Cosmological Tests with the Joint Lightcurve Analysis

TL;DR

This study compares cosmological models using Type Ia supernova data from the JLA catalog, revealing how model selection varies with data and highlighting potential systematic offsets between sources.

Contribution

It provides a detailed comparison of wCDM, LCDM, and R_h=ct models using combined supernova data, emphasizing the impact of data calibration and model assumptions.

Findings

Bayes Information Criterion favors R_h=ct over wCDM in some cases.

Systematic offset of ~0.04-0.08 mag observed between SNLS and SDSS-II.

Model selection results are sensitive to data calibration and assumptions about curvature.

Abstract

We examine whether a comparison between wCDM and R_h=ct using merged Type Ia SN catalogs produces results consistent with those based on a single homogeneous sample. Using the Betoule et al. (2014) joint lightcurve analysis (JLA) of a combined sample of 613 events from SNLS and SDSS-II, we estimate the parameters of the two models and compare them. We find that the improved statistics can alter the model selection in some cases, but not others. In addition, based on the model fits, we find that there appears to be a lingering systematic offset of ~0.04-0.08 mag between the SNLS and SDSS-II sources, in spite of the cross-calibration in the JLA. Treating wCDM, LCDM and R_h=ct as separate models, we find in an unbiased pairwise statistical comparison that the Bayes Information Criterion (BIC) favors the R_h=ct universe with a likelihood of 82.8% versus 17.2% for wCDM, but the ratio of likelihoods is reversed (16.2% versus 83.8%) when w_de=-1 (i.e., LCDM) and strongly reversed (1.0% versus 99.0%) if in addition k=0 (i.e., flat LCDM). We point out, however, that the value of k is a measure of the net energy (kinetic plus gravitational) in the Universe and is not constrained theoretically, though some models of inflation would drive k to 0 due to an expansion-enforced dilution. Since we here consider only the basic LCDM model, the value of k needs to be measured and, therefore, the pre-assumption of flatness introduces a significant bias into the BIC.