Is there any limitation in no of different type of elements while using OUTCAR

[chayanroy3]

Hello, I am using a OUTCAR generated from VASP simulation as initial data for training. There in the simmulation I have 27 different types of atoms. While making deepmd/raw file for training using dpdata is it taking only first 10 atom types into consideration. Is there any such limitation ? For training in GPU system we can use 128 types of atoms but while generating initial data for traing , it is not taking. Thanks in advanced for reply.

[wanghan-iapcm]

dpdata has no such limitation.

Please provide the minimal example to reproduce the issue so we may further help.

[chayanroy3]

Here we are trying to train a model to run lammps simulations on a system where we need to define different atoms of similar element. Please look into the files. We are beginner in this field, thanks for the help.
file: https://filetransfer.io/data-package/1UseUkEm#link

[wanghan-iapcm]

I did not see 27 types from your OUTCAR. atoms with the same name would not be distinguished as different types.

$ grep PBE OUTCAR  | grep TITEL | wc
     27     135    2727
$ grep PBE OUTCAR  | grep TITEL
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE O 08Apr2002                                                                     
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE O 08Apr2002                                                                     
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE C 08Apr2002                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE H 15Jun2001                                                                     
   TITEL  = PAW_PBE Na_pv 19Sep2006                                                                 
   TITEL  = PAW_PBE S 06Sep2000                                                                     
   TITEL  = PAW_PBE N 08Apr2002                                                                     
   TITEL  = PAW_PBE O 08Apr2002                                                                     
   TITEL  = PAW_PBE F 08Apr2002  

[chayanroy3]

Please go through the type.raw and type_map.raw. Only the first 90 atoms are read where we have 243 atoms in outcar and poscar.

[wanghan-iapcm]

The problem is this line

ions per type =               9   9   9   9   9   9   9   9   9   9 

the OUTCAR only contains the number of ions per type for the first 10 ion types. If we look at the vasp source code, we have

       '   ions per type =            ',10I4/)

namely only 10 number will be written. W

We cannot fix the issue from the dpdata side. Several workarounds :

1. try to load the xml file by using fmt='vasp/xml'
2. lump the atoms in different sections by their types. You only have 7 elements, the lumped outcar would have correct output of ions per type.

[chayanroy3]

Thank you very much for your quick response and detailed explaination. I will try as you have suggested.