convergence problem with 3d 4d transition metal

[ljzhang123]

Dear colleague:

when i do structure optimization for the 3d and 4d pure transition metal such as Zn with pv pesudopotential ,the electron loop couldn't convergence, can anyone help me with this problem? Thank you for your attention.

[admin]

Can you display the problematic POSCAR?

[ljzhang123]

CIF file
1.0
7.9946999550 0.0000000000 0.0000000000
-3.9973499775 6.9236132567 0.0000000000
0.0000000000 0.0000000000 9.8936004639
Zn
36
Direct
0.111110002 0.222220004 0.125000000
0.222220004 0.111110002 0.375000000
0.444440007 0.222220004 0.125000000
0.555559993 0.111110002 0.375000000
0.777779996 0.222220004 0.125000000
0.888890028 0.111110002 0.375000000
0.111110002 0.555559993 0.125000000
0.222220004 0.444440007 0.375000000
0.444440007 0.555559993 0.125000000
0.555559993 0.444440007 0.375000000
0.777779996 0.555559993 0.125000000
0.888890028 0.444440007 0.375000000
0.111110002 0.888890028 0.125000000
0.222220004 0.777779996 0.375000000
0.444440007 0.888890028 0.125000000
0.555559993 0.777779996 0.375000000
0.777779996 0.888890028 0.125000000
0.888890028 0.777779996 0.375000000
0.111110002 0.222220004 0.625000000
0.222220004 0.111110002 0.875000000
0.444440007 0.222220004 0.625000000
0.555559993 0.111110002 0.875000000
0.777779996 0.222220004 0.625000000
0.888890028 0.111110002 0.875000000
0.111110002 0.555559993 0.625000000
0.222220004 0.444440007 0.875000000
0.444440007 0.555559993 0.625000000
0.555559993 0.444440007 0.875000000
0.777779996 0.555559993 0.625000000
0.888890028 0.444440007 0.875000000
0.111110002 0.888890028 0.625000000
0.222220004 0.777779996 0.875000000
0.444440007 0.888890028 0.625000000
0.555559993 0.777779996 0.875000000
0.777779996 0.888890028 0.625000000
0.888890028 0.777779996 0.875000000

This is a 3*3*2 supercell of hcp-Zn.

[admin]

This is a numerical trap. Atoms placed in highly symmetric positions
(this is e.g. the case of pure metals, graphenes, etc.) can lead to singularities
and the numerical procedure of the electronic loop crashes.
Move the whole set of atoms slightly away from the "magic position" and the numerical
problem will disappear.