Type Ia Supernovae: Testing the Cosmological Principle and Improving Distance Measurements with Implications for the Hubble Constant

Type Ia supernovae (SN Ia) are standardizable candles and excellent distance indicators. For almost 25 years they have been used to measure the distance to extragalactic objects. When comparing distance versus velocity of distant objects, astronomers can put constraints on the content and dynamics of the universe.

A major assumption of the best fit cosmological model (Lambda-CDM) is that the universe is homogeneous and isotropic on large scales. That means there is no preferred location or direction. In relation to cosmology, isotropy means that at a sufficiently large scale, the relationship between distance and redshift should be uniform across the sky. That is, there should be no bulk flows at very large scales. Both theoretical and observational research in the past decade have begun to challenge this assumption.

In the first part of this dissertation, I re-analyze the detectability of this large scale bulk flows based upon the redshift-distance relation for SN Ia. These are commonly referred to as dark flows. This analysis uses a model based upon the angular dependence of the deviation from Hubble flow on the sky. I apply this analysis to the Union2.1 SN Ia data and to the SDSS-II supernova survey. Results for low redshift, z < 0.05, are consistent with previous searches. I find a local bulk flow of v_bf ~ 300 km/s in the direction of (l,b) ~ (270, 35) degrees. However, the search for a dark flow at z>0.05 is inconclusive.

Based upon simulated data sets, I deduce that the difficulty in detecting a dark flow at high redshifts arises primarily from the observational error in the distance modulus. Thus, even if it exists, a dark flow is not detectable at large redshifts with current SN Ia data sets. I estimate that a detection would require a SN Ia data set with both significant sky coverage, and a reduction in the effective distance modulus error from 0.2 mag to < 0.02 mag, at a redshift of 0.3. I estimate that a greatly expanded data sample of ~ 10^4 SN Ia might detect a dark flow as small as 300 km/s out to z = 0.3 even with a distance modulus error of 0.2 mag. This may be achievable in a next-generation large survey like the Large Synoptic Survey Telescope.

The second part of this dissertation focuses on improving the standardization of SN Ia. Since the mid 1990s, SN Ia have been corrected for light curve decline rate and photometric color. In the early 2000s, it started to become apparent that SN Ia peak luminosity was not just correlated with properties of the SN Ia but also with properties of the host galaxy. The residuals of the best fit cosmology should be the uncorrelated noise, but quantities like host stellar mass appear to be correlated with Hubble residuals.

I continue the effort to improve distance measurements by looking to see if the age of the local stellar environment trends with Hubble residuals. I use a Bayesian method to estimate the age of the supernova's local (and global) environment by matching observed SDSS SEDs to an FSPS stellar population. At ~8 Gyr, there appears to be a 0.114 +- 0.039 mag 'step' in the average Hubble residual. This step is seen in both the local environment age and the average age of the host galaxy. Using principal component analysis (PCA) on the SALT2 parameters, host stellar mass, and local environment age we see that a combination of light curve stretch (x1), host stellar mass, and local age have a significant (4.7-sigma) correlation with Hubble residual. I find that x1, host mass, and age should be combined into a single parameter that best corrects for residuals in the Hubble diagram. Because there is a difference in the average properties of galaxies in the Hubble flow and nearby hosts used to calibrate the luminosities of SN Ia, this parameter may have a strong impact on the measurement of the current Hubble constant (H0). There appears to be a ~1.3% overestimation of H0 using SN Ia, detected at a a ~2.5-sigma level. This effect is less then the current 1-sigma uncertainty of H0 and as such, does not relieve the full difference currently seen in the the measurement of H0 via the Cosmic Microwave Background radiation and with SN Ia.