Source-linked AI summary
Galaxia: a code to generate a synthetic survey of the Milky Way
Sanjib Sharma, Joss Bland-Hawthorn, Kathryn V. Johnston, James Binney
TL;DR
Synthetic Milky Way catalogs are needed to interpret observations, test models, and assess survey capabilities, but flexible methods for sampling analytic and N-body galaxy models are limited. Galaxia provides a fast population-synthesis framework that generates continuous stellar catalogs from both model types, and finds reliable six-dimensional stellar-halo samples for N < 10^5 despite smoothing effects in other regimes.
Problem
Synthetic stellar catalogs support observational interpretation, model testing, and instrument assessment, motivating fast methods that convert analytic and N-body galaxy models into catalogs.
Method
Galaxia samples analytic and N-body Milky Way models, uses theoretical isochrones to generate stars, and preserves N-body phase-space density through locally adaptive sampling.
Results
Galaxia produces reliable six-dimensional stellar-halo samples for N < 10^5, while limited numerical resolution can wash out structures and induce smoothing in other regimes.
Takeaways & Limitations
The framework supports synthetic large-scale survey analysis, including stellar-halo structure studies, when sample size and observational errors remain within the reported reliability conditions.
Takeaways & Limitations
The code does not model photometric errors or extinction during star spawning, so these effects must be added during post-processing.
Abstract
from arXiv · showhide
We present here a fast code for creating a synthetic survey of the Milky Way. Given one or more color-magnitude bounds, a survey size and geometry, the code returns a catalog of stars in accordance with a given model of the Milky Way. The model can be specified by a set of density distributions or as an N-body realization. We provide fast and efficient algorithms for sampling both types of models. As compared to earlier sampling schemes which generate stars at specified locations along a line of sight, our scheme can generate a continuous and smooth distribution of stars over any given volume. The code is quite general and flexible and can accept input in the form of a star formation rate, age metallicity relation, age velocity dispersion relation and analytic density distribution functions. Theoretical isochrones are then used to generate a catalog of stars and support is available for a wide range of photometric bands. As a concrete example we implement the Besancon Milky Way model for the disc. For the stellar halo we employ the simulated stellar halo N-body models of Bullock & Johnston (2005). In order to sample N-body models, we present a scheme that disperses the stars spawned by an N-body particle, in such a way that the phase space density of the spawned stars is consistent with that of the N-body particles. The code is ideally suited to generating synthetic data sets that mimic near future wide area surveys such as GAIA, LSST and HERMES. As an application we study the prospect of identifying structures in the stellar halo with a simulated GAIA survey. We plan to make the code publicly available at http://galaxia.sourceforge.net.
1. INTRODUCTION
The introduction motivates faster, accurate stellar-catalog generation for billion-star surveys and identifies limitations in existing disc and halo models. The paper therefore presents methods for converting analytic and N-body galaxy models into continuous synthetic catalogs.
- Motivation: Synthetic stellar catalogs interpret observations, test galaxy-formation theories, evaluate instruments and systematics, and guide strategies for reducing measurement errors.These uses motivate improved catalog-generation methods.
- Motivation: Future LSST and GAIA surveys plan to measure over 1 billion stars, creating a need for faster and accurate methods and a workable Milky Way model.The introduction describes a dynamically consistent model as unavailable but a working framework as fundamental to progress.
- Existing-model limitations: The Besançon model has limited applicability to wide-area surveys because user-specified radial and angular step sizes can change results, while smoother multi-disc models increase computational cost.Its scale height and velocity dispersion also vary with age, motivating more flexible sampling.
- Stellar-halo modeling: A smooth analytic halo cannot test substructure detection or accommodate structures such as the Sagittarius dwarf stream, whereas N-body models suit complex halo substructures.Accretion events are expected to leave detectable stellar-halo substructures.
- Paper contribution: The paper presents fast and accurate methods to convert analytic and N-body galaxy models into synthetic stellar catalogs, enabling model testing and rapid evaluation of new models.The stated aim is to reduce catalog-generation burdens for modelers and facilitate comparisons among models from different groups.
- Paper contribution: Its analytic-model sampling scheme generates continuous stellar positions and ages rather than discrete age-specific discs, applying the Besançon model to the disc and the Shu (1969) distribution function to improve disc kinematics.The Shu distribution function describes non-circular motion in the disc plane.
2. METHODS
The method models the Galaxy as distinct stellar components with parameterized distributions of position, velocity, age, metallicity, and mass, while also providing phase-space sampling for N-body realizations. It incorporates locally adaptive density estimation but has limitations involving dynamical self-consistency, birth radii, observational effects, extinction, and white dwarfs.
- Analytic framework: The Galaxy is represented as distinct components whose stellar distribution functions depend on position, velocity, age, metallicity, and mass.Components include the thin disc, thick disc, stellar halo, and bulge.
- Analytic framework: The analytic model parameterizes star formation, the initial mass function, spatial structure, velocity distributions, and metallicity distributions through age- and position-dependent functions.Metallicity may follow a log-normal distribution with an age-dependent mean and dispersion, while velocities may use triaxial Gaussian or Shu distributions.
- Model limitations: The analytic model is not dynamically self-consistent by itself and lacks explicit dependence on stellar birth radii.Self-consistency can be imposed externally, while birth-radius dependence can be added through an initial cylindrical radius parameter Ri.
- Observational limitations: Star spawning excludes photometric errors and extinction, and the code does not model white dwarfs in its present version.Users can apply different observational error laws and extinction models; white dwarfs are generally faint and rare but may matter for some applications.
- N-body sampling: N-body particles are oversampled by dispersing spawned stars while preserving the underlying phase-space density through kernel density estimation and locally adaptive covariance metrics.The position-to-velocity scale ratio varies by about 3 orders of magnitude, demonstrating why a global metric is inappropriate.
- Observational limitations: Schlegel extinction maps overestimate Galactic-plane extinction by about 30% where AV > 0.5 mag, complicating recalibration of the extinction treatment.The cited reports cover limited regions, and a properly recalibrated map was not yet available.
3. A WORKING MODEL OF THE GALACTIC COMPONENTS
Galaxia adopts the Besançon model as a flexible working description of Galactic components, parameterizing their star formation, kinematics, density profiles, and stellar populations. The model includes warped and flared discs, a G2 bar-shaped bulge, and either simulated or smooth stellar-halo components.
- Global model: Galaxia adopts the Besançon model’s star formation rate, age–velocity relation, initial mass function, and density profiles as its main Galactic-model ingredients.The model uses analytic functional forms rather than detailed self-consistent dynamical modeling.
- Disc components: The thin and thick discs reproduce Besançon kinematics while using a radial velocity-dispersion prescription from Schönrich & Binney (2009).The thin-disc dispersion saturates at 0.862σR0,φ0,z0, with τmax = 10.0 and τmin = 0.1.
- Disc components: The thin and thick discs include warp and flare prescriptions, with φmax = 90.0, γwarp = 0.18 kpc−1, Rflare = 1.12R⊙, and γflare = 0.0054 kpc−1.Stars beyond Rwarp are displaced from the plane, while those beyond Rflare have increased scale heights.
- Bulge component: The bulge is represented by the Besançon G2 boxy triaxial Gaussian bar, while its kinematics are modeled simply to approximate existing observations.This approach is intended for studying systematic effects rather than providing a fully self-consistent dynamical model.
- Halo component: The stellar halo can be sampled from eleven Bullock & Johnston (2005) N-body models or from a smooth oblate power law with ǫ = 0.76 and nH = −2.44.The simulated models describe satellite accretion, while the smooth-halo parameters can be altered for alternative observational prescriptions.
- Stellar populations: The adopted power-law IMF spans 0.07 < m < 100, and recalibration for a solar radius of 8 kpc gives an SFR of 2.37 M⊙/yr.The stellar-halo local mass density was increased by 10% to match Besançon star counts at large distances.
4. TESTS AND RESULTS
Galaxia reproduces established Besançon star-count and local kinematic distributions, while its solar-neighborhood catalog agrees well with Hipparcos aside from a 33% star-count excess. Its sampling efficiency enables billion-star GAIA-style surveys within hours.
- Observational comparisons: Galaxia’s thick-disc and stellar-halo star counts perfectly agree with Besançon, while the thin disc has subtle intermediate- and faint-magnitude discrepancies.The thin-disc excess and shortage also appear along other directions, shifted toward fainter magnitudes.
- Solar-neighborhood tests: Galaxia’s simulated solar-neighborhood color-magnitude, absolute-magnitude, color, and distance distributions show good agreement with Hipparcos data.The comparison uses stars within 100 pc and identifies a slight radial excess at r = 45 pc associated with the Hyades cluster.
- Solar-neighborhood tests: 33% higher is the simulated catalog’s star count relative to Hipparcos, primarily because stellar multiplicity was excluded from the Hipparcos catalog.For the stated selection limits, Hipparcos contains about 1549 sources flagged as binary.
- Kinematic tests: The model provides a reasonable fit to all three GCS velocity-component distributions, although it overestimates the negative U tail for V < −60 km s−1.Lowering σr0 reduces the discrepancy but makes the U distribution sharper, so the authors retain the model given high Poisson errors.
- Computational performance: 4 billion stars can be generated in about 6.5 hours for an all-sky GAIA-style survey with V < 20.A 3 × 10^7-star survey covering 10,000 sq degree toward the north galactic pole takes 220 seconds.
5. APPLICATIONS
Galaxia applications model the sky distribution of giants and red clump stars in RAVE- and SDSS-style surveys, and assess GAIA’s ability to identify stellar-halo substructures in E−Lz space. The halo analysis finds BJ05 structures less clustered than GHBL results and evaluates reliability limits caused by star-spawning scatter and simulation resolution.
- Giant and red-clump distributions: Galaxia maps giant and red-clump fractions and their ratio across RAVE-style and SDSS-style surveys, with and without dust extinction.Giants satisfy Teff <5000K and log(g) < 3.3; red clumps additionally satisfy 4500K< Teff <5000K and 1.625 < log(L/L⊙) < 1.825.
- Giant and red-clump distributions: 0.3 to 0.4 is the reported increase in the overall red-clump-to-giant ratio when extinction is included in the RAVE-style survey.The ratio is expected to be lower where the giant fraction is higher because red clumps are less luminous, and red clumps occur only in metal-rich populations.
- Giant and red-clump distributions: 0.016 is the mean SDSS giant fraction without extinction, increasing to 0.034 with extinction, while the red-clump-to-giant ratio changes from 0.44 to 0.42.The extinction effect on the giant fraction is opposite to that found in the RAVE simulation.
- Identification of substructures in the stellar halo in E−Lz space with GAIA: The 20 most luminous BJ05 satellite systems, comprising about 92% of the stellar mass, are significantly less clustered in E−Lz space than lower-luminosity systems.Compared with GHBL, BJ05 shows less conservation of energy and angular momentum and more high-energy radial orbits.
- Identification of substructures in the stellar halo in E−Lz space with GAIA: N < 105 provides excellent reliability for six-dimensional stellar-halo sampling, while N = 106 is moderately reliable only beyond 50 kpc.Slight smoothing remains reliable where observational errors exceed the smoothing induced by star spawning; unresolved massive accretion events can reduce structure contrast.
6. DISCUSSION
The discussion frames Galaxia as a tool for testing survey capabilities while emphasizing future refinement toward more physically self-consistent Milky Way models. Planned extensions include improved dynamical and chemical evolution, detailed Galactic structure, and recovering dispersed star clusters with HERMES.
- Galaxia generates particular model realizations useful for testing instrument capabilities, correcting systematics, and making observational predictions.The longer-term goal is to refine or fine-tune the model as new data become available.
- Dynamical consistency is not guaranteed when varying the star formation rate and other Galaxia parameters.The discussion points to Bienayme et al. (1987) and Just & Jahreiß (2010) as a self-consistent scheme for calculating local scale height or ellipticity as a function of age.
- Chemical evolution is presently not self-consistent, although Galaxia can in principle accommodate radial mixing.Galaxia does not currently include models with radial mixing; proposed semi-analytic schemes combine chemical evolution with radial mixing.
- Galaxia lacks a detailed central bar and spiral arms, while its thick disc is treated as a constant-age population.The missing structural components await better constraints from the AAOmega survey of the central disc.
- Galaxia is planned for recovering star clusters that have long since dispersed within the Galaxy as part of the HERMES project.This application can be tested with a multi-dimensional group-finding algorithm, whose power was demonstrated in simulated dwarf galaxies.
7. CONCLUSIONS
The paper presents a flexible, fast framework and Galaxia code for generating smooth analytical- and N-body-based stellar catalogs to support current and future surveys. Its validation shows reliable halo sampling in specified regimes, while limited simulation resolution can wash out structures.
- Conclusions: The framework is designed to adapt to evolving galaxy-dynamics theory, stellar-synthesis libraries, and N-body or semi-analytic simulations while rapidly generating large data sets and multiple realizations.It is intended to support preparation for upcoming surveys.
- Conclusions: Galaxia’s analytical-model algorithm generates smoothly distributed stars across finite fields, avoiding the cumbersome decomposition of wide areas into many lines of sight.Earlier schemes produced catalogs along specific lines of sight, limiting their use for large wide-area catalogs.
- Conclusions: The N-body sampling algorithm assigns spawned stars according to the underlying phase space density of N-body particles, enabling detailed predictions of stellar-halo structures from Bullock & Johnston (2005) models.This extends the framework beyond analytical density models.
- Conclusions: Significant washout of structures results from limited numerical resolution in the simulated stellar halos, and BJ05 satellite particles are less clustered than reported by G´omez et al. (2010).The latter issue is identified as requiring future investigation.
- Conclusions: N < 105 marks the sample-size regime where Galaxia’s 6d-space stellar-halo results are very reliable, while slight smoothing is expected in other regimes when observational errors remain controlled.The stated example is a survey of RR Lyrae or BHB stars; BJ05 simulations remain suitable for deep large-scale stellar-halo surveys.
- Conclusions: The authors plan to release Galaxia publicly to enable consistent comparison and evaluation of present and future stellar surveys.The planned release location is http://galaxia.sourceforge.net.
APPENDIX ROTATIONAL KINEMATICS OF THE DISC
The disc’s azimuthal velocities are modeled either with a Gaussian including asymmetric-drift correction or with the more thorough Shu distribution function. For the Shu-based treatment, the angular-momentum distributions are solved iteratively for a flat rotation curve and interpolated across stellar ages.
- Azimuthal velocity distribution: The simplest azimuthal-velocity model is a Gaussian with an asymmetric-drift correction.A more thorough treatment assumes the Shu (1969) distribution function.
- Shu distribution function: 0.0?The target distributions R(L) and Σ(L) are used to iteratively solve for Σ(L) and σR(L) following Dehnen (1999).
- Shu distribution function: The Shu formalism derives the distribution of angular momentum L, and hence vφ = L/R, at a given radius R.The effective potential is defined as Φeff(R, L) = Φ(R) + L2/2R2.
- Rotation curve: 226.84 is the assumed circular speed for the flat rotation curve.The gravitational potential is computed from the specified rotation curve.
- Age dependence: σR depends on age, so σR(L) and Σ(L) are calculated for finite age sets and evaluated at other ages by nearest-neighbor interpolation.This age dependence is incorporated when calculating P(L, R).