k-point grid
Posted: Fri Jul 04, 2014 5:04 pm
Hello,
I am trying model an alumina surface. Its space group (R-3c) has two representations: the rhombohedrical (primitive with 10 atoms) and the hexagonal (non-primitive with 30 atoms). Because I want the surface (001) I need the hexagonal representacion. First, I tried to study the convergence for increasing ecut with the bulk structure and I got several times the following
--- !ERROR
message: |
The eigenvector with band1 has zero norm.
This usually happens when the number of bands (nband) is comparable to the number of planewaves (mpw)
Action: Check the parameters of the calculation. If nband ~ mpw, then decrease nband or, alternatively, increase ecut
src_file: fxphas.F90
src_line: 164
checking out that information I realized that nband = 76 and npw > 2200, but anyway I increased the ecut and sometimes it works but soon or later (varying the k-point grid or the shifts, doing optimizations or introducing vacuum to model the surface) the same error message returns and I do not understand why.
Simplifying the question: How can change the k-point grid to avoid this problem? or What are the optimal shifts for the (001) surface of an hexagonal structure?
Attached is one of the simpliest input file which gives such an error.
Thank you in advance,
Jorge Vargas
I am trying model an alumina surface. Its space group (R-3c) has two representations: the rhombohedrical (primitive with 10 atoms) and the hexagonal (non-primitive with 30 atoms). Because I want the surface (001) I need the hexagonal representacion. First, I tried to study the convergence for increasing ecut with the bulk structure and I got several times the following
--- !ERROR
message: |
The eigenvector with band1 has zero norm.
This usually happens when the number of bands (nband) is comparable to the number of planewaves (mpw)
Action: Check the parameters of the calculation. If nband ~ mpw, then decrease nband or, alternatively, increase ecut
src_file: fxphas.F90
src_line: 164
checking out that information I realized that nband = 76 and npw > 2200, but anyway I increased the ecut and sometimes it works but soon or later (varying the k-point grid or the shifts, doing optimizations or introducing vacuum to model the surface) the same error message returns and I do not understand why.
Simplifying the question: How can change the k-point grid to avoid this problem? or What are the optimal shifts for the (001) surface of an hexagonal structure?
Attached is one of the simpliest input file which gives such an error.
Thank you in advance,
Jorge Vargas