On the impact of the supernova subsamples in reducing the Hubble tension

Abstract The persistent 4–6 σ discrepancy between early- and late-time measurements of the Hubble constant ( H 0 ) is known as the “Hubble tension” and represents one of the major open problems in modern cosmology. In this work, we investigate how differences in light-curve parameters such as colour ( c ) and stretch ( x 1 ), and host galaxy properties such as the host stellar mass (M) and specific star formation rate (sSFR) between the calibration and Hubble Flow (HF) Type Ia supernovae (SNe Ia) samples used by SH0ES affect the SN luminosity standardization and the H 0 estimation. To do that, we generate subsamples from both samples and use them to estimate H 0 , M B , α , β , Δ host and σ int . Both one- and multi-dimensional Kolmogorov–Smirnov tests are used to evaluate the consistency between the subsamples property distributions and analyze how the estimated parameters vary with the better matching of the subsamples. We find that the calibration sample is not fully representative of the HF sample, particularly in the host's M and sSFR. Improving consistency between subsamples leads to changes in H 0 , M B and σ int as the samples become more consistent, although overall values remain broadly stable. More consistent subsamples also tend to produce a mass step consistent with zero within 1 σ . When disentangling SN subpopulations using different approaches, we identify persistent differences in H 0 (∼ 2–3 σ ) and M B (∼ 2 σ ) between low- and high-stretch SN Ia subpopulations: we find H 0 = 75.27 ± 1.18 km s -1 Mpc -1 for low-stretch and H 0 = 71.25 ± 1.59 km s -1 Mpc -1 for high-stretch, resulting in a Hubble tension of 6.07 σ and 2.52 σ , respectively. These differences highlight how multiple SNIa sub-populations likely present variations in dust properties and intrinsic colour that are not captured by the standard single β parameterization and that affect cosmology. By considering these two SN subpopulations when estimating a single Hubble parameter for all SNe, we obtain H 0 = 73.78 ± 2.17 km s -1 Mpc -1 , with a much larger uncertainty than normally quoted and resulting in decreasing the tension with the Planck measurement from 5.87 σ to about 2.86 σ . Moreover, our results also suggest that the mass step may arise from an over-correction of more than one SN subpopulation associated to different environments.