with Prospector Johnson, Benjamin D.; Leja, Joel; Conroy, Charlie ...
The Astrophysical journal. Supplement series,
06/2021, Letnik:
254, Številka:
2
Journal Article
Recenzirano
Odprti dostop
Abstract
Inference of the physical properties of stellar populations from observed photometry and spectroscopy is a key goal in the study of galaxy evolution. In recent years, the quality and ...quantity of the available data have increased, and there have been corresponding efforts to increase the realism of the stellar population models used to interpret these observations. Describing the observed galaxy spectral energy distributions in detail now requires physical models with a large number of highly correlated parameters. These models do not fit easily on grids and necessitate a full exploration of the available parameter space. We present
Prospector
, a flexible code for inferring stellar population parameters from photometry and spectroscopy spanning UV through IR wavelengths. This code is based on forward modeling the data and Monte Carlo sampling the posterior parameter distribution, enabling complex models and exploration of moderate dimensional parameter spaces. We describe the key ingredients of the code and discuss the general philosophy driving the design of these ingredients. We demonstrate some capabilities of the code on several data sets, including mock and real data.
We present a new three-dimensional map of dust reddening, based on Gaia parallaxes and stellar photometry from Pan-STARRS 1 and 2MASS. This map covers the sky north of a decl. of −30°, out to a ...distance of a few kiloparsecs. This new map contains three major improvements over our previous work. First, the inclusion of Gaia parallaxes dramatically improves distance estimates to nearby stars. Second, we incorporate a spatial prior that correlates the dust density across nearby sightlines. This produces a smoother map, with more isotropic clouds and smaller distance uncertainties, particularly to clouds within the nearest kiloparsec. Third, we infer the dust density with a distance resolution that is four times finer than in our previous work, to accommodate the improvements in signal-to-noise enabled by the other improvements. As part of this work, we infer the distances, reddenings, and types of 799 million stars. (Our 3D dust map can be accessed at doi:10.7910/DVN/2EJ9TX or through the Python package dustmaps, while our catalog of stellar parameters can be accessed at doi:10.7910/DVN/AV9GXO. More information about the map, as well as an interactive viewer, can be found at argonaut.skymaps.info.) We obtain typical reddening uncertainties that are ∼30% smaller than those reported in the Gaia DR2 catalog, reflecting the greater number of photometric passbands that enter into our analysis.
Nonparametric star formation histories (SFHs) have long promised to be the "gold standard" for galaxy spectral energy distribution (SED) modeling as they are flexible enough to describe the full ...diversity of SFH shapes, whereas parametric models rule out a significant fraction of these shapes a priori. However, this flexibility is not fully constrained even with high-quality observations, making it critical to choose a well-motivated prior. Here, we use the SED-fitting code Prospector to explore the effect of different nonparametric priors by fitting SFHs to mock UV-IR photometry generated from a diverse set of input SFHs. First, we confirm that nonparametric SFHs recover input SFHs with less bias and return more accurate errors than do parametric SFHs. We further find that, while nonparametric SFHs robustly recover the overall shape of the input SFH, the primary determinant of the size and shape of the posterior star formation rate as a function of time (SFR(t)) is the choice of prior, rather than the photometric noise. As a practical demonstration, we fit the UV-IR photometry of ∼6000 galaxies from the Galaxy and Mass Assembly survey and measure scatters between priors to be 0.1 dex in mass, 0.8 dex in SFR100 Myr, and 0.2 dex in mass-weighted ages, with the bluest star-forming galaxies showing the most sensitivity. An important distinguishing characteristic for nonparametric models is the characteristic timescale for changes in SFR(t). This difference controls whether galaxies are assembled in bursts or in steady-state star formation, corresponding respectively to (feedback-dominated/accretion-dominated) models of galaxy formation and to (larger/smaller) confidence intervals derived from SED fitting. High-quality spectroscopy has the potential to further distinguish between these proposed models of SFR(t).
We present a uniform catalog of accurate distances to local molecular clouds informed by the Gaia DR2 data release. Our methodology builds on that of Schlafly et al. First, we infer the distance and ...extinction to stars along sightlines toward the clouds using optical and near-infrared photometry. When available, we incorporate knowledge of the stellar distances obtained from Gaia DR2 parallax measurements. We model these per-star distance-extinction estimates as being caused by a dust screen with a 2D morphology derived from Planck at an unknown distance, which we then fit for using a nested sampling algorithm. We provide updated distances to the Schlafly et al. sightlines toward the Dame et al. and Magnani et al. clouds, finding good agreement with the earlier work. For a subset of 27 clouds, we construct interactive pixelated distance maps to further study detailed cloud structure, and find several clouds which display clear distance gradients and/or are comprised of multiple components. We use these maps to determine robust average distances to these clouds. The characteristic combined uncertainty on our distances is 5%-6%, though this can be higher for clouds at greater distances, due to the limitations of our single-cloud model.
Accurate distances to local molecular clouds are critical for understanding the star and planet formation process, yet distance measurements are often obtained inhomogeneously on a cloud-by-cloud ...basis. We have recently developed a method that combines stellar photometric data with
Gaia
DR2 parallax measurements in a Bayesian framework to infer the distances of nearby dust clouds to a typical accuracy of ∼5%. After refining the technique to target lower latitudes and incorporating deep optical data from DECam in the southern Galactic plane, we have derived a catalog of distances to molecular clouds in Reipurth (2008, Star Formation Handbook, Vols. I and II) which contains a large fraction of the molecular material in the solar neighborhood. Comparison with distances derived from maser parallax measurements towards the same clouds shows our method produces consistent distances with ≲10% scatter for clouds across our entire distance spectrum (150 pc−2.5 kpc). We hope this catalog of homogeneous distances will serve as a baseline for future work.
Galaxy observations are influenced by many physical parameters: stellar masses, star formation rates (SFRs), star formation histories (SFHs), metallicities, dust, black hole activity, and more. As a ...result, inferring accurate physical parameters requires high-dimensional models that capture or marginalize over this complexity. Here we reassess inferences of galaxy stellar masses and SFRs using the 14-parameter physical model Prospector- built in the Prospector Bayesian inference framework. We fit the photometry of 58,461 galaxies from the 3D-HST catalogs at 0.5 < z < 2.5. The resulting stellar masses are ∼0.1-0.3 dex larger than the fiducial masses while remaining consistent with dynamical constraints. This change is primarily due to the systematically older SFHs inferred with Prospector. The SFRs are ∼0.1-1+ dex lower than UV+IR SFRs, with the largest offsets caused by emission from "old" (t > 100 Myr) stars. These new inferences lower the observed cosmic SFR density by ∼0.2 dex and increase the observed stellar mass growth by ∼0.1 dex, finally bringing these two quantities into agreement and implying an older, more quiescent universe than found by previous studies at these redshifts. We corroborate these results by showing that the Prospector- SFHs are both more physically realistic and much better predictors of the evolution of the stellar mass function. Finally, we highlight examples of observational data that can break degeneracies in the current model; these observations can be incorporated into priors in future models to produce new and more accurate physical parameters.
Abstract
We use the panchromatic spectral energy distribution (SED)-fitting code
Prospector
to measure the galaxy log
M
*–logSFR relationship (the
star-forming sequence
) across 0.2 <
z
< 3.0 using ...the COSMOS-2015 and 3D-HST UV-IR photometric catalogs. We demonstrate that the chosen method of identifying star-forming galaxies introduces a systematic uncertainty in the inferred normalization and width of the star-forming sequence, peaking for massive galaxies at ∼0.5 and ∼0.2 dex, respectively. To avoid this systematic, we instead parameterize the density of the full galaxy population in the log
M
*–logSFR–redshift plane using a flexible neural network known as a normalizing flow. The resulting star-forming sequence has a low-mass slope near unity and a much flatter slope at higher masses, with a normalization 0.2–0.5 dex lower than typical inferences in the literature. We show this difference is due to the sophistication of the
Prospector
stellar populations modeling: the nonparametric star formation histories naturally produce higher masses while the combination of individualized metallicity, dust, and star formation history constraints produce lower star formation rates (SFRs) than typical UV+IR formulae. We introduce a simple formalism to understand the difference between SFRs inferred from SED fitting and standard template-based approaches such as UV+IR SFRs. Finally, we demonstrate the inferred star-forming sequence is consistent with predictions from theoretical models of galaxy formation, resolving a long-standing ∼ 0.2–0.5 dex offset with observations at 0.5 <
z
< 3. The fully trained normalizing flow including a nonparametric description of
ρ
(
log
M
*
,
logSFR
,
z
)
is available online
20
20
https://github.com/jrleja/sfs_leja_trained_flow
to facilitate straightforward comparisons with future work.
There has been a long-standing factor-of-two tension between the observed star formation rate density and the observed stellar mass buildup after z ∼ 2. Recently, we have proposed that sophisticated ...panchromatic SED models can resolve this tension, as these methods infer systematically higher masses and lower star formation rates than standard approaches. In a series of papers, we now extend this analysis and present a complete, self-consistent census of galaxy formation over 0.2 < z < 3 inferred with the Prospector galaxy SED-fitting code. In this work, Paper I, we present the evolution of the galaxy stellar mass function using new mass measurements of ∼105 galaxies in the 3D-HST and COSMOS-2015 surveys. We employ a new methodology to infer the mass function from the observed stellar masses: instead of fitting independent mass functions in a series of fixed redshift intervals, we construct a continuity model that directly fits for the redshift evolution of the mass function. This approach ensures a smooth picture of galaxy assembly and makes use of the full, non-Gaussian uncertainty contours in our stellar mass inferences. The resulting mass function has higher number densities at a fixed stellar mass than almost any other measurement in the literature, largely owing to the older stellar ages inferred by Prospector. The stellar mass density is ∼50% higher than previous measurements, with the offset peaking at z ∼ 1. The next two papers in this series will present the new measurements of the star-forming main sequence and the cosmic star formation rate density, respectively.
We present a new technique to determine distances to major star-forming regions across the Perseus Molecular Cloud, using a combination of stellar photometry, astrometric data, and 12CO spectral-line ...maps. Incorporating the Gaia DR2 parallax measurements when available, we start by inferring the distance and reddening to stars from their Pan-STARRS1 and Two Micron All Sky Survey photometry, based on a technique presented by Green et al. and implemented in their 3D "Bayestar" dust map of three-quarters of the sky. We then refine their technique by using the velocity slices of a CO spectral cube as dust templates and modeling the cumulative distribution of dust along the line of sight toward these stars as a linear combination of the emission in the slices. Using a nested sampling algorithm, we fit these per-star distance-reddening measurements to find the distances to the CO velocity slices toward each star-forming region. This results in distance estimates explicitly tied to the velocity structure of the molecular gas. We determine distances to the B5, IC 348, B1, NGC 1333, L1448, and L1451 star-forming regions and find that individual clouds are located between 275 and 300 pc, with typical combined uncertainties of 5%. We find that the velocity gradient across Perseus corresponds to a distance gradient of about 25 pc, with the eastern portion of the cloud farther away than the western portion. We determine an average distance to the complex of 294 17 pc, about 60 pc further than the distance derived to the western portion of the cloud using parallax measurements of water masers associated with young stellar objects. The method we present is not limited to the Perseus Complex, but may be applied anywhere on the sky with adequate CO data in the pursuit of more accurate 3D maps of molecular clouds in the solar neighborhood and beyond.