MLFF issue with POMASS tag in INCAR and multiple group of same atoms in POSCAR

[ralf_tonner1]

Dear Admin,
1. I am running MLFF on a system containing H-atoms.
This post suggested "to increase the hydrogen mass in the POTCAR file".
But this best practice page in VASP wiki stated "increase the light element mass (POMASS) in the INCAR or the POTCAR file." The example INCAR file at the end of the same page defines the POMASS in the INCAR file. However, if I do not change the POMASS in the POTCAR as well I get the following warning in the standard output:

Code: Select all

 WARNING: type information on POSCAR and POTCAR are incompatible
 POTCAR overwrites the type information in POSCAR
 typ   3 type information:  H1 H
 WARNING: mass on POTCAR and INCAR are incompatible
  typ           3  Mass   8.00000000000000        1.00000000000000

Does POTCAR overwrites the POMASS as well? If so, then one must has to change the POMASS in POTCAR.
* Attached: TEST_POMASS inside TEST_MLFF.zip can be used to simulate the above (not with H but with Si and Ge).

2. MLFF is not working when using two groups of same 'Si' atoms in the POSCAR.

Code: Select all

Si1    Si2
63     1         

* Attached: TEST_Si1_Si2 inside TEST_MLFF.zip can be used to simulate the above.
* The similar worked when using e.g. 'H1 H2', 'O1 O2'. The tests are not attached.

VASP version: 6.4.2
* Tests were run only for 2 mins and, hence, the simulations were not finished.

Any suggestions/comments please.

Thank you

[ferenc_karsai]

1) POMASS in the INCAR succesfully overwrites the mass. You can see that from the OUTCAR file if you grap for POMASS that the ones specified in the INCAR file are used. There is only a warning message to make the user aware that the masses on the two files are not the same.

2) Unfortunately VASP only allows two characters for the type name. In your case you have 3 "Si" and the number. What the code does it just truncates the last character and for both you only see "Si" and it judges them as the same. So to make your example work you would have to set "S1" and "S2". It is clumsy, but it would be cumbersome to change everything in the code ad hoc. I will ask the bosses if we can add more characters (at least 1).

[ralf_tonner1]

1) Ah. Sorry. That was silly. I could have checked that. Thank you.
Just mere suggestion:
a. Can it be possible to include this information in the POMASS vasp wiki?
"POMASS in the INCAR overwrites the mass from POTCAR with a warning message - 'WARNING: mass on POTCAR and INCAR are incompatible' - in the standard output. You can see that from the OUTCAR file if you grep for POMASS that the ones specified in the INCAR file are used. There is only a warning message to make the user aware that the masses on the two files are not the same."
b. Can it be possible to print the information in the standard output?
"INCAR overwrites the mass information in POTCAR."
I personally got worried in the first place (without double checking with OUTCAR) with the warning messages as provided in my first post because after the 1st warning msg it printed "POTCAR overwrites type information".

2) Thank you.
Suggestion: Can it be possible to include this information in the wiki? Probably here and here.
Note: The TEST example that I provided would be useful in study such as to simulate defects in systems.
Remark: "VASP only allows two characters for the type name." => probably makes sense, given the elements in the periodic table. Not sure adding one more character would introduce unwanted consequences.
But thank you for your suggestion of using 'S1', 'S2'. I know it does not matter, but I wanted to avoid confusion with sulfur in my 'actual' system. Instead, I decided to use 'Si' and 'SI'. Notation ==> "Capital-small", "Capital-Capital" for all the elements with two characters.

Thank you

[ferenc_karsai]

Yes, we should describe both cases in the wiki. Thank you for pointing that out.