What are $Omh^2(z_1,z_2)$ and $Om(z_1,z_2)$ diagnostics telling us in light of $H(z)$ data?

TL;DR

This study evaluates two-point diagnostics using 36 $H(z)$ measurements from BAO and galaxy ages, finding tension with $ ext{Lambda}$CDM and highlighting systematic differences between methods, suggesting the need for better understanding before $H(z)$ data can be fully reliable.

Contribution

It provides a comprehensive analysis of two-point diagnostics on $H(z)$ data, emphasizing non-Gaussian uncertainties and systematic differences between measurement methods.

Findings

$ ext{Lambda}$CDM shows tension with $H(z)$ data.

Non-Gaussian uncertainties require median statistics.

Systematic differences between BAO and DA methods need further investigation.

Abstract

Two-point diagnostics $Om(z_i,z_j)$ and $Omh^2(z_i,z_j)$ have been introduced as an interesting tool for testing the validity of the $\Lambda$CDM model. Quite recently, Sahni, Shafieloo $\&$ Starobinsky (2014) combined two independent measurements of $H(z)$ from BAO data with the value of the Hubble constant $H_0$, and used the second of these diagnostics to test the $\Lambda$CDM model. Their result indicated a considerable tension between observations and predictions of the $\Lambda$CDM model. Since reliable data concerning expansion rates of the Universe at different redshifts $H(z)$ are crucial for the successful application of this method, we investigate both two-point diagnostics on the most comprehensive set of $N=36$ measurements of $H(z)$ coming from the BAO and differential ages (DA) of passively evolving galaxies. We discuss the uncertainties of two-point diagnostics and find that they are strongly non-Gaussian and follow the patterns deeply rooted in their very construction. Therefore we propose that non-parametric median statistics is the most appropriate way of treating this problem. Our results support the claims that $\Lambda$CDM is in tension with $H(z)$ data according to the two-point diagnostics developed by Shafieloo, Sahni and Starobinsky. However, other alternatives to the $\Lambda$CDM, such as wCDM or CPL models perform even worse. We also notice that there are serious systematic differences between BAO and DA methods which ought to be better understood before $H(z)$ measurements can become competitive to the other probes.