Coverage for local_installation/dynasor/trajectory/trajectory.py: 95%
153 statements
« prev ^ index » next coverage.py v7.3.2, created at 2024-09-27 15:43 +0000
« prev ^ index » next coverage.py v7.3.2, created at 2024-09-27 15:43 +0000
1__all__ = ['Trajectory', 'WindowIterator']
3import numpy as np
5from collections import deque
6from itertools import islice, chain
7from os.path import isfile
8from typing import Callable, Dict, Union, List
10from dynasor.trajectory.atomic_indices import parse_gromacs_index_file
11from dynasor.trajectory.ase_trajectory_reader import ASETrajectoryReader
12from dynasor.trajectory.extxyz_trajectory_reader import ExtxyzTrajectoryReader
13from dynasor.trajectory.lammps_trajectory_reader import LammpsTrajectoryReader
14from dynasor.trajectory.mdanalysis_trajectory_reader import MDAnalysisTrajectoryReader
15from dynasor.trajectory.trajectory_frame import TrajectoryFrame
16from dynasor.logging_tools import logger
19class Trajectory:
20 """Instances of this class hold trajectories in a format suitable for
21 the computation of correlation functions. They behave as
22 iterators, where each step returns the next frame as a
23 :class:`TrajectoryFrame` object. The latter hold information
24 regarding atomic positions, types, and velocities.
26 Parameters
27 ----------
28 filename
29 Name of input file
30 trajectory_format
31 Type of trajectory. Possible values are:
32 ``'lammps_internal'``, ``'extxyz'``, ``'ase'`` or one of the formats supported by
33 `MDAnalysis <https://www.mdanalysis.org/>`_ (except for ``'lammpsdump'``,
34 which can be called by specifying ``'lammps_mdanalysis'`` to avoid ambiguity)
35 atomic_indices
36 Specify which indices belong to which atom type. Can be
37 (1) a dictionary where the keys specicy species and the values are list of atomic indices,
38 (2) ``'read_from_trajectory'``, in which case the species are read from the trajectory or
39 (3) the path to an index file.
40 length_unit
41 Length unit of trajectory (``'Angstrom'``, ``'nm'``, ``'pm'``, ``'fm'``). Necessary for
42 correct conversion to internal dynasor units if the trajectory file does not contain unit
43 information.
44 time_unit
45 Time unit of trajectory (``'fs'``, ``'ps'``, ``'ns'``). Necessary for correct conversion to
46 internal dynasor units if the trajectory file does not contain unit information.
47 frame_start
48 First frame to read; must be larger or equal ``0``.
49 frame_stop
50 Last frame to read. By default (``None``) the entire trajectory is read.
51 frame_step
52 Read every :attr:`frame_step`-th step of the input trajectory.
53 By default (``1``) every frame is read. Must be larger than ``0``.
55 """
56 def __init__(
57 self,
58 filename: str,
59 trajectory_format: str,
60 atomic_indices: Union[str, Dict[str, List[int]]] = None,
61 length_unit: str = 'Angstrom',
62 time_unit: str = 'fs',
63 frame_start: int = 0,
64 frame_stop: int = None,
65 frame_step: int = 1
66 ):
68 if frame_start < 0:
69 raise ValueError('frame_start should be positive')
70 if frame_step < 0:
71 raise ValueError('frame_step should be positive')
73 self._frame_number = frame_start
74 self._frame_step = frame_step
75 self._frame_stop = frame_stop
77 # setup trajectory reader
78 if not isfile(filename):
79 raise IOError(f'File {filename} does not exist')
80 self._filename = filename
82 if trajectory_format == 'lammps_internal':
83 reader = LammpsTrajectoryReader
84 elif trajectory_format == 'extxyz':
85 reader = ExtxyzTrajectoryReader
86 elif trajectory_format == 'lammps_mdanalysis':
87 reader = MDAnalysisTrajectoryReader
88 trajectory_format = 'lammpsdump'
89 elif trajectory_format == 'ase': 89 ↛ 90line 89 didn't jump to line 90, because the condition on line 89 was never true
90 reader = ASETrajectoryReader
91 elif trajectory_format == 'lammps':
92 raise IOError('Ambiguous trajectory format, '
93 'did you mean lammps_internal or lammps_mdanalysis?')
94 else:
95 reader = MDAnalysisTrajectoryReader
97 logger.debug(f'Using trajectory reader: {reader.__name__}')
98 if reader == MDAnalysisTrajectoryReader:
99 self._reader_obj = reader(self._filename, trajectory_format,
100 length_unit=length_unit, time_unit=time_unit)
101 else:
102 self._reader_obj = reader(self._filename, length_unit=length_unit, time_unit=time_unit)
104 # Get two frames to set cell etc.
105 frame0 = next(self._reader_obj)
106 frame1 = next(self._reader_obj)
107 self._cell = frame0.cell
108 self._n_atoms = frame0.n_atoms
110 # Make sure cell is not changed during consecutive frames
111 if not np.allclose(frame0.cell, frame1.cell):
112 raise ValueError('The cell changes between the first and second frame. '
113 'The concept of q-points becomes muddy if the simulation cell is '
114 'changing, such as during NPT MD simulations, so trajectories where '
115 'the cell changes are not supported by dynasor.')
117 # setup iterator slice (reuse frame0 and frame1 via chain)
118 self.number_of_frames_read = 0
119 self._reader_obj = islice(chain([frame0, frame1], self._reader_obj),
120 frame_start, self._frame_stop, self._frame_step)
122 # setup atomic indices
123 if atomic_indices is None: # Default behaviour
124 atomic_indices = {'X': np.arange(0, self.n_atoms)}
125 elif isinstance(atomic_indices, str): # Str input
126 if atomic_indices == 'read_from_trajectory':
127 if frame0.atom_types is None:
128 raise ValueError('Could not read atomic indices from the trajectory.')
129 else:
130 uniques = np.unique(frame0.atom_types)
131 atomic_indices = {uniques[i]: (frame0.atom_types == uniques[i]).nonzero()[0]
132 for i in range(len(uniques))}
133 else:
134 atomic_indices = parse_gromacs_index_file(atomic_indices)
135 elif isinstance(atomic_indices, dict): # Dict input
136 pass
137 else:
138 raise ValueError('Could not understand atomic_indices.')
139 self._atomic_indices = atomic_indices
141 # sanity checks for atomic_indices
142 for key, indices in self._atomic_indices.items():
143 if np.max(indices) > self.n_atoms:
144 raise ValueError('maximum index in atomic_indices exceeds number of atoms')
145 if np.min(indices) < 0:
146 raise ValueError('minimum index in atomic_indices is negative')
147 if '_' in key:
148 # Since '_' is what we use to distinguish atom types in the results, e.g. Sqw_Cs_Pb
149 raise ValueError('The char "_" is not allowed in atomic_indices.')
151 # log info on trajectory and atom types etc
152 logger.info(f'Trajectory file: {self.filename}')
153 logger.info(f'Total number of particles: {self.n_atoms}')
154 logger.info(f'Number of atom types: {len(self.atom_types)}')
155 for atom_type, indices in self._atomic_indices.items():
156 logger.info(f'Number of atoms of type {atom_type}: {len(indices)}')
157 logger.info(f'Simulation cell (in Angstrom):\n{str(self._cell)}')
159 def __iter__(self):
160 return self
162 def __next__(self):
163 frame = next(self._reader_obj)
164 new_frame = TrajectoryFrame(self.atomic_indices, frame.frame_index, frame.positions,
165 frame.velocities)
166 self.number_of_frames_read += 1
167 return new_frame
169 def __str__(self) -> str:
170 s = ['Trajectory']
171 s += ['{:12} : {}'.format('filename', self.filename)]
172 s += ['{:12} : {}'.format('natoms', self.n_atoms)]
173 s += ['{:12} : {}'.format('frame_step', self.frame_step)]
174 s += ['{:12} : [{}\n {}\n {}]'
175 .format('cell', self.cell[0], self.cell[1], self.cell[2])]
176 return '\n'.join(s)
178 def _repr_html_(self) -> str:
179 s = [f'<h3>{self.__class__.__name__}</h3>']
180 s += ['<table border="1" class="dataframe">']
181 s += ['<thead><tr><th style="text-align: left;">Field</th><th>Value</th></tr></thead>']
182 s += ['<tbody>']
183 s += [f'<tr"><td style="text-align: left;">File name</td><td>{self.filename}</td></tr>']
184 s += [f'<tr><td style="text-align: left;">Number of atoms</td><td>{self.n_atoms}</td></tr>']
185 s += [f'<tr><td style="text-align: left;">Cell metric</td><td>{self.cell}</td></tr>']
186 s += [f'<tr><td style="text-align: left;">Frame step</td><td>{self.frame_step}</td></tr>']
187 s += [f'<tr><td style="text-align: left;">Atom types</td><td>{self.atom_types}</td></tr>']
188 s += ['</tbody>']
189 s += ['</table>']
190 return '\n'.join(s)
192 @property
193 def cell(self):
194 """ Simulation cell """
195 return self._cell
197 @property
198 def n_atoms(self):
199 """ Number of atoms """
200 return self._n_atoms
202 @property
203 def filename(self):
204 """ The trajectory filename """
205 return self._filename
207 @property
208 def atomic_indices(self):
209 """ Return copy of index arrays """
210 atomic_indices = dict()
211 for name, inds in self._atomic_indices.items():
212 atomic_indices[name] = inds.copy()
213 return atomic_indices
215 @property
216 def atom_types(self) -> List[str]:
217 return sorted(self._atomic_indices.keys())
219 @property
220 def frame_step(self):
221 """ Frame to access, trajectory will return every :attr:`frame_step`-th snapshot """
222 return self._frame_step
225def consume(iterator, n):
226 """ Advance the iterator by :attr:`n` steps. If :attr:`n` is ``None``, consume entirely. """
227 # From the python.org
228 if n is None: 228 ↛ 229line 228 didn't jump to line 229, because the condition on line 228 was never true
229 deque(iterator, maxlen=0)
230 else:
231 next(islice(iterator, n, n), None)
234class WindowIterator:
235 """Sliding window iterator.
237 Returns consecutive windows (a window is represented as a list
238 of objects), created from an input iterator.
240 Parameters
241 ----------
242 itraj
243 Trajectory object
244 width
245 Length of window (``window_size`` + 1)
246 window_step
247 Distance between the start of two consecutive window frames
248 element_processor
249 Enables processing each non-discarded object; useful if ``window_step >
250 width`` and ``map_item`` is expensive (as compared to directly passing
251 ``map(fun, itraj)`` as ``itraj``); if ``window_step < width``, you could as
252 well directly pass ``map(fun, itraj)``.
253 """
254 def __init__(self,
255 itraj: Trajectory,
256 width: int,
257 window_step: int = 1,
258 element_processor: Callable = None):
260 self._raw_it = itraj
261 if element_processor:
262 self._it = map(element_processor, self._raw_it)
263 else:
264 self._it = self._raw_it
265 assert window_step >= 1
266 assert width >= 1
267 self.width = width
268 self.window_step = window_step
269 self._window = None
271 def __iter__(self):
272 return self
274 def __next__(self):
275 """ Returns next element in sequence. """
276 if self._window is None:
277 self._window = deque(islice(self._it, self.width), self.width)
278 else:
279 if self.window_step >= self.width:
280 self._window.clear()
281 consume(self._raw_it, self.window_step - self.width)
282 else:
283 for _ in range(min((self.window_step, len(self._window)))):
284 self._window.popleft()
285 for f in islice(self._it, min((self.window_step, self.width))):
286 self._window.append(f)
288 if len(self._window) == 0:
289 raise StopIteration
291 return list(self._window)