Interpolation of BSE Hamiltonian
Posted: Thu Jun 13, 2019 2:52 pm
Dear developers:
I'm learning to use the interpolation technique of BSE Hamiltonian within ABINIT.
Firstly, I have tried the tutorials t31.in~35.in attached below the description of related variables (bs_interp_mode, etc.). All these tests can be completed properly only when submitted to a serial job, any parallelization will give error like this:
--- !ERROR
src_file: m_hexc.F90
src_line: 507
mpi_rank: 0
message: |
Parallelization not available in interpolation
...
But a realistic calculation with fully converged parameters can hardly be performed by a serial job, isn't it? So can we actually perform a parallel BSE calculation with interpolation technique?
After that, I try to use the BSE interpolation on other materials with more realistic calculation parameters. This time, the self-consistent and non-self-consistent ground state calculation as well as the "preparation" BSE on the coarse k gird are all successfully completed. However, another issue occurs as:
--- !ERROR
src_file: m_double_grid.F90
src_line: 205
mpi_rank: 43
message: |
Coarse mesh works only with nshiftk=1
...
But I'm sure that I do use nshiftk=1 in all the calculations, hence I'm rather confused about this comment.
I'm very appreciative if anyone could tell me how to solve these two problems, thank you very much~!
I'm learning to use the interpolation technique of BSE Hamiltonian within ABINIT.
Firstly, I have tried the tutorials t31.in~35.in attached below the description of related variables (bs_interp_mode, etc.). All these tests can be completed properly only when submitted to a serial job, any parallelization will give error like this:
--- !ERROR
src_file: m_hexc.F90
src_line: 507
mpi_rank: 0
message: |
Parallelization not available in interpolation
...
But a realistic calculation with fully converged parameters can hardly be performed by a serial job, isn't it? So can we actually perform a parallel BSE calculation with interpolation technique?
After that, I try to use the BSE interpolation on other materials with more realistic calculation parameters. This time, the self-consistent and non-self-consistent ground state calculation as well as the "preparation" BSE on the coarse k gird are all successfully completed. However, another issue occurs as:
--- !ERROR
src_file: m_double_grid.F90
src_line: 205
mpi_rank: 43
message: |
Coarse mesh works only with nshiftk=1
...
But I'm sure that I do use nshiftk=1 in all the calculations, hence I'm rather confused about this comment.
I'm very appreciative if anyone could tell me how to solve these two problems, thank you very much~!