Constraints on the dark energy using multiple observations : snare of principal component analysis

Abstract

We explore snares in determining the equation of state of dark energy ($\omega$) when one uses the so-called principal component analysis for multiple observations. We demonstrated drawbacks of principal component analysis in an earlier paper. We used the Hubble parameter data generated from a fiducial model using the so-called Chevallier-Polarski-Linder parameterization. We extend our previous consideration to multiple observations, the Hubble parameter and the luminosity distance. We find that the principal component analysis produces the almost constant $\omega$ even when a fiducial model is a rapidly varying $\omega$. Thus, resolution of dynamical property of $\omega$ through PCA is degraded especially when one fits to several observations.