Hi,
I think this difference is numerics, probably from the mkl.
Do you use ifort + mkl (which version) ??
It would be interesting to
i) use gfortran + openblas
ii) gfortran + self-compiled blas/lapack (from fortran sources) - will
be very slow
iii) use compiler options with increased precision or no optimization
(-O0). I think our default ifort options do not fully explore machine
precision but speed. Laurence Marks may know better than me how to set
compilation parameters with increased precision.
Are the momentum matrix elements very different or also only with
numerical noise ?
Actually I think that they can be different, only the squared mommat
elements should be fine (not sure about this statement).
Best regards
Peter
Am 26.04.2023 um 05:25 schrieb 曹迎迎:
Dear Prof. Blaha,
Thank you for your reply.
I have checked the eigenvalue and made sure that it is not
degenerate. Actually, the second eigenvalue is -5.633 and differs from
the first value -5.833 by about 0.2 Ry.
I undersatand the phase freedom of eigenvactor, but it seems like
that the difference not only came from diagonalization. After trace back
the output of the program, i find that the Hamiltonian matrix elements
are not the same. In the 2-kpt case, the Hamiltonian(HS) given by hamilt
in lapw1 is
...
-0.296635514229139 -0.296635514229139 -7.955504967907867E-003
...,
while in the 101-kpt case, it is
...
-0.296635514229139 -0.296635514229139 -7.955504967907853E-003
....
Is this a numerical error? Or did i do something wrong?
In fact , I want to get finally the momment matrix case.mommat by the
OPTIC module. And i find that the two mommat files are different to each
other at the same k vector from two klists. That is why i check the
eigenvectors. If there is only the phase freedom in the eigenvector,
does it affect the momment matrix?
Once again, thank you for your response and support. I look forward
to hearing back from you soon.
Best regards, Yingying Cao
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
--
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300
Email: peter.bl...@tuwien.ac.at WIEN2k: http://www.wien2k.at
WWW: http://www.imc.tuwien.ac.at
-------------------------------------------------------------------------
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html