Discrepancies in NELMDL Values in OUTCAR Files for Identical Optimization Jobs

[nidhi_tiwari]

In the context of the same optimization task, I noticed variations in the OUTCAR files. Upon comparing these files, I observed that the NELMDL tag was 0 in one instance and -12 in another, despite identical inputs (ISTART = 0, INIWAV = 1, IALGO = 48), with no explicit specification of the NELMDL tag in the INCAR. The expected default value for NELMDL in this scenario is -12. I am seeking an explanation for why NELMDL is registering as 0 in one case while maintaining the default -12 in the other. What factors could be contributing to these distinct values of NELMDL?

[fabien_tran1]

Without knowing more details about your calculations it is not possible to give any explanation. In particular:
Are all other tags in INCAR the same?
Were the files present in the directory when starting the calculation the same for both calculations?
Was the same VASP version used for both calculations?

[nidhi_tiwari]

Indeed, the input files are identical. I have attached the .zip folders containing two single-point energy calculations, each with a different POSCAR, executed on the same cluster. Please examine the OUTCAR files and specifically look for the NELMDL parameter. You will observe that in one OUTCAR, its value is 0, while in the other, it is -12.

[fabien_tran1]

By default, NELMDL=0 if the calculation starts with a WAVECAR file already present in the directory (which is not indicated on the wiki). Thus, your calculations IS_1 and IS_2 were started with and without a WAVECAR file present, respectively.

Nevertheless, the problem is that "NELMDL=0" is printed in OUTCAR even if the WAVECAR file that was present has a size of 0 (like in your IS_1 case). However, according to my tests it does not seem to affect the calculations, since it seems that NELMDL=-12 is used internally. Thus, "NELMDL=0" seems to be a wrong information in OUTCAR, and we should probably correct this in the code.

[nidhi_tiwari]

Thank you for your prompt response and assistance. Just to clarify, in both of these cases, I utilized only four input files without employing the WAVECAR.

[fabien_tran1]

If NELMDL=0 that means that there was a WAVECAR file in the directory, but probably it was an empty one.

[nidhi_tiwari]

Thank you; your explanation has clarified things for me. It seems I may have unintentionally retained a WAVECAR from a previous job. I'll ensure to be more attentive in the future.

[fabien_tran1]

The case "NELMDL=0" has just been added to the page NELMDL.