1. VASP优化结构的内部过程怎样？哪些参数会影响结构优化的结果？学习从摘录开始

（1）LiH6

ThePerdew-Burke-Enzerhof (PBE) exchange–correlation functional (26), an energycutoff of 500 eV, and all-electronplane-wave basis sets within the projector augmented wave (PAW) method (27)have been used.

The k-pointgrids were generated using the Monkhorst–Pack scheme, and the number ofdivisions along each reciprocal lattice vector was chosen so that the product of thisnumber with the corresponding real lattice constant was ~50 Å. 【怎么算的】

(2) WH6

  For the plane-wave basis-set expansion, an energy cutoff of 750 eV was used.Theaccuracy of sampling in the first Brillouin zone (FBZ) was ensured by using dense k-point grids resulting from a series ofconvergence tests. In all cases, the structures were fullyoptimized with respect to lattice parameters and atomic positions until the forces acting on atoms converged to0.001 eV Å^-1 or less.

A prerequisite【先决条件】 for thetheoretical study of tungsten hydrides under pressure is that the method ofcalculating electronic structure used should be able to reproduce correctly the electronic structure of metallictungsten itself. Thus we confirmed that the computationalmethodology described above accounts for the measured Fermi surface andassociated bands in pure tungsten at one atmosphere【我已做过类似测试】. The agreement is very good.

（3） CaH6

To analyze the results of structuresimulations, we selected a number of distinct lowest-enthalpy structures andoptimized their structures as a function of pressures using VASP code. To reduce the errors, the cut-off energyfor the expansion of wavefunctions into plane waves was set to 400 eV in allcalculations,and the Monkhorst–Pack1 grid with amaximum spacing of 0.03 Å−1 was individually adjusted in reciprocal space tothe size of each computational cell. This usually gave well-convergedtotal energies (within ~ 1 meV per atom).

（4）High pressure ices

For a proposed structure the enthalpies are computed bydensity functional theory (DFT) with the VASP software package (21), making use of the Perdew-Burke-Ernzerhofparametrization of the exchangecorrelation energy density (22), andthe ProjectorAugmented Wave (PAW) method to describe the electron-ion interaction  (23, 24).

We use “hard” PAW datasets, with outermost cutoff radii for hydrogen and oxygen of 0.8 and1.1 Bohr radii, respectively. A plane wave energycutoff of 800 eV was employed, and Brillouin zones were sampled with a linear k-point densityof 20 Å.  (Auto 20 ---KPOINTS方式)

(5) High pressurepartially ionic phase of water ice

Total energy calculations were performed inthe framework of density functional theory within the Perdew–Burke–Ernzerhof29 parameterization ofgeneralized gradient approximation30 as implemented in the VASP (Vienna Ab Initio simulationpackage) code31.

The projector-augmented wave (PAW) method32,33was adopted with the PAW potentials taken from the VASP 【cut off 等于多少？！】library where 1s1 and 2s22p4are treated as valence electrons for H and O atoms, respectively.

The validity of the used PAW potential atcurrently studied pressures ( < 20 Mbar) has earlier been carefully checkedby comparing with the full-potential all-electron calculations through Excitingcode9.

The use of the plane-wave kinetic energy cutoff of 1,200 eV and dense k-pointsampling34, adopted here, were shown to give excellent convergence of totalenergies.

（6）  LiNH₂

the electron−ion interaction was treated with the Projector AugmentedWave (PAW) method.34,35 The PAW potentials represent the valence electrons ofLi (1s2, 2s1), N (2s2, 2p3), and H (1s1). Aplane-wave basis set cutoff of 650 eV and selfconsistent field toleranceof 0.1 × 10⁻⁶ eV/atomwas set. The Brillouin zone was sampled on a grid of spacing 2π × 0.04 Å⁻¹. [2π/(Na), N为分割数]

With these settings, thestructures were fully relaxed until the forces on atomswere less than 10^-3 eV/Å.

该部分参数解读： EDIFF=1.0xe-7

  The three qualities of k-point separation for CASTEP (1/Angstrom): coarse.le..08 & medium.le..05 & fine.le..04

KPOINTS设置解读：

 Automatic mesh

  0

  Auto

 40

上面的40 表示倒空间的k-point density （1/40= 0.025 Å^-1），即每Å^-1上分布的k点数目，40 /Å^-1,表示1 Å^-1长度上分布有40个k点（40 per Å^-1），假设倒空间中晶胞大小为1x1x1 Å^-3，则在此设置下不考虑对称性时，产生的k点数将达到40x40x40= 64000个！所以40 已经是很好的设置了。

VASP手册： Useful values of the length vary between 10 (large gap insulators) and 100 (d-metals).

 →   2π*0.025 Å^-1中2π怎么来的？

可见，当晶格常数如a₁在压力作用下减小时，为保持k-pointdensity不变，分割点数N₁将增加。

只有k-point density选择的fine，则Enthalpy-pressure曲线上将不会出现跳跃点；改变截断能，只要保证在相同截断能下计算，则不同结构之间的焓差大小不会随截断能的改变而显著改变(Prasad语)。

同一体系，不同结构之间比较，用不同的k点方式，能量差之间差别大吗？如果k-points density相近，则能量相差不大，最多几个meV.【Prasad语】

网上解读：

EDIFF=1E-5是电子自洽步（SCF）的收敛标准，表示0.01meV/cell；
EDIFFG表示离子弛豫的收敛标准，若为正值，表示以能量为收敛标准，意义同上，默认值为EDIFF的10倍；若为负值，表示以离子受力作为收敛标准，单位eV/A，这个标准不能太高，否则很难收敛。
一般情况下，你可以先设置EDIFFG为正值，跑完后用p4vasp查看一下，能看到能量和力随弛豫步的变化（跟MS类似），能量最低的时候力也基本上就是最小的了，而且对稍微复杂一点的体系，力很难收敛到0.001eV/A。

   （7）Noble hydirdes   JPC2012

【通过晶格参数比较证实参数选择合理？】

The lattice parametersof the predicted various structures for PtH are listed in Table 1. It is found that thelattice data for hcp-PtH at 42.9 GPa are in excellent agreement withexperiments (e.g., our theoretical a = 2.795 Å and c = 4.769 Å versusexperimental data of a = 2.773 Å and c = 4.713 Å with a maximum deviation of1.1%).15

This fact strongly supported the choice of PAW potentials and GGA for thecurrent study. Our current results also confirmed previous experiences that GGA provides better cohesiveenergies than the local density approximation in the cases of platinummetal(Pt, Ir,Os, Ru, Rh, and Pd) nitrides.33

(8) NaHn  PRL2011

Calculations wereperformed with the VASP 4.6 and 5.2 codes, [11], along with the PBE functional [12], an energy cutoffof 1000 eV, and plane-wave basis sets within the projector augmentedwave (PAW) method [13].

For Na the 2s, 2p, 3selectrons were treated explicitly, and for H a hard all-electron PAW was used. The k-pointgrids were generated using the _-centeredMonkhorst-Pack scheme,

and thenumber of divisions along each reciprocal lattice vector was chosen so that theproduct of this numberwith the real lattice constant was 50A° .