Analysis of recent type Ia supernova data based on evolving dark energy models

TL;DR

This paper investigates recent type Ia supernova data using evolving dark energy models with changing equations of state, revealing that observed deviations from the LCDM model are due to calibration differences and data distribution biases.

Contribution

It introduces a double-transition model for dark energy's equation of state and analyzes calibration effects and biases in supernova data using MCMC methods.

Findings

Deviations from LCDM are due to calibration differences, not new data.

Biases in dark energy parameters are quantified using mock data analysis.

Peak distribution of arithmetic means is an unbiased estimator for bias detection.

Abstract

We study characters of recent type Ia supernova (SNIa) data using evolving dark energy models with changing equation of state parameter w. We consider sudden-jump approximation of w for some chosen redshift spans with double transitions, and constrain these models based on Markov Chain Monte Carlo (MCMC) method using the SNIa data (Constitution, Union, Union2) together with baryon acoustic oscillation A parameter and cosmic microwave background shift parameter in a flat background. In the double-transition model the Constitution data shows deviation outside 1 sigma from LCDM model at low (z < 0.2) and middle (0.2 < z < 0.4) redshift bins whereas no such deviations are noticeable in the Union and Union2 data. By analyzing the Union members in the Constitution set, however, we show that the same difference is actually due to different calibration of the same Union sample in the Constitution set, and is not due to new data added in the Constitution set. All detected deviations are within 2 sigma from the LCDM world model. From the LCDM mock data analysis, we quantify biases in the dark energy equation of state parameters induced by insufficient data with inhomogeneous distribution of data points in the redshift space and distance modulus errors. We demonstrate that location of peak in the distribution of arithmetic means (computed from the MCMC chain for each mock data) behaves as an unbiased estimator for the average bias, which is valid even for non-symmetric likelihood distributions.