[QE-users] eigenvectors failed to converge

[┰☉ ♭ə: №▌]

Dear pwscf users
Recently I am calculating a system that needs to consider SOC, I use the Norm conservation pseudopotential with full relativistic, when I run vc-relax, I have such kind error:


Initial potential from superposition of free atoms
starting charge  103.98524, renormalised to  104.00000
Starting wfcs are   66 randomized atomic wfcs +   94 random wfcs
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Error in routine cdiaghg (159):
eigenvectors failed to converge
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
stopping ...


But when I add:
noncolin = .true.
lspinorb = .true.
in the input file, the error will disappear and the program can run normally, however, SOC will be considered in this way in vc-relax. I notice that SOC is not supported in structural optimization. I want to know how to use the Norm conservation pseudopotential with full relativistic for structural optimization?


Thanks in advance,
Dongze fan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20220912/0b633da4/attachment.html>