Hu Zhan (National Astronomical Observatories of China, China)
Non-parametric reconstruction is useful for constraining quantities that are potentially varying with time. A common practice is to replace the continuous function with an interpolation over a set of points or bins. This approach often involves many degrees of freedom, so that the constraints become too weak to be informative. In this presentation, we discuss priors that can help reduce the uncertainties of reconstructed Hubble parameters and dark energy equation of state. Specifically, we propose a physically motivated prior that requires the Hubble expansion rate to be a non-decreasing function of redshift. Tests using SDSS-III and SNLS supernova data show significant improvement with such a prior, and the reconstructed Hubble parameters are consistent with measurements from ages of passively evolving galaxies.