题记： I am lucky to stay with Andreas Hermann for nealy five months.  Today maybe the last day to stay with him. Thanks for his help, for him to share his luck to me. Light at the end of the tunnel.

1. Symmetry

……………………

Analysis of symmetry for initialpositions (statically):

RoutineSETGRP: Setting up the symmetry group for a simple orthorhombic supercell.

Subroutine GETGRP returns: Found 8 space group operations (whereof  4 operations were pure pointgroup operations) out of a pool of  8 trial point group operations.

The static configuration has thepoint symmetry C_2h.

The point group associated withits full space group is D_2h.

Analysis of symmetry fordynamics (positionsand initial velocities):

Subroutine DYNSYM returns: Found 8 space group operations (whereof  4 operations were pure pointgroup operations) out of a pool of  8 trial space group operations(whereof  4 operations were pure point group operations) and found also    2 'primitive' translations

The dynamic configuration has thepoint symmetry C_2h.

The point group associated withits full space group is D_2h

Address: /home/xy83/work/strucSearch/optimization/done-opt/H2-contcar-opt/Cmca-4-contcar-afteropt2/pra-cmca-4/pra-cmca-4-opt2/4000-1

INCAR

SYSTEM = Various- local optimisation

PREC = Accurate

ENCUT=800.0

EDIFF= 1E-7

EDIFFG = -1E-3    →   EDIFFG = -2E-5   （time expensive? Not reasonalbe?）

#SYMPREC=1e-3   →   SYMPREC=1e-5 （Default value ）

IBRION= 2

POTIM= 0.1

ISIF= 3

NSW= 100

PSTRESS = 0

#for insulator

#ISMEAR=0

#SIGMA= 0.05

#for metal

ISMEAR= 1

SIGMA= 0.2

ISYM=0      （time expensive? Not reasonalbe?）

#LREAL= .FALSE.

LCHARG = FALSE

LWAVE = FALSE

~

Andreas回答：

 Twoways:

1) Decrease the SYMPREC，usually, you will find a symmetryhigher than P1.   Be careful to increasethis parameter, otherwise you will get a wrong higher symmetry that will changethe POSCAR to one that totally different from the initial one.  The SYMPEC are illustrated as in thefollowing figure.  It is similar to the toleranceof symmetry we use in CASTEP.

【SYMPREC参数表征了离子偏离格点的程度，如下图所示】

2) Use ISYM=0, often can deal with thisproblem.  The symmetry will be kept atP1. However, it  may be time expensive.

3) Copy CONTCAR to POSCAR, and continue tooptimize the structure.

Prasad回答：

1) We can use ISYM=1； ( ISYM=0~3.)

2) We can use VESTA to change the symmetryto P1.

3) Use gamma KPOINTS generated methods.

Duck Young Kim回答：

(1) 调节Tolerance  (SYMPREC=1e-3)，使变小，这样VASP找的对称性更精确，不至于出现对称性不匹配问题；

(2) 使用ISYM=0，完全不考虑对称性，这样就不会出现对称性不匹配问题；但会更耗时。

2. Different KPOINTS generatedmethods will give different enthalpies? (Monkhorst-Pack, Auto, Gamma)

The cut-off energy for the expansion of thewave function into plane waves and Monkhorst-Pack k meshes were chosen toensure that all enthalpy calculations converged to better than 1 meV/atom.

/c2m-sch6-300G-21-12/first_opt> grep NKPT OUTCAR_*

OUTCAR_100:  k-points         NKPTS =   1625  

OUTCAR_300[kbar]:   k-points          NKPTS =   1625

OUTCAR_800:   k-points          NKPTS =    860     【在此压力点，对称性可能改变了】

OUTCAR_1500:  k-points         NKPTS =   1625

OUTCAR_2000:  k-points           NKPTS =    860

OUTCAR_2500:  k-points           NKPTS =    860

Andreas回答：

1) Different KPOINTS generated methods will give different enthalpies?  (Monkhorst-Pack, Auto, Gamma)

Usually, it will give several meV difference.  Therefore, please choose the same KPOINTS generated methods for the same structure under different pressure.This will make the Enthalpy-pressure curve look not smoothly.

2) The changes of NKPTS are unreasonable. It may be due to the changes of thesymmetries of the structures. Usually, we can use (N1,N2,N3) at  the highest pressure for all pressurestructures.

3. ZBRENT: fatalerror in bracketing

 N      E                    dE             deps       ncg     rms         rms(c)

DAV:  1    -0.248484889542E+02   -0.59182E-03  -0.76643E-05 48336   0.106E-01    0.715E-03

DAV:  2    -0.248484881092E+02    0.84499E-06  -0.21719E-06 39664   0.134E-02

71 F= -.43273373E+01 E0= -.43270216E+01 d E =-.422912E-06

curvature:  0.00 expect dE= 0.667-308 dE for cont linesearch  0.112-308

ZBRENT:fatal error in bracketing

   pleasererun with smaller EDIFF, or copy CONTCAR

   to POSCARand continue

Head-admin:

   pleasehave a look at the OUTCAR file, at the forces of the last ionic step beforeBrent's algorithm fails 
-- if the calculation is very well-converged already, please proceed using IBRION=1;ADDGRID=.TRUE. and a larger ENCUT

-- if this error appears at an early stage of the relaxation, pelase check if your input geometry was reasonable, eg by having a look at the interatomic distances (writtenin OUTCAR) or the forces of the first geometries.

  It may thenhelp to increase the accuracy (PREC= Accurate and ahigher ENCUT) and to decrease POTIM to avoid too large stepwidhts in the firstrelaxation step. 

Andreas回答：

It means the symmetry failed during theoptimization process. COPY CONTCAR POSCAR will be helpful due to the historicaleffect of VASP.  

The method of “ increase theaccuracy (PREC= Accurate and a higher ENCUT) and to decrease POTIM” will make the enthalpy change better to get/reach the optimumstructure as show in the following figure.

4.Convex hull

Andreas回答：

Maybe convex hull is nonsense.

Different compounds can be stable under thesame pressure range as shown in the above figure .  The relative stability depends on the initialmixture ratio. For example:

For example, a tie line between LiH₆and LiH would show that a mixture ofthese could yield LiH₂ (but not LiH₃) at P =150GPa, providedthat the kinetic barriers are not too high.

  LiH₆ + LiH→LiH₂

  LiH₂+2 H₂→ LiH₆

【类似与相图中的杠杆定则，稳定性与分解产物有关，convex hull phasediagram】

From absolute formationenthalpy  to relative formation enthalpy,we often lift up the lower enthalpy compound (such as B) to the ZERO place.

【相比常规的relative convex hull phase diagram,可以认为上图是absolute convex hull phase diagram】

Binary Compositional Phase Diagrams

Figure 2: Calculated Fe-PPhase Diagram

Figure2 is an example of a calculated binary Fe-P phase diagram at 0K and 0 atm usingthe PDApp.

Binaryphase diagrams show the complete convex hull for the system, where the y-axis is theformation energy per atom and the x-axis is the composition (e.g., in the Fe-Pphase diagram, the x-axis is the fraction of P).

The black lines show the convex hullconstruction, which connects stable phases.Unstable phases will always appear above theconvex hull line;one measure of the thermodynamic stability of an arbitrary compound is its distance from the convex hullline, which predicts the decomposition energy of that phase into the most stable phases.

 解读：

（1）在tie-line上的都是稳定的物相，画tie-line时，下一点的选择应保证连接线是所有最近邻点连线中位置最低的(hf最低)。

（2） 在tie-line上的都是稳定的物相，在其上方的相是不稳定的，在其下方的相？？？？

【选作基准相的物质必须稳定，不会转变为相图中的其他相 ？ 但自身可发生同质异构转变！】

From Prof . Miaomao Shen

For the most stable structures ateach pressure, the enthalpy of formation per atom is calculated using the followingformula:

hf(CsF_n) = (H(CsF_n) - H(CsF)- (n -1)H(F₂)/2)/(n + 1).[28, 29]

We choose to use the enthalpy of CsF instead of Cs (see the above formula) because ofthe substantial stability of CsF throughout the studied pressure range.

Thus, changing Cs to CsF does not change theconvexity of the compounds with higher F compositions, although the absolutevalue of hf does change.[29]