Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userSPECULATOR: Emulating stellar population synthesis for fast and accurate galaxy spectra and photometry
78
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
We present SPECULATOR - a fast, accurate, and flexible framework for emulating stellar population synthesis (SPS) models for predicting galaxy spectra and photometry. For emulating spectra, we use principal component analysis to construct a set of basis functions, and neural networks to learn the basis coefficients as a function of the SPS model parameters. For photometry, we parameterize the magnitudes (for the filters of interest) as a function of SPS parameters by a neural network. The resulting emulators are able to predict spectra and photometry under both simple and complicated SPS model parameterizations to percent-level accuracy, giving a factor of $10^3$-$10^4$ speed up over direct SPS computation. They have readily-computable derivatives, making them amenable to gradient-based inference and optimization methods. The emulators are also straightforward to call from a GPU, giving an additional order-of-magnitude speed-up. Rapid SPS computations delivered by emulation offers a massive reduction in the computational resources required to infer the physical properties of galaxies from observed spectra or photometry and simulate galaxy populations under SPS models, whilst maintaining the accuracy required for a range of applications.
rate research
Read More
We present EzGal, a flexible python program designed to easily generate observable parameters (magnitudes, colors, mass-to-light ratios) for any stellar population synthesis (SPS) model. As has been demonstrated by various authors, the choice of input SPS models can be a significant source of systematic uncertainty. A key strength of EzGal is that it enables simple, direct comparison of different models sets. EzGal is also capable of generating composite stellar population models (CSPs) and can interpolate between metallicities for a given model set. We have created a web interface to run EzGal and generate observables for a variety of star formation histories and model sets. We make many commonly used SPS models available from this interface; the BC03 models, an updated version of these models, the Maraston models, the BaSTI models, and finally the FSPS models. We use EzGal to compare magnitude predictions for the model sets as a function of wavelength, age, metallicity, and star formation history. We recover the well-known result that the models agree best in the optical for old, solar metallicity models, with differences at the ~0.1 magnitude level. The most problematic regime for SPS modeling is for young ages (<2 Gyrs) and long wavelengths (lambda >7500 Angstroms) where scatter between models can vary from 0.3 mags (Sloan i) to 0.7 mags (Ks). We find that these differences are best understood as general uncertainties in SPS modeling. Finally we explore a more physically motivated example by generating CSPs with a star formation history matching the global star formation history of the universe. We demonstrate that the wavelength and age dependence of SPS model uncertainty translates into a redshift dependent model uncertainty, highlighting the importance of a quantitative understanding of model differences when comparing observations to models as a function of redshift.
To study the effect of supermassive black holes (SMBHs) on their host galaxies it is important to study the hosts when the SMBH is near its peak activity. A method to investigate the host galaxies of high luminosity quasars is to obtain optical spectra at positions offset from the nucleus where the relative contribution of the quasar and host are comparable. However, at these extended radii the galaxy surface brightness is often low (20-22 mag per arcsec$^{2}$) and the resulting spectrum might have such low S/N that it hinders analysis with standard stellar population modeling techniques. To address this problem we have developed a method that can recover galaxy star formation histories (SFHs) from rest frame optical spectra with S/N $sim$ 5~AA$^{-1}$. This method uses the statistical technique diffusion k-means to tailor the stellar population modeling basis set. Our diffusion k-means minimal basis set, composed of 4 broad age bins, is successful in recovering a range of galaxy SFHs. Additionally, using an analytic prescription for seeing conditions, we are able to simultaneously model scattered quasar light and the SFH of quasar host galaxies (QHGs). We use synthetic data to compare results of our novel method with previous techniques. We also present the modeling results on a previously published QHG and show that galaxy properties recovered from a diffusion k-means basis set are less sensitive to noise added to this quasar host galaxy spectrum. Our new method has a clear advantage in recovering information from QHGs and could also be applied to the analysis of other low S/N galaxy spectra such as those typically obtained for high redshift objects or integral field spectroscopic surveys.
GALAH is a large-scale magnitude-limited southern stellar spectroscopic survey. Its second data release (GALAH DR2) provides values of stellar parameters and abundances of 23 elements for 342,682 stars (Buder et al.). Here we add a description of the public release of radial velocities with a typical accuracy of 0.1 km/s for 336,215 of these stars, achievable due to the large wavelength coverage, high resolving power and good signal to noise ratio of the observed spectra, but also because convective motions in stellar atmosphere and gravitational redshift from the star to the observer are taken into account. In the process we derive medians of observed spectra which are nearly noiseless, as they are obtained from between 100 and 1116 observed spectra belonging to the same bin with a width of 50 K in temperature, 0.2 dex in gravity, and 0.1 dex in metallicity. Publicly released 1181 median spectra have a resolving power of 28,000 and trace the well-populated stellar types with metallicities between -0.6 and +0.3. Note that radial velocities from GALAH are an excellent match to the accuracy of velocity components along the sky plane derived by Gaia for the same stars. The level of accuracy achieved here is adequate for studies of dynamics within stellar clusters, associations and streams in the Galaxy. So it may be relevant for studies of the distribution of dark matter.
We present SPISEA (Stellar Population Interface for Stellar Evolution and Atmospheres), an open-source Python package that simulates simple stellar populations. The strength of SPISEA is its modular interface which offers the user control of 13 input properties including (but not limited to) the Initial Mass Function, stellar multiplicity, extinction law, and the metallicity-dependent stellar evolution and atmosphere model grids used. The user also has control over the Initial-Final Mass Relation in order to produce compact stellar remnants (black holes, neutron stars, and white dwarfs). We demonstrate several outputs produced by the code, including color-magnitude diagrams, HR-diagrams, luminosity functions, and mass functions. SPISEA is object-oriented and extensible, and we welcome contributions from the community. The code and documentation are available on GitHub and ReadtheDocs, respectively.
This paper is the 4th in a series describing the latest additions to the BaSTI stellar evolution database, which consists of a large set of homogeneous models and tools for population synthesis studies. Here we present a new set of low and high resolution synthetic spectra based on the BaSTI stellar models, covering a large range of simple stellar populations (SSPs) for both scaled solar and alpha-enhanced metal mixtures. This enables a completely consistent study of the photometric and spectroscopic properties of both resolved and unresolved stellar populations, and allows us to make detailed tests on their integrated properties. Our low resolution spectra are suitable for deriving broadband magnitudes and colors in any photometric system. These spectra cover the full wavelength range (9-160000nm) and include all evolutionary stages up to the end of AGB evolution. Our high resolution spectra are suitable for studying the behaviour of line indices and we have tested them against a large sample of Galactic globular clusters. We find that the range of ages, iron abundances [Fe/H], and degree of alpha-enhancement predicted by the models matches observed values very well. We have also tested the global consistency of the BaSTI models by making detailed comparisons between ages and metallicities derived from isochrone fitting to observed CMDs, and from line index strengths, for the Galactic globular cluster 47Tuc and the open cluster M67. For 47Tuc we find reasonable agreement between the 2 methods, within the estimated errors. From the comparison with M67 we find non-negligible effects on derived line indices caused by statistical fluctuations, which are a result of the specific method used to populate an isochrone and assign appropriate spectra to individual stars. (abridged)
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
