Cannot continue the AIMD calculation with DPKS

deepmodeling / abacus-develop

An electronic structure package based on either plane wave basis or numerical atomic orbitals.

http://abacus.ustc.edu.cn

GNU Lesser General Public License v3.0

161 stars 127 forks source link

Cannot continue the AIMD calculation with DPKS #2661

Closed iduygnay closed 1 year ago

iduygnay commented 1 year ago

Describe the bug

The issue is when I turn on the md_restart 1 (since it was out of memory before, I had to continue the calculation), it is reported that I cannot find the orbit files for descriptors.

I am attaching the file of the AIMD example with DPKS. I've tried if I turn off the DPKS, it works. If I turn off the md_restart, it also works. continue.zip

Expected behavior

No response

To Reproduce

No response

Environment

No response

Additional Context

No response

YuLiu98 commented 1 year ago

I have submitted PR #2670 to fix this issue.

As a temporary solution, you can add a part to the file OUT.ABACUS/STRU/STRU_MD_780, just like STRU:

NUMERICAL_DESCRIPTOR
jle.orb

iduygnay commented 1 year ago

I tried to add the numerical descriptor lines in the STRU file but still did not work.

iduygnay commented 1 year ago

new.tar.zip Maybe you can try this

YuLiu98 commented 1 year ago

I tried to add the numerical descriptor lines in the STRU file but still did not work.

In the case you provided. you should modify OUT.ABACUS/STRU/STRU_MD_1695 as follows:

ATOMIC_SPECIES
C 12.011 C_ONCV_PBE-1.0.upf upf201
N 14.007 N_ONCV_PBE-1.0.upf upf201
H 1.008 H_ONCV_PBE-1.0.upf upf201
Pb 207.2 Pb_ONCV_PBE-1.0.upf upf201
I 126.904 I_ONCV_PBE-1.0.upf upf201

NUMERICAL_ORBITAL
C_gga_9au_100Ry_2s2p1d.orb
N_gga_9au_100Ry_2s2p1d.orb
H_gga_9au_100Ry_2s1p.orb
Pb_gga_9au_100Ry_2s2p2d1f.orb
I_gga_9au_100Ry_2s2p2d1f.orb

NUMERICAL_DESCRIPTOR
jle.orb

LATTICE_CONSTANT
1.889716

LATTICE_VECTORS
12.724 0 0 #latvec1
0 12.724 0 #latvec2
0 0 12.724 #latvec3

Between the part NUMERICAL_ORBITAL and LATTICE_CONSTANT, please add a part of NUMERICAL_DESCRIPTOR, just like the format of STRU in your working directory.

iduygnay commented 1 year ago

This is the input STRU file

This is the output

I think I've done what you mentioned, but it still did not work. These files I've sent you before in the new.tar. If I did anything wrong just let me know, please. Thank you so much!

YuLiu98 commented 1 year ago

This is the input STRU file

This is the output

I think I've done what you mentioned, but it still did not work. These files I've sent you before in the new.tar. If I did anything wrong just let me know, please. Thank you so much!

You should modify OUT.ABACUS/STRU/STRU_MD_1695 in the new case, which is output latest in the directory OUT.ABACUS/STRU/.

And you must NOT move the modified file!

iduygnay commented 1 year ago

Now it works, sorry for the misunderstanding. Thank you so much!