Kernel dependence of the Gaussian Process reconstruction of late Universe expansion history

Abstract: In this work, we discuss model-independent reconstruction of the expansion history of the late Universe. We use Gaussian Process Regression to reconstruct the evolution of various cosmological parameters such as H(z) and slow-roll parameter using observational data to train the GP model. We look at the GP reconstruction of these parameters using stationary and non-stationary kernel functions. We examine the effect of the choice of kernel functions on the reconstructions. We find that non-stationary kernels such as polynomial kernels might be a better choice for the reconstruction if the training data set is noisy (such as H(z) data) as it helps to avoid fitting the error in the data. We also look at the kernel dependence of other cosmological parameters such as the redshift of transition to the accelerated expansion. This has been achieved by reconstructing the derivatives of the expansion history (H(z)) such as the deceleration parameter/slow-roll parameter.