[ase-users] Optimization with the genetic algorithm and VASP

[Kondov, Ivan (SCC)]

Dear all,

i am trying to get the GA optimization working with VASP as calculator. When i just take the sample scripts from

https://wiki.fysik.dtu.dk/ase/tutorials/ga/ga_optimize.html

and substitute the EMT calculator with VASP i get the following error right after the first structure is optimized:

Relaxing starting candidate 2
Traceback (most recent call last):
  File "./ga-example.py", line 52, in <module>
    da.add_relaxed_step(a)
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/ga/data.py", line 116, in add_relaxed_step
    data=a.info['data'])
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/db/core.py", line 140, in new_method
    return method(self, *args, **kwargs)
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/db/core.py", line 201, in write
    id = self._write(atoms, kvp, data)
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/db/sqlite.py", line 190, in _write
    row = AtomsRow(atoms)
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/db/row.py", line 67, in __init__
    dct = atoms2dict(dct)
  File "/myhome/anaconda4/lib/python2.7/site-packages/ase/db/row.py", line 49, in atoms2dict
    dct['calculator_parameters'] = atoms.calc.todict()
AttributeError: Vasp instance has no attribute 'todict'

The VASP calculations and BFGS work with no error. Obviously the Vasp calculator cannot be saved in the database together with the Atoms object because todict() method is missing. I will appreciate very much if somebody has a clue how to deal with this.

Best regards,
Ivan