Input File for a Ferromagnetism system

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
Najmeh
Posts: 6
Joined: Mon May 14, 2018 3:20 pm

Input File for a Ferromagnetism system

Post by Najmeh » Wed Nov 28, 2018 9:18 am

Hi Dear

I want to optimize a Ferromagnetism system. My input file is attached. It is around 2 weeks I run it with - np 32 but there is no answering.
Is my input file correct?

Other question:
Do the spin polarization and the LDA+U needed for the optimize?
I mean, can I optimize the ferromagnetism system without considering spin and +U in the input file?

Best regards,
Najmeh
Attachments
Gd2SiO5_Opt.in
(2.1 KiB) Downloaded 257 times

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: Input File for a Ferromagnetism system

Post by ebousquet » Wed Dec 05, 2018 10:17 am

Dear Najmeh,
In your case with Gd-f electrons, it is probably impossible to run without a +U.
Your input looks OK, what a "grep ETOT" on your output gives to see the SCF behavior?
It might be a convergence problem and thus some diemix, diemixmag, nilne, etc might be helpfull to converge the calculation.
Best wishes,
Eric

Najmeh
Posts: 6
Joined: Mon May 14, 2018 3:20 pm

Re: Input File for a Ferromagnetism system

Post by Najmeh » Sun Dec 23, 2018 1:30 pm

Dear Eric
Thanks a lot for your kindly attention,
According to your suggestion, I added diemix, diemixmag in my input file.
My input and output files are attached. I have a convergence problem yet, and I do not know what can I do.
I would appreciate it if you please help me.

Best wishes,
Najmeh
Attachments
Gd2SiO5_Opt.in
(2.06 KiB) Downloaded 246 times
Gd2SiO5_Opt.out
(399.11 KiB) Downloaded 228 times

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: Input File for a Ferromagnetism system

Post by ebousquet » Sun Jan 27, 2019 10:41 am

Dear Najmeh,
I do not understand, your input file does not correspond to your output file (much more atoms in the output than the input).
Regarding diemixmag, use rather positive value and a bit smaller than diemix. Start with a larger magnetic moment on spinat (10 instead of 7 to enforce high spin configuration, in your output the low spin is picked up at the end) and increase the nline to 10. If this is not enough use nnsclo=-4 or -5 (only allowed in the latest version of Abinit. Otherwise you can put positive values but this will increase a lot the time for one SCF, thought convergence will be better...).
Rare-earth with f in the valence are always tricky, let us know if this improves the convergence (it should).
Eric

Locked