Type Ia supernova growth-rate measurement with LSST simulations: intrinsic scatter systematics

TL;DR

This study forecasts $f\sigma_8$ constraints from a full 10-year LSST SN Ia survey by embedding a realistic peculiar velocity field and host-galaxy correlations in eight Uchuu-based realizations and evaluating four intrinsic-scatter models. It contrasts a simple Tripp-based framework with a BBC bias-correction approach to build the SN Ia Hubble diagram and extract PVs via maximum likelihood, incorporating a redshift-space damping parameter $\sigma_u$. The results show that achromatic and chromatic scatter models recover unbiased $f\sigma_8$ with roughly $11$–$14\%$ precision, while the dust-based BS21 model introduces non-Gaussian HD residuals that bias $f\sigma_8$ by about $-20$ to $-26\%$, highlighting the need for non-Gaussian treatments of residuals. Systematics are dominated by the damping parameter $\sigma_u$ (contributing around $6\%$ to the $f\sigma_8$ error), whereas BS21-parameter uncertainties have a minor impact, underscoring the value of improved redshift-space distortion modeling and robust bias-correction methods for LSST-era velocity cosmology.

Abstract

Measurement of the growth rate of structures ($\fsig$) with Type Ia supernovae (\sns) will improve our understanding of the nature of dark energy and enable tests of general relativity. In this paper, we generate simulations of the 10 year \sn\ dataset of the Rubin-LSST survey, including a correlated velocity field from a N-body simulation and realistic models of \sns\ properties and their correlations with host-galaxy properties. We find, similar to SN~Ia analyses that constrain the dark energy equation-of-state parameters $w_0w_a$, that constraints on $\fsig$ can be biased depending on the intrinsic scatter of \sns. While for the majority of intrinsic scatter models we recover $\fsig$ with a precision of $\sim13 - 14\%$, for the most realistic dust-based model, we find that the presence of non-Gaussianities in Hubble diagram residuals leads to a bias on $\fsig$ of about $\sim-20\%$. When trying to correct for the dust-based intrinsic scatter, we find that the propagation of the uncertainty on the model parameters does not significantly increase the error on $\fsig$. We also find that while the main component of the error budget of $\fsig$ is the statistical uncertainty ($>75\%$ of the total error budget), the systematic error budget is dominated by the uncertainty on the damping parameter, $σ_u$, that gives an empirical description of the effect of redshift space distortions on the velocity power spectrum. Our results motivate a search for new methods to correct for the non-Gaussian distribution of the Hubble diagram residuals, as well as an improved modeling of the damping parameter.

Figures (11)

• Figure 1: Distribution of host masses in the Uchuu UniverseMachine catalog (yellow) and of the SN Ia host masses from our LSST simulation (blue).
• Figure 2: HEALPix pixels representation of the LSST survey WFD program. Pixels that contain less than 500 and more than 1100 visits are cut. HEALPix pixels are then sampled by OpSimSummaryV2 to build the SNANA simulation input.
• Figure 3: Angular distribution of the simulated SN Ia hosts. The colormap represents their peculiar velocities on the line of sight.
• Figure 4: Results of the fit for $\sigma_u$ on true velocities from Uchuu simulated galaxies. The red points represent the result in the case where $f\sigma_8$ is fixed to the fiducial value of 1. The blue points represent the case where $f\sigma_8$ is fitted along with $\sigma_u$.
• Figure 5: Histograms of the Hubble diagram residuals (left) and velocity bias residuals (right). Results for the simple fit are in blue and results for the BBC fit are in red. We added in legend the standard deviation of the samples.