Implications of two type Ia supernova populations for cosmological measurements

Recent work suggests that Type Ia supernovae (SNe) are composed of two distinct populations: prompt and delayed. By explicitly incorporating properties of host galaxies, it may be possible to target and eliminate systematic differences between these two putative populations. However, any resulting post-calibration shift in luminosity between the components will cause a redshift-dependent systematic shift in the Hubble diagram. Utilizing an existing sample of 192 SNe Ia, we find that the average luminosity difference between prompt and delayed SNe is constrained to be (4.5 +/- 8.9)%. If the absolute difference between the two populations is 0.025 mag, and this is ignored when fitting for cosmological parameters, then the dark energy equation of state (EOS) determined from a sample of 2300 SNe Ia is biased at similar to 1 sigma. By incorporating the possibility of a two-population systematic, this bias can be eliminated. However, assuming no prior on the strength of the two-population effect, the uncertainty in the best-fit EOS is increased by a factor of 2.5, when compared to the equivalent sample with no underlying two-population systematic. To avoid introducing a bias in the EOS parameters, or significantly degrading the measurement accuracy, it is necessary to control the postcalibration luminosity difference between prompt and delayed SN populations to better than 0.025 mag.