Source-linked AI summary
SNANA: A Public Software Package for Supernova Analysis
Richard Kessler, Joseph P. Bernstein, David Cinabro, Benjamin Dilday, Joshua A. Frieman, Saurabh Jha, Stephen Kuhlmann, Gajus Miknaitis, Masao Sako, Matt Taylor, Jake Vanderplas
TL;DR
SN Ia analyses require reliable and reproducible tools because light-curve-fitting systematics limit cosmological-parameter determination. SNANA addresses this need with a public framework combining simulation, light-curve fitting, and cosmology fitting across models and surveys, while documenting important scope limitations.
Problem
Light-curve-fitting systematics limit cosmological-parameter determination, while differing software versions and analysis details complicate reproducible comparisons.
Method
SNANA combines a light-curve fitter, Monte Carlo simulation, and cosmology fitter using shared model implementations, common technical utilities, and adaptable survey descriptions.
Results
SNANA provides a public, adaptable framework incorporating existing models and supporting new or improved light-curve models across surveys.
Takeaways & Limitations
SNANA supports shared analyses of cosmological parameters, SN rates, non-Ia contamination, and future-survey optimization by researchers and collaborations.
Takeaways & Limitations
SNANA does not currently include a utility for converting non-Ia light curves into spectral surfaces, and its spectroscopic-selection efficiency extraction requires software outside SNANA.
Abstract
from arXiv · showhide
We describe a general analysis package for supernova (SN) light curves, called SNANA, that contains a simulation, light curve fitter, and cosmology fitter. The software is designed with the primary goal of using SNe Ia as distance indicators for the determination of cosmological parameters, but it can also be used to study efficiencies for analyses of SN rates, estimate contamination from non-Ia SNe, and optimize future surveys. Several SN models are available within the same software architecture, allowing technical features such as K-corrections to be consistently used among multiple models, and thus making it easier to make detailed comparisons between models. New and improved light-curve models can be easily added. The software works with arbitrary surveys and telescopes and has already been used by several collaborations, leading to more robust and easy-to-use code. This software is not intended as a final product release, but rather it is designed to undergo continual improvements from the community as more is learned about SNe. Below we give an overview of the SNANA capabilities, as well as some of its limitations. Interested users can find software downloads and more detailed information from the manuals at http://www.sdss.org/supernova/SNANA.html .
1. INTRODUCTION
SN Ia light curves can yield relative distance moduli with approximately 15% dispersion after accounting for shape, color, and luminosity correlations. SNANA was developed as a public, adaptable framework combining fitting, simulation, and cosmology tools to improve the availability and reproducibility of such analyses.
- Scientific motivation: Approximately 15% dispersion remains in SN Ia relative distance modulus estimates after correcting for light-curve shape, color, and luminosity correlations.Common fitting models combine template light curves and spectra with parameters such as stretch and color.
- Scientific motivation: Systematic uncertainties in light-curve fitting limit cosmological-parameter determination and complicate comparisons across groups using different software versions or analysis details.Differences in selection criteria and K-corrections can obscure comparisons between methods.
- SNANA framework: SNANA combines a light-curve fitter, Monte Carlo simulation, and cosmology fitter in a publicly available package.It incorporates existing light-curve models and supports adding new or improved models.
- SNANA framework: Each light-curve model is implemented for both fitting and simulation, keeping them synchronized and allowing new models to be added through new functions.Rest-frame models use a common K-correction utility based on spectral templates and filter bandpasses.
- SNANA framework: SNANA is designed to simulate arbitrary surveys through standard observing-condition descriptions and to add new surveys without code modifications.The survey-addition principle is explicitly stated as requiring no code modifications.
2. SIMULATING SN LIGHT CURVES
SNANA simulates supernova light curves using actual survey conditions, configurable light-curve and spectral models, and modeled detection and spectroscopic-selection efficiencies. It supports diverse surveys, future-survey optimization, infrared and photometric-redshift studies, and non-Ia contamination estimates, while retaining stated limitations in non-Ia modeling and efficiency calibration.
- Survey conditions and observing noise: SNANA’s fast light-curve simulation uses actual survey conditions instead of pixel-level image simulations, accounting for non-photometric conditions and enabling many systematic or survey-design variations.Survey conditions are stored in SIMLIB files, which can contain thousands of observation sequences for large fields.
- Survey conditions and observing noise: SIMLIB observations convert model magnitudes into fluxes using zeropoints, while CCD readout noise, sky noise, PSF aperture effects, non-Gaussian PSF tails, and zeropoint dispersion determine uncertainties.For the SDSS-II simulation, non-Gaussian PSF tails increase total noise by about 10% relative to a single-Gaussian approximation.
- Model generation and transformations: The simulation generates shape and color properties, rest-frame magnitudes, optional redshift evolution and intrinsic smearing, then applies K-corrections and SIMLIB-based conversion to observed fluxes and uncertainties.K-corrections can use Hsiao et al. templates or any user-selected spectral template.
- Extended simulation capabilities: SNANA supports infrared light curves through extended mlcs2k2 templates, host-galaxy photo-Z priors through HOSTLIB, and non-Ia contamination studies using spectral-template libraries.Non-Ia template diversity is expected to improve as more well-observed non-Ia light curves become available.
- Detection and selection efficiencies: A configurable trigger simulation reproduces discovery logic and supports estimates of discovery rates, while spectroscopic selection is empirically modeled from observed and simulated redshift distributions.The SDSS-II example requires detections in at least two gri-filter combinations and at two or more epochs.
- Detection and selection efficiencies: Spectroscopic-efficiency parameters are inferred by matching simulated and observed redshift distributions, but SNANA does not include the software needed to extract those parameters and precision is limited by the SN rate versus redshift.The simulation provides NSIM(z,x_sim) and NDATA(z), while extracting x_model requires separate software.
3. LIGHT CURVE FITTING
SNANA fits observed and simulated light curves with multiple models, supports photometric-redshift estimation, and provides diagnostic and interpolation tools for survey analyses.
- Core fitting: The fitter processes data and simulations identically, using shared algorithms for host-galaxy extinction and K-corrections across rest-frame models.
- Photometric redshifts: For spectroscopically unconfirmed events, SNANA offers constrained photo-Z and cosmology-photoZ fits, with the latter fitting both redshift and distance modulus.The constrained fit derives distance modulus under a user-fixed cosmology, while the cosmology-photoZ fit has five fitted parameters.
- Examples: Example mlcs2k2 fits cover optical g, r, i, z bands and infrared Y, J bands for simulated SNe Ia.
- Photometric redshifts: Including host-galaxy photo-Z information as a prior produces a smaller scatter in fitted-minus-true redshift for simulated DES SNe.
- Core fitting: The fitter minimizes data-model χ2 with minuit and can optionally marginalize parameter estimates, although marginalization increases CPU time.
- Derived measurements: SNANA generates interpolated fluxes and uncertainties at arbitrary epochs, accounting for covariances among model parameters.
4. COSMOLOGY FITTER
SNANA includes cosmology-fitting tools that infer standard cosmological parameters from fitted supernova distances and can incorporate external priors and covariance information.
- Basic fitter: The basic cosmology fitter reads redshifts and distance moduli from light-curve fits and determines w and ΩM within the Friedmann-Robertson-Walker framework.
- External information: Baryon Acoustic Oscillation and Cosmic Microwave Background measurements can provide priors, while peculiar-velocity covariances can be included.
- Extended models: A separate Monte Carlo Markov Chain fitter supports time-dependent dark energy and non-zero curvature models.
5. SOFTWARE OVERVIEW
SNANA is implemented as an extensible public software package with general analysis features and attention to processing speed for large supernova surveys.
- Implementation: SNANA is written primarily in C and Fortran, uses CERNLIB, and includes tools for comparing results, translating formats, plotting, sorting, generating SIMLIBs, and converting χ2 grids into contours.
- Extensibility: The package integrates model-specific features such as priors and marginalization in a manner intended for use with any light-curve model.
- Performance: The simulation generates more than 10 light curves per second, while fitting takes about 1 second per light curve for minimization and 10–100 seconds for marginalization.
6. CONCLUSION
SNANA has broad use across supernova analyses and is intended to improve through community contributions, while its current fitting models depend on external training procedures that cannot be duplicated.
- Adoption: About a dozen SN-related analyses currently use SNANA, with users including students, postdocs, and faculty.
- Limitation: The current SN fitting models rely on externally performed training procedures that cannot be duplicated, motivating development of turn-key training programs.