Hello,
I am trying to use icutcoul to get rid of the interaction along the z direction for a 2D bilayer MoS2. The calculation runs perfectly without icutcoul. Nevertheless, the calculation cannot converge once I turned on icutcol.
My input is as below:
2 acell 3.1845414639 3.1845414639 51.6705341339 Angstrom
3 rprim 1.0 0.0 0.0
4 -0.500000 0.86602540378 0.0
5 0.0 0.0 1.0
6 chkprim 0
7 ecut 21.0
8 pawecutdg 42.0
9 #Definition of the k-point grid
10 nshiftk 1
11 ngkpt 16 16 1
12 npkpt 50
13 shiftk 0 0 0
14 kptopt 3
22 nstep 50 # Maximal number of SCF cycles
23 toldfe 1.0d-10 # Will stop when, twice in a row, the difference
27 vdw_xc 5
28 vdw_tol 1d-10
29 diemac 12.0 # Although this is not mandatory, it is worth to
30 # precondition the SCF cycle. The model dielectric
31 # function used as the standard preconditioner
32 pp_dirpath “./”
33 pseudos “Mo.GGA_PBE-JTH.xml, S.GGA_PBE-JTH.xml”
34 prtvha 1
35 prtvpsp 1
36 ndtset 2
37 icutcoul1 3
38 icutcoul2 2
39 vcutgeo2 1 1 0
40 #Definition of the atom types
41 natom 6 # There are five atoms
42 ntypat 2
43 typat 21 42
44 znucl 42 16 # The keyword “znucl” refers to the atomic number of the
45 xred
46 0.333333333 0.666666667 0.56242311
47 0.666666667 0.333333333 0.43757688
48 0.666666667 0.333333333 0.59242476
49 0.000000000 0.000000000 0.46760109
50 0.666666667 0.333333333 0.53239891
51 0.000000000 0.000000000 0.40757524
I doubt this might be related to pseudopotential, so I also tried HGH normconserving pseudopotential(cp2k / Code / [r18621] /trunk/potentials/Goedecker/abinit/pbe). However, the problem there is: the convergence speed with and without icutcoul is similar, and it just cannot converge. Does anybody have good suggestion? Thanks for your help!