MDAnalysis, OVITO and notebooks

pydseams.adapters connects the engine to the tools a trajectory already lives in.

MDAnalysis

IceStates is an AnalysisBase subclass over an oxygen AtomGroup; ions are a second group read against the assignment. Boxes must be orthorhombic.

import MDAnalysis as mda
from pydseams.adapters import IceStates

u = mda.Universe("brine.gro", "brine.xtc")
an = IceStates(u.select_atoms("name OW"), ions=u.select_atoms("name NA CL")).run()
an.results.states        # frames x oxygens, codes in an.results.state_names
an.results.features      # frames x len(an.results.names)
an.results.ion_states    # frames x ions, codes in an.results.ion_state_names

OVITO

ice_states is a Python modifier function. Append it to a pipeline and colour by the Ice state particle property (0 water, 1 cubic, 2 hexagonal, 3 mixed; -1 outside the oxygen type).

from functools import partial
from ovito.io import import_file
from pydseams.adapters import ice_states

pipeline = import_file("dump.lammpstrj")
pipeline.modifiers.append(partial(ice_states, oxygen_type=1, cutoff=3.5))
data = pipeline.compute()
print(data.particles["Ice state"][...])

Triclinic OVITO cells pass through as LAMMPS bound spans and tilts.

Command line

Without Python, seams cages dump.lammpstrj --per-atom labelled.lammpstrj appends a dump frame with a cage column (0 water, 1 hexagonal cage, 2 double-diamond cage, 3 both); fingerprint writes class or label and ions writes state. Any visualiser that reads LAMMPS dumps colours by that column.

Notebook

notebooks/classify_ice.ipynb installs the wheel, fetches one frame of the cubic mW lattice and runs the assignment, CHILL+ and a fingerprint; the README carries a Colab badge for it.