A multiscale deep learning model is proposed for fluid flow in porous media. The method is formulated on hierarchical grid system, that is, a coarse grid and a fine grid. Deep learning network is used to train data on the coarse gird. Source term and permeability field is treated as input parameter and coarse-scale solution is treated as output parameter. We construct multiscale basis functions by solving local flow problems within coarse gridcells. Heterogeneity and interactions between matrix and fracture are captured by basis functions. Oversampling technique is applied to get more accurate small-scale details. Numerical experiments show that the multiscale deep learning model is promising for flow simulation in heterogeneous and fractured porous media.

Conduction band structure of silicon under uniaxial[110] stress is studied with two band k·p perturbation theory. Splitting energy of conduction band minima and electron effective mass as a function of stress and direction electron mobility in uniaxial stressed silicon are obtained with relax time approximate theory. Intervalley scattering, intravalley scattering, and ionized impurity scattering are considered in calculation. It is demonstrated that as uniaxial[110] stress is applied on silicon crystal, a significant anisotropy in electron mobility can be observed. Among crystal directions[001],[110], and[110], electron mobility along[110] direction under uniaxial[110] tensile stress has a profound enhancement, which increase from 1 450 cm²·Vs^-1 to 2 500 cm²·Vs^-1 as stress change from 0 to 2 GPa. Electron mobility enhancement is mainly due to uniaxial stress induced conduction effective mass reduction, while suppression of intervalley scattering plays a minor role.

Considering complexity of coupled fluid-solid calculation of low supersonic aircraft under unsteady flight condition, real time aerodynamic heat by outer fluid field is converted into floated third type boundary conditions for decoupled calculation. Take three-dimensional head cone experiencing accelerating diving as example, floated temperature and radiation equilibrium methods are adopted to extract surface heat transfer coefficients for decoupling calculation. Results are compared with coupled results to validate reliability of decoupling methods. It shows that with dispersed state points representing unsteady process, temperature distributions of cone surface at different state points by decoupling methods agree well with coupling ones. Calculating efficiency of decoupling methods is superior to coupling ones. Maximum relative errors of decoupling methods are both within 2%, even as external aerodynamic environment changes dramatically.

Disturbance equations of velocity and temperature fields are built based on Boussinesq approximation,in which second order and higher-order disturbance of Bénard convection are ignored. Transcendental equations controlling Bénard convection are obtained with non-dimensional treatment of variables. Specific parameters of Bénard convection are obtained,which provide reference and guidance for Bénard convection experiments.

Possible equilibrium geometries of W_nNi_m(n+m=8) clusters are optimized with density functional theory at B3LYP/LANL2DZ level.For ground state structures,molecular orbital energy level,HOMO level,LUMO level,aromaticity and thermodynamics are analyzed.It shows that molecular orbits between Alpha and Beta in clusters W₅Ni₃ and W₆Ni₂ are completely degenerate,and all electrons are strictly paired.W₂Ni₆ cluster,with low energy gap,has strong chemical activity.Its component for frontier orbital are basically the same.W₆Ni₂cluster,with high energy gap,has weak chemical activity;Clusters W₁Ni₇,W₅Ni₃,W₆Ni₂,W₇Ni₁ have aromaticity,while W₂Ni₆ cluster has antiaromaticity.Formation enthalpy of all clusters W_nNi_m(n+m=8) are negative.They are stable Oil thermodynamics.

Possible equilibrium geometries of W_nC^0,±(n=1,…,6) clusters are optimized with density functional theory at B3LYP/LANL2DZ level.Stability and electronic properties of ground state structures are analyzed.It shows that as n > 3 cluster undergoes a transition from a two-dimensional structure to a three-dimensional structure,and carbon atom remains on the surface of cluster.Stability of W_n cluster is increased with doped carbon atom.Moreover W₃C is the most stable one among W_nC^0,±(n=1,…,6) clusters,and it is taken as a basic structure.Electronic properties of W_nC^0,±(n=1,…,6) clusters are characterized by analyzing energy gaps,vertical electron affinities(VEA) and vertical ionization potentials(VIP) of W_nC with pure Wnclusters.W_nC clusters get electron easily and shows higher nonmetallicity than W_n clusters.