ZTF SN Ia DR2: Cosmology-independent constraints on Type Ia supernova standardisation from supernova siblings

Basic Information

• Paper ID: 2406.01434
• Title: ZTF SN Ia DR2: Cosmology-independent constraints on Type Ia supernova standardisation from supernova siblings
• Authors: S. Dhawan, E. Mortsell, J. Johansson, A. Goobar, M. Rigault, M. Smith, K. Maguire, J. Nordin, G. Dimitriadis, P.E. Nugent, L. Galbany, J. Sollerman, T. de Jaeger, J.H. Terwel, Y.-L. Kim, U. Burgaz, G. Helou, J. Purdum, S.L. Groom, R. Laher, B. Healy
• Classification: astro-ph.CO (Cosmology), astro-ph.HE (High Energy Astrophysics)
• Publication Date: June 4, 2024
• Journal: Astronomy & Astrophysics (submitted)
• Paper Link: https://arxiv.org/abs/2406.01434

Abstract

This paper utilizes 25 pairs of Type Ia supernovae "siblings" (multiple Type Ia supernovae in the same host galaxy) observed by the Zwicky Transient Factory (ZTF) to infer the standardization relationship parameters α and β for Type Ia supernovae in a cosmology-independent manner. The study finds: when allowing diversity in the standardization relationship, α = 0.218 ± 0.055, β = 3.084 ± 0.312; assuming a single global standardization relationship, α = 0.228 ± 0.029, β = 3.160 ± 0.191. The intrinsic scatter of the sibling pair sample σ_int ≤ 0.10 (95% confidence level) is significantly smaller than the 0.22 ± 0.04 obtained by traditional methods. The β value is consistent with cosmological samples, but the α value is higher by approximately 2.5-3.5σ, a discrepancy primarily driven by low x₁ pairs, suggesting that fast and slow-declining Type Ia supernovae may have different width-luminosity relationship slopes.

Research Background and Motivation

Research Questions

1. Core Problem: Precisely determining the standardization relationship parameters α (width-luminosity relation) and β (color-luminosity relation) for Type Ia supernovae, which is crucial for using Type Ia supernovae as cosmic distance indicators.
2. Scientific Significance:
   • Type Ia supernovae are key tools for cosmological research, used to discover cosmic acceleration and measure dark energy and the Hubble constant
   • The standardization relationship reduces the peak brightness scatter of Type Ia supernovae from its natural state to approximately 15%
   • Current Type Ia supernova cosmology is limited by systematic errors; understanding the standardization relationship is essential for future Stage-IV dark energy missions
3. Limitations of Existing Methods:
   • Traditional Cosmological Fitting: Simultaneously fitting α, β, and cosmological parameters, affected by K-corrections, selection effects, redshift uncertainties, and Galactic extinction errors
   • Cross-Calibration Issues: Cross-calibration systematic errors from heterogeneous observational samples are significant
   • Model Dependence: Requires detailed simulation inputs to correct for selection effects
   • β Value Controversy: The β ≈ 3.04 inferred from cosmological samples is far lower than the Galactic dust R_B ≈ 4.1
4. Research Motivation:
   • Independent Verification: The sibling pair method provides cosmology-independent measurements of the standardization relationship
   • Systematic Error Elimination: Supernovae in the same galaxy share identical cosmological distance, peculiar velocity, and overall host galaxy properties, which automatically cancel in pairwise comparisons
   • Uniform Observations: ZTF provides uniform observations in a single photometric system, minimizing cross-calibration errors
   • Diversity Exploration: Can investigate whether the standardization relationship has a single value across the population or exhibits diversity

Core Contributions

1. Largest Uniform Sibling Pair Sample: Constructs the largest sample of Type Ia supernova sibling pairs observed with a single instrument to date (25 pairs from ZTF DR2)
2. Cosmology-Independent Standardization Parameter Measurement:
   • First simultaneous constraints on α and β and their scatter
   • α = 0.228 ± 0.029, β = 3.160 ± 0.191 (single relationship assumption)
   • Scatter upper limits: σ(α) ≤ 0.195, σ(β) ≤ 0.923 (95% C.L.)
3. Discovery of Extremely Low Intrinsic Scatter: Sibling pair comparison shows σ_int ≤ 0.10, approximately 50% lower than traditional Hubble residual methods
4. Discovery of Anomalous α Value: α value is 2.5-3.5σ higher than cosmological samples, primarily driven by low x₁ pairs, suggesting possible discontinuity in the width-luminosity relationship
5. Methodological Innovation: Develops a Bayesian inference framework that marginalizes over true x₁ and c values, correctly handling measurement uncertainties and intrinsic scatter

Detailed Methodology

Task Definition

Input: Multi-band light curves of two Type Ia supernovae in the same host galaxy Output: Standardization relationship parameters α, β and their scatters σ(α), σ(β), and intrinsic scatter σ_int Constraints: Both supernovae have identical distance modulus μ, therefore Δμ = 0

Core Method Principles

SALT2 Light Curve Fitting

Using the SALT2 model (Guy et al. 2010) to fit light curves, obtaining:

• m_B: B-band peak apparent magnitude
• x₁: light curve width parameter
• c: color parameter
• t₀: peak time

Standardization relationship:

μ = m_B + αx₁ - βc - M_B

Sibling Pair Constraint Equation

For two supernovae in the same galaxy with identical distance modulus:

Δμ = 0 = Δm_B + αΔx₁ - βΔc

where Δ denotes the difference in parameters between the two supernovae.

Likelihood Function Construction

Considering measurement uncertainties and intrinsic scatter, true parameter differences:

Δx₁ = Δx₁^obs + δx₁
Δc = Δc^obs + δc

Distance modulus difference:

Δμ = Δm_B + αΔx₁^obs - βΔc^obs + αδx₁ - βδc

Complete likelihood function (after marginalizing δx₁ and δc):

χ² = Σ[a - b²(c_α + c_β)/(c·c_α + c·c_β + c_α·c_β) + log(2π(c·c_α + c·c_β + c_α·c_β)/(c·c_α·c_β))]

where:

• a = (Δμ^obs)²/σ²_m
• b = Δμ^obs/σ²_m
• c = 1/σ²_m
• c_α = 1/(α²σ²_δx₁)
• c_β = 1/(β²σ²_δc)

Error terms include:

• σ_fit: SALT2 fitting error
• σ_int: intrinsic scatter
• σ_α·Δx₁, σ_β·Δc: scatter contributions from standardization relationships

Technical Innovations

1. Complete Uncertainty Propagation:
   • Correctly marginalizes uncertainties in true x₁ and c values
   • Includes scatter in α and β themselves
   • Terms in denominator involving α and β ensure proper likelihood normalization
2. Cosmology Independence:
   • No need to assume cosmological model
   • No peculiar velocity corrections required
   • Overall host galaxy properties (e.g., mass step) automatically cancel
3. Bayesian Inference Framework:
   • Uses PyMultiNest for parameter inference
   • Simultaneously fits five parameters: α, β, σ(α), σ(β), σ_int
   • Employs uninformative priors
4. Sample Selection Strategy:
   • Strict early observation requirements (observations at least 3 days before peak)
   • Parameter recovery verification through downsampling tests
   • Removal of objects with peak time uncertainty σ(t₀) > 2 days

Experimental Setup

Dataset

Sample Construction

1. Initial Selection:
   • Starting from ZTF spectroscopically confirmed Type Ia supernova catalog
   • Query other transient sources in the same host galaxy within 100 arcsecond radius
   • Remove sources with variability exceeding 60 days (excluding AGN, TDE)
2. Final Sample:
   • Spec subsample: 12 pairs (both spectroscopically classified)
   • Photo-spec subsample: 13 pairs (one spectroscopic, one photometric classification)
   • Total: 25 Type Ia supernova pairs
   • Redshift range: 0.01 < z < 0.1
3. Photometric Classification Verification:
   • Fit SN Ib/c, IIn, IIP templates
   • Retain only objects where Type Ia template provides better fit (Δχ² ≥ 5, typically Δχ² ≈ 50)
4. Quality Control:
   • Peak time error σ(t₀) ≤ 2 days
   • Observations in at least two bands
   • Observations at least 3 days before peak
   • Remove insufficiently sampled objects

Observational Characteristics

• Instrument: ZTF 48-inch Samuel Oschin Telescope
• Bands: g, r, i
• Survey Depth: ~20.5 magnitude
• Survey Cadence: 3 days for ZTF-I, 2 days for ZTF-II
• Special Objects: 3 "co-pixel" sibling pairs (angular separation smaller than pixel size)

Data Processing

1. Light Curve Construction: Using variant of IPAC forced photometry pipeline (Masci et al. 2019)
2. Galactic Extinction Correction:
   • Using extinction values from Schlafly & Finkbeiner (2011)
   • Adopting CCM extinction law (Cardelli et al. 1989)
   • R_V = 3.1 (Galactic standard value)
3. SALT2 Fitting:
   • Version: Updated by Taylor et al. (2021), implemented via sncosmo v2.1.0
   • Iterative fitting: First iteration determines peak time, second iteration fits within -10 to +40 days phase range
   • Includes model covariance matrix
4. Host Galaxy Mass:
   • Using g-i color and i-band absolute magnitude
   • Taylor et al. (2011) relation: log(M/M_⊙) = 1.15 + 0.7(m_g - m_i) - 0.4M_i

Evaluation Metrics

1. Standardization Parameters: α (width-luminosity), β (color-luminosity)
2. Scatter: σ(α), σ(β)
3. Intrinsic Scatter: σ_int
4. Statistical Significance: Evaluated using confidence intervals (C.L.)

Comparison Methods

1. Cosmological Samples:
   • Pantheon+ (Brout et al. 2022a)
   • DES (Vincenzi et al. 2024)
   • ZTF DR2 complete sample
2. Historical Sibling Pair Studies:
   • Burns et al. (2020): 12 heterogeneous pairs
   • Biswas et al. (2022): Single pair β constraint
   • Scolnic et al. (2020, 2022)
3. Traditional Hubble Residual Method: Cosmology-dependent method applied to same sample for comparison

Experimental Results

Main Results

Standardization Parameter Measurements

1. Fitting with Scatter Allowed (Full Sample):

• α = 0.218 ± 0.055
• β = 3.084 ± 0.312
• σ(α) ≤ 0.195 (95% C.L.)
• σ(β) ≤ 0.923 (95% C.L.)
• σ_int ≤ 0.103 (95% C.L.)

Key Finding: Although scatter median is large, it is consistent with zero scatter within 1σ range.

2. Single Standardization Relationship Fitting (Full Sample):

• α = 0.228 ± 0.029
• β = 3.160 ± 0.191
• σ_int ≤ 0.088 (95% C.L.)

3. Subsample Results:

Both subsamples show high consistency, supporting combined analysis.

Comparison with Cosmological Samples

β Value:

• Sibling pairs: β = 3.160 ± 0.191
• Pantheon+: β = 3.04 ± 0.04
• Conclusion: <1σ agreement

α Value:

• Sibling pairs: α = 0.228 ± 0.029
• Pantheon+: α ≈ 0.14-0.16
• DES: α ≈ 0.15
• Conclusion: 2.3-3σ higher

Intrinsic Scatter:

• Sibling pair method: σ_int ≤ 0.10
• Traditional Hubble residual: σ_int = 0.22 ± 0.04
• Improvement: Approximately 50% scatter reduction

Host Galaxy Mass Dependence

Splitting sample at log(M*/M_⊙) = 10.57:

• Low-mass subsample: α and β consistent with high-mass sample within 1.5σ
• High-mass subsample: β value corresponds to R_V ≈ 2.2, although 3σ higher than Brout & Scolnic (2021), remains reasonable when considering scatter
• Conclusion: No strong evidence for mass dependence

Ablation Studies

x₁ Discontinuity Analysis

Introducing discontinuity at x₁ = -0.49 (based on ZTF DR2 overall analysis):

• Low x₁ (slow-declining): α_low = 0.274 ± 0.045
• High x₁ (fast-declining): α_high = 0.133 ± 0.072
• Significance: Substantial difference between α values

Robustness Testing:

• Varying discontinuity point from -1 to 0
• Including median -0.27
• Results remain consistent

Physical Interpretation: Suggests fast and slow-declining Type Ia supernovae may have different width-luminosity relationship slopes.

SALT2 Version Comparison

• Using older SALT2 version from Betoule et al. (2014)
• Inferred parameters consistent with Taylor et al. (2021) version
• Conclusion: Results robust to SALT2 version choice

Galactic R_V Sensitivity Test

For high Δc pair (ZTF20acehyxd+ZTF21abouuow, E(B-V)_MW = 0.463):

• Testing R_V from 2.5 to 3.5
• No significant change in inferred β value
• Conclusion: Insensitive to Galactic extinction assumptions

Case Studies

High Δx₁ Pair: ZTF18abdmgab + ZTF20abqefja

• Δx₁ = 3.537 (maximum)
• Δc = 0.014 (minimum)
• Contribution: Primarily constrains α parameter
• Light Curve Features: Distinctly different decline rates, significant differences in r-band shoulder and i-band secondary maximum timing

High Δc Pair: From Biswas et al. (2022)

• High Δc, low Δx₁
• Contribution: Primarily constrains β parameter
• Orthogonality: Forms orthogonal constraints with high Δx₁ pair (shown in Figure 6)

Co-pixel Pair: ZTF20abatows + ZTF20abcawtk

• Angular separation: 9.7 arcseconds
• Peak time difference: ~5 days (smallest in sample)
• Characteristics: Simultaneously detectable, extremely similar local environment

Experimental Findings

1. Extremely Low Intrinsic Scatter: Sibling pair comparison method significantly reduces scatter, validating Burns et al. (2020) findings
2. Systematically High α Value: Possible causes:
   • Selection effects favoring low x₁ supernovae
   • Non-linearity in width-luminosity relationship
   • Sample statistical fluctuations
3. β Value Consistency: Supports color-luminosity relationship in current cosmological analyses
4. Scatter Upper Limits: Although median is large, consistent with zero scatter within 1σ, suggesting possible single standardization relationship
5. Parameter Correlations: Correlation coefficient between Δm_B and Δc (r=0.42) significantly higher than with Δx₁ (r=-0.045), explaining stronger β constraints
6. Mass Distribution Bias: Sibling pair sample biased toward high-mass galaxies (expected, larger galaxies more likely to produce multiple supernovae)

Related Work

Type Ia Supernova Cosmology

1. Landmark Studies:
   • Riess et al. (1998), Perlmutter et al. (1999): Discovery of cosmic acceleration
   • Phillips (1993): Width-luminosity relation
   • Tripp (1998): Color-luminosity standardization
2. Modern Large Samples:
   • Pantheon+ (Brout et al. 2022a): ~1700 Type Ia supernovae
   • DES 5-year results (Vincenzi et al. 2024)
   • ZTF DR2 (Rigault et al. in prep): ~3000 Type Ia supernovae

Sibling Pair Research History

1. Early Studies:
   • Hamuy et al. (1991): Single or small samples
   • Stritzinger et al. (2010): 3 normal + 1 peculiar supernovae in NGC 1316
2. Systematic Studies:
   • Anderson & Soto (2013), Kelsey (2023): Historical sibling pair compilations
   • Burns et al. (2020): 12 heterogeneous pairs, discovering smaller scatter
   • Scolnic et al. (2020, 2022): No scatter difference found
3. Single Pair In-Depth Analysis:
   • Gall et al. (2018): SN2007on and SN2011iv, optical and near-infrared distance discrepancies
   • Biswas et al. (2022): First β = 3.5 ± 0.3 inference from single pair

Standardization Relationship Diversity

1. Host Galaxy Dependence:
   • Kelly et al. (2010), Sullivan et al. (2010): Mass step
   • Brout & Scolnic (2021): Environmental dependence of β and R_V
   • Rigault et al. (2020), Briday et al. (2022): Relationship with progenitor age
2. Local Environment Effects:
   • Roman et al. (2018), Kelsey et al. (2021): Local properties influence
   • This study: Overall mass effects cancel in sibling pairs, but local effects may remain

Selection Effects and Systematic Errors

1. Simulation Corrections:
   • Kessler et al. (2019), Popovic et al. (2021): Selection effect simulations
   • Scolnic & Kessler (2016): Population models
2. Cross-Calibration:
   • Scolnic et al. (2022), Brout et al. (2022b): Important systematic error source
   • This study advantage: Single photometric system

Relative Advantages of This Work

1. Largest Uniform Sample: 25 pairs, all from ZTF
2. Simultaneous α and β Constraints: Previous work only constrained β
3. Cosmology Independence: Significantly reduced systematic errors
4. Scatter Quantification: First quantification of standardization relationship diversity

Conclusions and Discussion

Main Conclusions

1. Standardization Parameter Measurement:
   • Cosmology-independently measured α = 0.228 ± 0.029 and β = 3.160 ± 0.191
   • β value shows excellent agreement with cosmological samples, validating current standardization methods
   • α value systematically 2.5-3.5σ higher, possibly indicating complexity in width-luminosity relationship
2. Scatter Constraints:
   • Standardization relationship scatter upper limits: σ(α) ≤ 0.195, σ(β) ≤ 0.923
   • Consistent with single standardization relationship within 1σ range
   • Intrinsic scatter σ_int ≤ 0.10, approximately 50% lower than traditional methods
3. Physical Implications:
   • x₁ discontinuity analysis suggests fast and slow-declining supernovae may have different α values
   • β ≈ 3 corresponds to R_V ≈ 2, significantly lower than Galactic value, supporting diverse dust properties
4. Method Validation: Sibling pair method is a robust, independent tool for constraining standardization relationships

Limitations

1. Sample Size Constraints:
   • 25 pairs limit statistical precision
   • Few high Δx₁ and high Δc pairs (approximately 2-3 each)
   • α constraint precision inferior to β
2. Selection Effects:
   • Sibling pairs biased toward high-mass galaxies
   • x₁ distribution differs from DR2 overall (KS test p=0.014)
   • Possible unidentified selection biases
3. Residual Systematic Errors:
   • Local environment effects not fully canceled
   • Photometric classification sample contamination risk
   • Early observation requirements may introduce bias
4. Uncertainty in α Anomaly:
   • Physical origin of 2.5-3.5σ deviation unclear
   • May be statistical fluctuation
   • Larger sample needed to confirm x₁ discontinuity
5. Redshift Range:
   • Limited to z < 0.1
   • Cannot probe high-redshift evolution effects

Future Directions

1. LSST Prospects:
   • Expected ~800 sibling pairs
   • α and β uncertainties reducible by ~3 times
   • Can definitively confirm or refute x₁ discontinuity
2. Triplets and Multiplets:
   • Possible discovery of systems with >2 supernovae
   • Single galaxy simultaneously constraining α and β
   • Stronger internal consistency checks
3. Multi-Wavelength Extension:
   • Near-infrared observations (e.g., Dwomoh et al. 2023)
   • Joint UV-optical-near-infrared constraints
   • Better understanding of dust properties
4. Physical Interpretation:
   • Exploring physical origins of α and β diversity
   • Connecting progenitor properties (age, metallicity)
   • Improving explosion physics models
5. Method Improvements:
   • Including local environment parameters
   • Non-parametric standardization relationships
   • Machine learning approaches

In-Depth Evaluation

Strengths

1. Methodological Innovation:
   • Cosmology-independent constraints represent important methodological contribution
   • Complete Bayesian framework correctly handles uncertainties
   • Simultaneous scatter measurement is novel
   • Mathematical derivation for marginalizing true parameter values rigorous
2. Sample Quality:
   • Single instrument observations eliminate cross-calibration errors
   • Strict quality control standards
   • Detailed photometric classification verification
   • Uniform data processing pipeline
3. Experimental Sufficiency:
   • Multiple subsample analyses (spec/photo-spec, mass splitting)
   • Systematic exploration of x₁ discontinuity
   • Robustness tests for SALT2 version and Galactic R_V
   • Comparison with traditional Hubble residual method
4. Result Convincingness:
   • Excellent β value agreement with cosmological samples enhances credibility
   • Intrinsic scatter reduction finding important and robust
   • Conservative uncertainty estimates (95% C.L. upper limits)
   • Subsample consistency supports conclusions
5. Writing Clarity:
   • Detailed, transparent method derivation
   • Information-rich figures (light curves, parameter distributions, constraint contours)
   • Comprehensive literature comparison
   • Honest limitation discussion

Weaknesses

1. Insufficient Explanation of α Anomaly:
   • 2.5-3.5σ deviation is important finding but physical explanation limited
   • x₁ discontinuity hypothesis needs more theoretical support
   • Possible systematic error sources insufficiently discussed
2. Sample Representativeness Issues:
   • x₁ distribution differs from DR2 but impact not deeply analyzed
   • High-mass galaxy bias systematic effects inadequately assessed
   • Lack of quantitative selection effect simulations
3. Statistical Testing:
   • Scatter-zero consistency only at 1σ, insufficient evidence
   • No Bayes factors comparing different models
   • Limited statistical power for subsample splitting
4. Missing Physical Models:
   • No attempt to explain results with physical models (progenitor models, dust models)
   • β-R_V connection discussion superficial
   • Lack of exploration of intrinsic color-luminosity relationship
5. Experimental Design:
   • Early observation requirement (3 days before peak) may introduce bias but not quantified
   • Photometric classification sample systematic error assessment insufficient
   • Lack of Monte Carlo simulations validating methods

Impact

1. Field Contribution:
   • High Importance: Provides independent verification for Type Ia supernova cosmology
   • Strong Timeliness: Methodological preparation before next-generation surveys like LSST
   • Pioneering: First large-sample sibling pair simultaneous α and β constraints
2. Practical Value:
   • Guidance for future survey design
   • Sibling pair method as systematic check for cosmological analysis
   • Practical significance of low scatter finding for precision cosmology
3. Reproducibility:
   • Detailed method description, complete mathematical derivation
   • Commitment to public data release via GitHub (ZTF DR2 subset)
   • Uses public software (sncosmo, PyMultiNest)
   • Rating: High
4. Follow-up Research:
   • Companion papers already available (Ginolin et al., Dimitriadis et al.)
   • Establishes foundation for LSST sibling pair science
   • May advance physical understanding of standardization relationships

Applicable Scenarios

1. Direct Applications:
   • Sibling pair analysis for time-domain surveys (ZTF, ATLAS, LSST)
   • Type Ia supernova cosmology systematic error assessment
   • Independent standardization relationship calibration
2. Method Transfer:
   • Sibling pair studies of other standard candles (Type II supernovae)
   • Host-internal comparisons of variable stars (Cepheids, RR Lyrae)
   • Multi-event gravitational wave source host studies
3. Limiting Conditions:
   • Requires long-term survey accumulating sufficient sibling pairs
   • Demands uniform high-quality photometric observations
   • Best suited for low-redshift samples z < 0.2
4. Extension Potential:
   • Incorporating spectroscopic information (progenitor type)
   • Multi-wavelength (UV to near-infrared)
   • Machine learning enhanced photometric classification

References

Key Citations

1. Methodological Foundation:
   • Guy et al. (2010): SALT2 model
   • Tripp (1998): Standardization relationship formula
   • Brout & Scolnic (2021): Environmental dependence of β
2. Sibling Pair Studies:
   • Burns et al. (2020): 12 heterogeneous pairs
   • Biswas et al. (2022): Single pair β constraint
   • Scolnic et al. (2020): LSST predictions
3. Cosmological Samples:
   • Brout et al. (2022a): Pantheon+
   • Vincenzi et al. (2024): DES 5-year
   • Riess et al. (1998), Perlmutter et al. (1999): Acceleration discovery
4. ZTF Related:
   • Graham et al. (2019): ZTF survey description
   • Rigault et al. (in prep): ZTF DR2
   • Smith et al. (in prep): Light curve processing

Overall Assessment: This is a high-quality observational astrophysics research paper with rigorous, innovative methodology and important, reliable results. While limited sample size and incomplete physical explanation of the α anomaly require further work, as pioneering research independently verifying Type Ia supernova standardization relationships, it possesses significant scientific value and methodological contributions. Particularly, the discovery of extremely low intrinsic scatter and independent β confirmation provide important support for current cosmological analysis. With the advent of next-generation surveys like LSST, this work establishes a solid foundation for sibling pair cosmology.