[Re:01] Nd-Fe-B Convergence

Posted on : March 25, 2015 (Wed) 05:39:17

by Hitoshi GOMI

> So my first question is it really ok to use that penmp ?

Please edit the makefile, in order not to use the openMP.

# If openMP is not installed the following two lines should be
# replaced by
omp =
nomp =
#omp = -openmp
#nomp = -openmp-stubs

> Could anyone of you please give me some feedbacks regarding the convergence problem.

I have little confidence in my answer, because I am not a developer, and I have never tried such a complicated system...

Basically, more than 1 is required for bzqlty to converge. bzqlty=0 is used only for test calculation. And, mxl=3 is required for Nd. Spin-orbit coupling is possibly important.

How about trying this?

go data/nd2fe14b_srals
st 16.63, 1.3852, , , , ,
0.001 1.4 srals mjw mag 2nd
update 1 200 0.02

9
B_g 1 0 0 2 5 100
Fe_c 1 0 0 2 26 100
Fe_e 1 0 0 2 26 100
Fe_j1 1 0 0 2 26 100
Fe_j2 1 0 0 2 26 100
Fe_k1 1 0 0 2 26 100
Fe_k2 1 0 0 2 26 100
Nd_f 1 0 0 3 60 100
Nd_g 1 0 0 3 60 100
...

Hitoshi