dmft.x stopped with an error on "cor_at" numbers

[Yuan Yin]

Dear developers,

I got an error during running dmft.x program. The error shows that:

Traceback (most recent call last):
  File "/data/home/yinyuan/DMFTwDFT-old/bin/RUNDMFT.py", line 432, in <module>
    DMFT.Update_Nlatt(TB, p["nspin"])
  File "/data/home/yinyuan/DMFTwDFT-old/bin/DMFT_MOD.py", line 520, in Update_Nlatt
    self.N_latt[i,j]+=float(line[idx])
ValueError: could not convert string to float: #

The error indicates that there is no data in the INFO_DM file (exactly, no determinant element in INFO_DM after the error), but I don't know why the program stopped at this stage. My crystal structure is a 36-atom disordered iron supercell derived from molecular dynamics calculations. This error came when I specified more than 26 atoms as cor_at in INPUT.py. If I just add less than 16 atoms in cor_at in INPUT.py, the task runs well. Besides, the n_tot equals 288 (36*8) in INPUT.py. 

Could anyone help me to figure out what's wrong with my INPTU.py file?

Thank you so much!

Best wishes,

Yuan

[Uthpala Herath]

Hello Yuan, 

This is interesting, can you send me your files please and I shall take a look? 

Thank you,

Best,

Uthpala

--
You received this message because you are subscribed to the Google Groups "DMFTwDFT" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dmftwdft+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dmftwdft/324efc3d-76cd-442f-b283-7bed8a4c7da6n%40googlegroups.com.

--

Uthpala Herath

Postdoctoral Associate

Department of Mechanical Engineering and Materials Science

Duke University

Durham, NC

Tel. (304) 216-2535

Email: uthpala...@duke.edu

Website: https://uthpalaherath.github.io/ 

[Yuan Yin]

Hello, sir

Happy to hear from you. There are my calculation files in the zipped attachment. When I add more than 19 atoms in the "cor_at," i.g., Fe1 to Fe20, the dmft.x program will die abnormally at somewhile. You can find some valuable info in the error1.dat file.

Moreover, it is also weird that when I run the program on two or more nodes, such as 2 nodes* 24 cores/node, the program usually stops at dmft.x or ctqmc stage (precisely, only one ctqmc/dmft.x zombie process exists).  But if I kill this process id, the program will continue the calculation to the next stage, and everything works well till the next break on the stage of ctqmc (mainly). It is a random break that makes me have to keep an eye on the screen and be ready to kill the process immediately. Otherwise, the program will always stop at the stage of ctqmc or dmft.x but not automatically move to the next stage. This is terrible and sad.

Thanks, these are all my present obstacles to using dmftwdft, though it works well in most conditions.

Best wishes

Yuan

[Uthpala Herath]

Thanks for the files, Yuan. 

I will run this and see if I can reproduce this issue. 

Best,

Uthpala

To view this discussion on the web visit https://groups.google.com/d/msgid/dmftwdft/d18cdb45-78e6-4ca9-8d9e-ba1c414b1107n%40googlegroups.com.

[Uthpala Herath]

Hello Yuan, 

Sorry for the late response. I was caught up on some other work.

So I ran the calculation from my side and faced the exact same issue you did.

I'm trying to figure out what's going on so I'll get back to you soon.

I had one question for you. Why do you only consider 21 Fe atoms to be correlated and not the others?

Best,

Uthpala

On Thu, Feb 23, 2023 at 3:27 AM Yuan Yin <13030...@gmail.com> wrote:

Hello sir:

Do you reproduce this issue and see anything for the "cor_at" wrongs? I would like to know if this program can deal with a disordered supercell with 36 and more atoms.

Best!

Yuan

--

Uthpala Herath, Ph.D. 

[Uthpala Herath]

Hello Yuan, 

Did you set "Niter":   xx, to be larger than 1?

In your DMFT directory for a CSC run your vasp.out should display something like:

  LDA part: xc-table for Ceperly-Alder, standard interpolation
  found WAVECAR, reading the header
  POSCAR, INCAR and KPOINTS ok, starting setup
  FFT: planning ...
  reading WAVECAR
  the WAVECAR file was read successfully
  initial charge from wavefunction
  DFT+DMFT run with DMFTwDFT
  entering main loop
     ¦  N       E                     dE             d eps       ncg     rms          rms(c)
 DAV:   1    -0.824877003963E+02   -0.82488E+02   -0.90857E-02 12984   0.558E+00    0.151E+00
 DAV:   2    -0.819347040654E+02    0.55300E+00   -0.20385E-01 12966   0.906E+00    0.168E+00
 DAV:   3    -0.822241920828E+02   -0.28949E+00   -0.45741E-02 12378   0.483E+00    0.497E-01

Do you find this in your case?

Best,

Uthpala

On Thu, Feb 23, 2023 at 9:06 AM Yuan Yin <13030...@gmail.com> wrote:

Hello, developer:

I got a problem when I use the CSC model to calculate the LaNiO example. The DFT charge density was not updated from the DMFT charge density! I fully followed the compilation steps as shown on your githup website. During compilations, no wrongs. I want to know which subroutine is for updating the charge density to VASP? And if no ICHARG = 1, how the VASP run with the updated the charge density? I see another manu saying that dmft.x and libdmft.a are both to update charge density. So, what is the workflow of this step?

Thanks

Yours

Yuan

在2022年12月20日星期二 UTC+8 00:18:42<ukh0001> 写道：

--

Uthpala Herath, Ph.D. 

[Yuan Yin]

Hi developer:

I always got into trouble with calculations on a disordered 36-atom supercell of iron. There is always a mistake when the "cor_at" is set to more than 20 atoms. A report from ksum_error_dmft.x is shown as

I guess the INFO_DM file is unable to be written when the correlated atoms are beyond 20 because the second line in the INFO_DM file is more than 4000 columns (20 atoms * 20colu/orbital* 5 orbitals (for 3d) * 2 (spin)) in length. Thus, dmft.x program can not come out correct INFO_DM file but an empty one. Then, RUNDMFT.py got an error when reading the INFO_DM file.

Is that a bug, or does my INPUT file have any wrong format? hope you can tell me how to deal with this.

Best,

Yuan