Attempts to use NS calculator
Nested sampling package brings its own calculator (modified from LAMMPS calculator). The way I was using is ilustrated in the following code:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
from ase import Atom, Atoms
from lammpslib import LAMMPSlib # <<-- NS calculator
import os
cmds = ["pair_style mlip mlip.ini",\
"pair_coeff * * " ]
positions = [ (0.0, 0.0, 0.0), (0.0, 0.0, 5.0) ]
atoms = Atoms( 'AlW', positions=positions , cell=(a, a, a), pbc=True )
struct.get_positions()
absoletPath2file = os.getcwd() + "/log.txt"
mylammps = LAMMPSlib(lmpcmds=cmds, atom_types={13:1, 74:2}, keep_alive=True, log_file=absoletPath2file)
atoms.set_calculator(mylammps)
atoms.get_potential_energy(atoms)
# atoms.set_calculator(mylammps)
atoms.get_forces(atoms)
# atoms.set_calculator(mylammps)
atoms._calc.get_forces(atoms)
where 13 and 74 are the atomic numbers of the elements defined in atoms. I think I should had used atoms.get_potential_energy() instead of atoms.get_potential_energy(atoms). It is important the keyword keep_alive=True, otherwise the calculator would need to be attached repeatedly as shown in the comments above.
The atoms_types keyword is not needed in the LAMMSP calculator (LAMMPS determines it automatically). NS code asks for this keyword because it uses its own system to call the participant atoms in the calculation, i.e, 1, 2, …