Decoding the Early-Time Light Curves of Type Ia Supernovae. II. Population Parameters of One Thousand ZTF Supernovae

Technical Summary: Decoding the Early-Time Light Curves of Type Ia Supernovae. II. Population Parameters of One Thousand ZTF Supernovae

Problem Statement
Type Ia supernovae (SNe Ia) are standardizable cosmological distance indicators, yet their underlying explosion mechanisms and progenitor systems remain debated. While late-time light curves near peak brightness are well-characterized, early-time emission (within the first week) encodes critical information about the outer ejecta, the distribution of radioactive $^{56}\text{Ni}$, and potential interactions with circumstellar material (CSM) or companions. Historically, studies of early-time SNe Ia have been limited by small sample sizes, selection biases (e.g., Malmquist bias), and the aggregation of individually fitted parameters, which can introduce spurious correlations and systematics. Furthermore, the relationship between early-time rise morphology and the global light-curve stretch parameter ($x_1$) has not been rigorously quantified across a statistically complete population. This paper addresses the need for a robust, population-level characterization of early-time SNe Ia using a volume-complete sample to break degeneracies between explosion channels and progenitor properties.

Methodology
The authors apply a hierarchical Bayesian framework, previously developed in "Paper I" of this series, to a volume-complete sample of 972 spectroscopically normal SNe Ia from the Zwicky Transient Facility Data Release 2 (ZTF DR2, $z < 0.06$). This sample is an order of magnitude larger than previous datasets used for similar analyses.

Key methodological components include:

• Power-Law Rise Model: Early-time light curves are modeled using an empirical power-law rise ($F \propto (t - t_{\text{fl}})^\alpha$) up to a truncation threshold (30%, 40%, or 50% of peak flux). The model includes parameters for time of first light ($t_{\text{fl}}$), amplitude, and a filter-specific rise index ($\alpha$).
• Hierarchical Inference: Instead of fitting individual SNe and aggregating results, the framework simultaneously infers individual parameters and population-level hyperparameters (means, scatters, and covariances). This approach naturally down-weights sparsely sampled or noisy events, mitigating selection biases and Malmquist bias without requiring strict data cuts.
• Uncertainty Handling: The likelihood function incorporates a scaling factor ($\beta_d$) to correct for underestimated photometric uncertainties in the ZTF pipeline and propagates uncertainties from the peak time ($t_0$) inferred from full light-curve fits (SALT2).
• Correlation Analysis: The study investigates correlations between early-time parameters ($t_{\text{rise}}$, $\alpha$, color evolution $\alpha_g - \alpha_r$) and the SALT2 stretch parameter ($x_1$) using a Bayesian errors-in-variables regression model with a broken linear fit to identify structural breaks.
• Theoretical Comparison: Observed correlations are compared against synthetic light curves from two theoretical frameworks: 1D near-Chandrasekhar mass (near-$M_{\text{Ch}}$) models with varying density profiles and $^{56}\text{Ni}$ mixing (TURTLS code), and 2D sub-Chandrasekhar mass ($M_{\text{Ch}}$) double-detonation models with varying viewing angles.

Key Results

1. Population Parameters: The study provides robust constraints on the early-time population:

   • Mean rise time ($t_{\text{rise}}$): $18.55 \pm 0.08$ days.
   • Mean rise index in $r$-band ($\alpha_r$): $2.10 \pm 0.04$ (steeper than the fireball model $\alpha=2$).
   • Mean color evolution ($\alpha_g - \alpha_r$): $0.20 \pm 0.02$, indicating early-time flux evolves more steeply in $g$-band than $r$-band (bluer evolution).
   • Significant intrinsic scatter exists in all parameters ($\sigma_{t_{\text{rise}}} \approx 1.42$ days, $\sigma_{\alpha_r} \approx 0.48$).

2. Bifurcation with Stretch ($x_1$): The relationship between early-time morphology and light-curve stretch exhibits a distinct structural break near $x_1 \approx 0.2\text{--}0.5$:

   • High-stretch population ($x_1 > x_{\text{brk}}$): Shows clear trends where higher $x_1$ correlates with shallower rises (lower $\alpha$), more persistent blue colors (lower $\alpha_g - \alpha_r$), and a flattening of the $t_{\text{rise}}$ vs. $x_1$ correlation.
   • Low-stretch population ($x_1 < x_{\text{brk}}$): Lacks significant correlations between early-time parameters and $x_1$.
   • An "adjusted rise time" (measured from a fractional flux epoch rather than extrapolated first light) reveals a continuous, strong correlation with $x_1$ without a break, suggesting the break in standard $t_{\text{rise}}$ is driven by variations in the early flux evolution ($t_{X\%}$).

3. Anomalies and Peculiar Subclasses:

   • 03fg-like events: Exhibit significantly longer rise times and steeper rise indices, consistent with short-duration early flux excesses from ejecta-CSM interactions in violent double white dwarf mergers.
   • 02cx-like and 91bg-like events: Show shorter rise times and shallower indices, consistent with partial deflagrations and low ejecta masses.
   • Normal SNe Ia Anomalies: The authors identify normal SNe Ia with unusually long rise times (potentially short-duration excesses) and a subset of high-stretch, 91T-like events with near-linear rises ($\alpha \approx 1$). These linear risers appear to occupy the tail of the distribution of luminous, slow-evolving normal SNe rather than forming a distinct subclass, suggesting widespread outward $^{56}\text{Ni}$ mixing in the high-stretch population.

4. Model Comparison:

   • Near-$M_{\text{Ch}}$ Models: While capable of producing diverse light curves, 1D models with varying density and $^{56}\text{Ni}$ mixing fail to reproduce the observed broken correlations between early-time parameters and stretch.
   • Sub-$M_{\text{Ch}}$ Double-Detonations: 2D models with asymmetric ejecta and viewing angle effects successfully reproduce a broken correlation in color evolution ($\alpha_g - \alpha_r$) but fail to match the observed rise times (predicting shorter rises) and do not reproduce the break in the rise index relation.

Significance and Claims
The paper claims to provide the first comprehensive, population-level characterization of early-time SNe Ia using a volume-complete sample, moving beyond the limitations of small, biased samples. By employing a hierarchical Bayesian framework, the authors demonstrate that strict quality cuts can introduce selection biases (e.g., skewing samples toward high-stretch events) and that robust population inferences can be made even with sparsely sampled data.

The primary significance lies in the discovery of a dichotomy in the early-time behavior of SNe Ia based on light-curve stretch. The observed break suggests that the physical mechanisms driving early emission differ fundamentally between low- and high-stretch populations. Specifically, the prevalence of shallow rises and persistent blue colors in the high-stretch population points to widespread outward mixing of $^{56}\text{Ni}$ in the outer ejecta layers.

The authors conclude that current theoretical models (both 1D near-$M_{\text{Ch}}$ and 2D sub-$M_{\text{Ch}}$) are insufficient to fully reproduce the observed parameter space, particularly the specific correlations with stretch. They argue that multi-dimensional explosion models with more realistic progenitor setups and treatments of $^{56}\text{Ni}$ distribution and ejecta asymmetry are necessary to explain the observed diversity. The paper positions these empirical constraints as a critical benchmark for future theoretical developments and highlights the potential of upcoming large-scale surveys (e.g., Rubin Observatory) to further refine these population demographics.