dmftwdft-project / dmftwdft Goto Github PK

Dear developers,
I'm trying to repeat the example CSC calculation of LaNiO3. I find that though the dmft.x part works as well as in NCSC calculation, running vaspDMFT results in the follow errors:

...
LDA part: xc-table for Pade appr. of Perdew
found WAVECAR, reading the header
POSCAR, INCAR and KPOINTS ok, starting setup
FFT: planning ...
reading WAVECAR
the WAVECAR file was read successfully
initial charge from wavefunction
DFT+DMFT run with DMFTwDFT
entering main loop
N E dE d eps ncg rms rms(c)
DAV: 1 -0.325174780448E+04 -0.32517E+04 -0.15030E+04 2920 0.177E+03BRMIX: very serious problems
the old and the new charge density differ
old charge density: 29.94236 new 31.09228
0.108E+03
DAV: 2 -0.117777953625E+04 0.20740E+04 -0.20858E+03 2340 0.404E+02 Fail to find mu, increase mu_iter
Fail to find mu, increase mu_iter
Fail to find mu, increase mu_iter
Fail to find mu, increase mu_iter
...
Projection 26 done.
Projection 27 done.
Projection 28 done.
MLWF_WANNIER90: WARNING: wannier90.win seems to contains
a UNIT_CELL_CART block already, I hope it applies to the current setup ...

MLWF_WANNIER90: WARNING: wannier90.win seems to contains
a ATOMS_CART block already, I hope it applies to the current setup ...

MLWF_WANNIER90: WARNING: wannier90.win seems to contains
the MP_GRID tag already, I hope it applies to the current setup ...

MLWF_WANNIER90: WARNING: wannier90.win seems to contains
a KPOINTS block already, I hope it applies to the current setup ...

BRMIX: very serious problems
the old and the new charge density differ
old charge density: 31.54367 new 78.00000
1 F= -.44645718E+03 E0= -.44642453E+03 d E =-.652956E-01
writing wavefunctions

I've tried to increase mu_iter in the INPUT.py from 100 to 1000, but the charge number is still not correct (there are 78 valence electrons.
Thanks for any suggestion!

Dear developers:
According to the tutorial of https://dmftwdft-project.github.io/DMFTwDFT, I compiled DMFTwDFT based on Python/2.7.9 and Wannier90/2.0.0. No error was reported during compilation.
However, when I tried to use the command "DMFT.py - h", the following error occurred:
Traceback (most recent call last):
File "/WORK/bin/g-vasp/DMFTwDFT-master/bin/DMFT.py", line 18, in
from INPUT import *
ImportError: No module named INPUT
How can I solve this problem?
Many thanks!

The VASP and wannier90 version I used are 5.4.4 and 1.2. After DFMTwDFT and wannier90 compiled successfully, the files under DMFTwDFT-master/sources/CSC-mods/vasp are copied to vasp source folder, changes are mde in vasp makefile.include according to DMFTwDFT manual(https://dmftwdft-project.github.io/DMFTwDFT/installation.html). However, the vasp compilation failed with following erro message:

make[2]: Entering directory /data/home/xinyuanwei/software/DMFTwDFTsource/vasp.5.4.4/build/std'
mpiifort -free -names lowercase -assume byterecl -w -O2 -xCORE-AVX512 -I/opt/compiler/intel2018u4/compilers_and_libraries_2018.5.274/linux/mkl/include/fftw -c charge.f90
charge.F(48): error #7002: Error in opening the compiled module file. Check INCLUDE paths. [MLWF]
USE mlwf
--------^
charge.F(328): error #7002: Error in opening the compiled module file. Check INCLUDE paths. [MLWF]
USE mlwf
--------^
charge.F(329): error #7002: Error in opening the compiled module file. Check INCLUDE paths. [PEAD]
USE pead
--------^
charge.F(364): error #6404: This name does not have a type, and must have an explicit type. [WDES_FULL_PEAD]
WORIG%WDES=>WDES_FULL_PEAD
----------------^
charge.F(364): error #6796: The variable must have the TARGET attribute or be a subobject of an object with the TARGET attribute, or it must have the POINTER attribute. [WDES_FULL_PEAD]
WORIG%WDES=>WDES_FULL_PEAD
----------------^
charge.F(406): error #6404: This name does not have a type, and must have an explicit type. [NUM_BAND_MAX]
num_band_max=band_win_(2,NK)-band_win_(1,NK)+1
----------^
charge.F(406): error #6404: This name does not have a type, and must have an explicit type. [BAND_WIN_]
num_band_max=band_win_(2,NK)-band_win_(1,NK)+1
-----------------------^
charge.F(407): error #6385: The highest data type rank permitted is INTEGER(KIND=8). [NUM_BAND_MAX]
ALLOCATE(CR_TOT(WDES1%GRID%MPLWV,num_band_max))
-------------------------------------------^
charge.F(427): error #6404: This name does not have a type, and must have an explicit type. [DMFT_EVEC_]
W1%CR(:)=W1%CR(:)+CR_TOT(:,N2)DMFT_EVEC_(N2,NBG-band_win_(1,NK)+1,NK)
------------------------------------------------^
charge.F(429): error #6404: This name does not have a type, and must have an explicit type. [DMFT_EVAL_]
WEIGHT=WORIG%WDES%RSPINKPOINTS_FULL%WTKPT(NK)*DMFT_EVAL_(NBG-band_win_(1,NK)+1,NK)
--------------------------------------------------------------^
compilation aborted for charge.f90 (code 1)
make[2]: *** [charge.o] Error 1
make[2]: Leaving directory /data/home/xinyuanwei/software/DMFTwDFTsource/vasp.5.4.4/build/std' cp: cannot stat ‘vasp’: No such file or directory make[1]: *** [all] Error 1 make[1]: Leaving directory /data/home/xinyuanwei/software/DMFTwDFTsource/vasp.5.4.4/build/std'
make: *** [std] Error 2`

I guess this error might related with the modifed vasp files (charge.F et al.) provided by DMFTwDFT. Does anyone compile the VASP with DMFTwDFT successfully? Any suggestions are appreciated. Many Thanks.

Xinyuan Wei
Jul. 18 2020

Hi, Thank you for the installation tutorial(https://dmftwdft-project.github.io/DMFTwDFT/installation.html). I want to know which versions of these dependencies(matplotlib numpy scipy weave mpi4py requests) are suitable.

I tried the command (pip install matplotlib numpy scipy weave mpi4py requests) but get errors.

And I want know that whether python3 is ok for the installation.

I'm looking forward to your reply.

Dear developers,
Here are two problems when I repeat the example calculations of LaNiO3_vasp.
I follow the steps in the appendix of your published paper (https://doi.org/10.1016/j.cpc.2020.107778) for dos and band calculation. The dos is ok, but I fail the band calculations.
So far, two problems I've met.

1. In the paper, Interp_Sig.py is used but I don't find it in the bin/.
2. when I ran dmft_ksum_band it told me two files including ksum,.input and SigMoo_real.out are required. So how to generate these files?
   Thanks a lot for any suggestion

Dear developer,

I find there is a bug in the subroutine MLWF_WANNIER90_AMN in sources/CSC-mods/vasp/mlwf.F .
You forgot to allocate M_matrix which will cause a segment fault error.

You can just fix the bug by adding after the line #541 in sources/CSC-mods/vasp/mlwf.F
ALLOCATE(M_matrix(num_bands,num_bands,nntot,num_kpts,WDES%ISPIN))
and by adding after the line #918.
DEALLOCATE(M_matrix)

Best,

Changming Yue

In the line #333 of DMFTwDFT/sources/src/read_inputs.F90
You should use
Overlap=MATMUL(TRANSPOSE(CONJG(UNI_loc)),UNI_loc)
instead of
Overlap=MATMUL(UNI_loc,TRANSPOSE(CONJG(UNI_loc)))
to check the unitarity of matrix of a_matrix, Amn(k).

The bug is due to dimension mismatch between Overlap (dimension nwannnwann) and MATMUL(UNI_loc,TRANSPOSE(CONJG(UNI_loc))) (nband_maxnband_max).