In this paper, we have constrained the neutrino mass and mass hierarchy in the $\Lambda$CDM cosmology with the neutrino mass hierarchy parameter $\Delta$, which represents different mass orderings, by using the {\it Planck} 2015 + BAO + SN + $H_{0}$ data set, together with the neutrino oscillation and neutrinoless double beta decay data. We find that the mass of the lightest neutrinos and the total neutrino mass are no more than $0.035$eV and $0.133$ eV at $95\%$ confidence level, respectively. Comparing the result of our joint analysis with that obtained using cosmological data alone, we find that, by adding the neutrino oscillation and neutrinoless double beta decay data, the tendency for normal hierarchy has increased a lot. By means of importance sampling, three other priors are taken into account, i.e., the flat logarithmic prior on the absolute value of the neutrino hierarchy parameter $\Delta$, the flat linear prior on the total neutrino mass $\Sigma m_{\nu}$, and the flat logarithmic prior on $\Sigma m_{\nu}$. We find that the preference for the normal hierarchy is in agreement whatever what kinds of priors we choose. Finally, we make a Bayesian model analysis about four priors and we find that flat-linear and the flat logarithmic priors on $\Sigma m_{\nu}$ are the most favored priors.

Relativistic hydrodynamics is a powerful tool to simulate the evolution of the quark gluon plasma (QGP) in relativistic heavy ion collisions. Using 10000 initial and final profiles generated from 2+1-d relativistic hydrodynamics VISH2+1 with MC-Glauber initial conditions, we train a deep neural network based on stacked U-net, and use it to predict the final profiles associated with various initial conditions, including MC-Glauber, MC-KLN and AMPT and TRENTo. A comparison with the VISH2+1 results shows that the network predictions can nicely capture the magnitude and inhomogeneous structures of the final profiles, and nicely describe the related eccentricity distributions $P(\varepsilon_n)$ (n=2, 3, 4). These results indicate that deep learning technique can capture the main features of the non-linear evolution of hydrodynamics, showing its potential to largely accelerate the event-by-event simulations of relativistic hydrodynamics.