8594096068

Committed 17 May 2023 05:43AM UTC coverage: 81.442% (+0.5%) from 80.955%

Build # 8594096068

Build Type

push

github

Committed by

web-flow

Commit Message

use pre-commit.ci (#597)

* use pre-commit.ci

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* disable codespell, enable ruff autofix

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* disable some pre-commits

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* reset

* fix comment

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Run Details

1050 of 1379 branches covered (76.14%)

Branch coverage included in aggregate %.

2 of 2 new or added lines in 2 files covered. (100.0%)

2 existing lines in 2 files now uncovered.

4124 of 4974 relevant lines covered (82.91%)

0.83 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

96.67

/mdsuite/calculators/einstein_diffusion_coefficients.py

"""
MDSuite: A Zincwarecode package.

License
-------
This program and the accompanying materials are made available under the terms
of the Eclipse Public License v2.0 which accompanies this distribution, and is
available at https://www.eclipse.org/legal/epl-v20.html

SPDX-License-Identifier: EPL-2.0

Copyright Contributors to the Zincwarecode Project.

Contact Information
-------------------
email: zincwarecode@gmail.com
github: https://github.com/zincware
web: https://zincwarecode.com/

Citation
--------
If you use this module please cite us with:

Summary
-------
Module for the computation of self-diffusion coefficients using the Einstein method.
"""
from __future__ import annotations

import logging
from abc import ABC
from dataclasses import dataclass
from typing import Any, List, Union

import numpy as np
import tensorflow as tf
from bokeh.models import HoverTool, LinearAxis, Span
from bokeh.models.ranges import Range1d
from bokeh.plotting import figure
from tqdm import tqdm

from mdsuite import utils
from mdsuite.calculators.calculator import call
from mdsuite.calculators.trajectory_calculator import TrajectoryCalculator
from mdsuite.database.mdsuite_properties import mdsuite_properties
from mdsuite.utils.calculator_helper_methods import fit_einstein_curve

log = logging.getLogger(__name__)


@dataclass
class Args:
    """Data class for the saved properties."""

    data_range: int
    correlation_time: int
    atom_selection: np.s_
    tau_values: np.s_
    molecules: bool
    species: list
    fit_range: int


class EinsteinDiffusionCoefficients(TrajectoryCalculator, ABC):
    """
    Class for the Einstein diffusion coefficient implementation.

    Attributes
    ----------
    msd_array : np.ndarray
            MSd data updated during each ensemble computation.

    See Also
    --------
    mdsuite.calculators.calculator.Calculator class

    Examples
    --------
    project.experiment.run.EinsteinDiffusionCoefficients(data_range=500,
                                                         plot=True,
                                                         correlation_time=10)
    """

    def __init__(self, **kwargs):
        """
        Parameters
        ----------
        experiment :  Experiment
                Experiment class to call from
        experiments :  Experiment
                Experiment classes to call from
        """
        super().__init__(**kwargs)
        self.scale_function = {"linear": {"scale_factor": 150}}
        self.loaded_property = mdsuite_properties.unwrapped_positions
        self.x_label = r"$$\text{Time} / s$$"
        self.y_label = r"$$\text{MSD} / m^{2}$$"
        self.result_keys = [
            "diffusion_coefficient",
            "uncertainty",
            "gradient",
            "intercept",
        ]
        self.result_series_keys = ["time", "msd", "gradients", "gradient_errors"]
        self.analysis_name = "Einstein Self-Diffusion Coefficients"
        self._dtype = tf.float64

        self.msd_array = None

        log.info("starting Einstein Diffusion Computation")

    @call
    def __call__(
        self,
        plot: bool = True,
        species: list = None,
        data_range: int = 100,
        correlation_time: int = 1,
        atom_selection: np.s_ = np.s_[:],
        molecules: bool = False,
        tau_values: Union[int, List, Any] = np.s_[:],
        fit_range: int = -1,
    ):
        """

        Parameters
        ----------
        plot : bool
                if true, plot the output.
        species : list
                List of species on which to operate.
        data_range : int
                Data range to use in the analysis.
        correlation_time : int
                Correlation time to use in the window sampling.
        atom_selection : np.s_
                Selection of atoms to use within the HDF5 database.
        molecules : bool
                If true, molecules are used instead of atoms.
        tau_values : Union[int, list, np.s_]
                Selection of tau values to use in the window sliding.

        Returns
        -------
        None
        """
        if species is None:
            if molecules:
                species = list(self.experiment.molecules)
            else:
                species = list(self.experiment.species)

        if fit_range == -1:
            fit_range = int(data_range - 1)
        # set args that will affect the computation result
        self.args = Args(
            data_range=data_range,
            correlation_time=correlation_time,
            atom_selection=atom_selection,
            tau_values=tau_values,
            molecules=molecules,
            species=species,
            fit_range=fit_range,
        )
        self.plot = plot
        self.system_property = False

    def ensemble_operation(self, ensemble):
        """
        Calculate and return the msd.

        Parameters
        ----------
        ensemble : tf.Tensor
                An ensemble of data to be operated on.

        Returns
        -------
        MSD of the tensor_values.
        """
        msd = tf.math.squared_difference(
            tf.gather(ensemble, self.args.tau_values, axis=1), ensemble[:, None, 0]
        )
        self.count += msd.shape[0]
        # average over particles, sum over dimensions
        # msd = tf.reduce_sum(tf.reduce_mean(msd, axis=0), axis=-1)
        msd = tf.reduce_sum(tf.reduce_sum(msd, axis=0), axis=-1)

        # sum up ensembles to average in post processing
        return np.array(msd)

    def fit_diff_coeff(self):
        """Apply unit conversion, fit line to the data, prepare for database storage."""
        # self.msd_array /= int(self.n_batches) * self.ensemble_loop
        self.msd_array /= self.count

        self.msd_array *= self.experiment.units.length**2
        self.time *= self.experiment.units.time

        fit_values, covariance, gradients, gradient_errors = fit_einstein_curve(
            x_data=self.time, y_data=self.msd_array, fit_max_index=self.args.fit_range
        )
        error = np.sqrt(np.diag(covariance))[0]

        data = {
            self.result_keys[0]: 1 / 6.0 * fit_values[0],
            self.result_keys[1]: 1 / 6.0 * error,
            self.result_keys[2]: fit_values[0],
            self.result_keys[3]: fit_values[1],
            self.result_series_keys[0]: self.time.tolist(),
            self.result_series_keys[1]: self.msd_array.tolist(),
            self.result_series_keys[2]: (np.array(gradients) / 6).tolist(),
            self.result_series_keys[3]: (np.array(gradient_errors) / 6).tolist(),
        }
        return data

    def run_calculator(self):
        """Run analysis."""
        self._run_dependency_check()
        for species in self.args.species:
            # Here for now to avoid issues. Should be moved out when calculators become
            # species-wise
            self.time = None
            self.time = self._handle_tau_values()
            dict_ref = str.encode("/".join([species, self.loaded_property.name]))
            batch_ds = self.get_batch_dataset([species])
            self.msd_array = np.zeros(self.data_resolution)
            self.count = 0
            # loop over batches to get MSD
            for i, batch in tqdm(
                enumerate(batch_ds),
                ncols=70,
                desc=species,
                total=self.n_batches,
                disable=self.memory_manager.minibatch,
            ):
                ensemble_ds = self.get_ensemble_dataset(batch, species)

                for ensemble in ensemble_ds:
                    if not ensemble[dict_ref].shape[1] == self.args.data_range:
                        continue
                    else:
                        self.msd_array += self.ensemble_operation(ensemble[dict_ref])
                        self.count += 1

            # self.msd_array = np.array(tf.reduce_sum(self.msd_array, axis=0))
            fit_results = self.fit_diff_coeff()
            self.queue_data(data=fit_results, subjects=[species])

    def plot_data(self, data):
        """
        Plot the Einstein fits.

        Parameters
        ----------
        data

        Returns
        -------

        """
        for selected_species, val in data.items():
            fig = figure(x_axis_label=self.x_label, y_axis_label=self.y_label)

            gradients = np.array(val[self.result_series_keys[2]])
            gradient_errors = np.array(val[self.result_series_keys[3]])

            time = np.array(val[self.result_series_keys[0]])
            msd = np.array(val[self.result_series_keys[1]])

            fig.y_range = Range1d(-0.0, 1.1 * max(msd))

            # Compute the span
            span = Span(
                location=time[self.args.fit_range],
                dimension="height",
                line_dash="dashed",
            )
            # Compute msd and fit lines
            fig.line(
                time,
                msd,
                color=utils.Colour.ORANGE,
                legend_label=(
                    f"{selected_species}: {val[self.result_keys[0]]: 0.3E} +-"
                    f" {val[self.result_keys[1]]: 0.3E}"
                ),
            )
            fit_data = val[self.result_keys[2]] * time + val[self.result_keys[3]]
            fig.line(time, fit_data, color=utils.Colour.PAUA, legend_label="Curve fit.")
            fig.extra_y_ranges = {
                "Diff_range": Range1d(
                    start=0.999 * min(gradients), end=1.001 * max(gradients)
                )
            }

            fig.add_layout(
                LinearAxis(
                    y_range_name="Diff_range",
                    axis_label=r"$$\text{Diffusion Coefficient} / m^{2}s^{-1}$$",
                ),
                "right",
            )
            grad_time = time[-len(gradients) :]
            fig.line(
                grad_time,
                gradients,
                y_range_name="Diff_range",
                color=utils.Colour.MULBERRY,
            )
            fig.varea(
                grad_time,
                gradients - gradient_errors,
                gradients + gradient_errors,
                alpha=0.3,
                color=utils.Colour.ORANGE,
                y_range_name="Diff_range",
            )

            fig.add_tools(HoverTool())
            fig.add_layout(span)
            self.plot_array.append(fig)

1	"""
2	MDSuite: A Zincwarecode package.
3
4	License
5	-------
6	This program and the accompanying materials are made available under the terms
7	of the Eclipse Public License v2.0 which accompanies this distribution, and is
8	available at https://www.eclipse.org/legal/epl-v20.html
9
10	SPDX-License-Identifier: EPL-2.0
11
12	Copyright Contributors to the Zincwarecode Project.
13
14	Contact Information
15	-------------------
16	email: zincwarecode@gmail.com
17	github: https://github.com/zincware
18	web: https://zincwarecode.com/
19
20	Citation
21	--------
22	If you use this module please cite us with:
23
24	Summary
25	-------
26	Module for the computation of self-diffusion coefficients using the Einstein method.
27	"""
28	from __future__ import annotations	1✔
29
30	import logging	1✔
31	from abc import ABC	1✔
32	from dataclasses import dataclass	1✔
33	from typing import Any, List, Union	1✔
34
35	import numpy as np	1✔
36	import tensorflow as tf	1✔
37	from bokeh.models import HoverTool, LinearAxis, Span	1✔
38	from bokeh.models.ranges import Range1d	1✔
39	from bokeh.plotting import figure	1✔
40	from tqdm import tqdm	1✔
41
42	from mdsuite import utils	1✔
43	from mdsuite.calculators.calculator import call	1✔
44	from mdsuite.calculators.trajectory_calculator import TrajectoryCalculator	1✔
45	from mdsuite.database.mdsuite_properties import mdsuite_properties	1✔
46	from mdsuite.utils.calculator_helper_methods import fit_einstein_curve	1✔
47
48	log = logging.getLogger(__name__)	1✔
49
50
51	@dataclass	1✔
52	class Args:	1✔
53	"""Data class for the saved properties."""
54
55	data_range: int	1✔
56	correlation_time: int	1✔
57	atom_selection: np.s_	1✔
58	tau_values: np.s_	1✔
59	molecules: bool	1✔
60	species: list	1✔
61	fit_range: int	1✔
62
63
64	class EinsteinDiffusionCoefficients(TrajectoryCalculator, ABC):	1✔
65	"""
66	Class for the Einstein diffusion coefficient implementation.
67
68	Attributes
69	----------
70	msd_array : np.ndarray
71	MSd data updated during each ensemble computation.
72
73	See Also
74	--------
75	mdsuite.calculators.calculator.Calculator class
76
77	Examples
78	--------
79	project.experiment.run.EinsteinDiffusionCoefficients(data_range=500,
80	plot=True,
81	correlation_time=10)
82	"""
83
84	def __init__(self, **kwargs):	1✔
85	"""
86	Parameters
87	----------
88	experiment : Experiment
89	Experiment class to call from
90	experiments : Experiment
91	Experiment classes to call from
92	"""
93	super().__init__(**kwargs)	1✔
94	self.scale_function = {"linear": {"scale_factor": 150}}	1✔
95	self.loaded_property = mdsuite_properties.unwrapped_positions	1✔
96	self.x_label = r"$$\text{Time} / s$$"	1✔
97	self.y_label = r"$$\text{MSD} / m^{2}$$"	1✔
98	self.result_keys = [	1✔
99	"diffusion_coefficient",
100	"uncertainty",
101	"gradient",
102	"intercept",
103	]
104	self.result_series_keys = ["time", "msd", "gradients", "gradient_errors"]	1✔
105	self.analysis_name = "Einstein Self-Diffusion Coefficients"	1✔
106	self._dtype = tf.float64	1✔
107
108	self.msd_array = None	1✔
109
110	log.info("starting Einstein Diffusion Computation")	1✔
111
112	@call	1✔
113	def __call__(	1✔
114	self,
115	plot: bool = True,
116	species: list = None,
117	data_range: int = 100,
118	correlation_time: int = 1,
119	atom_selection: np.s_ = np.s_[:],
120	molecules: bool = False,
121	tau_values: Union[int, List, Any] = np.s_[:],
122	fit_range: int = -1,
123	):
124	"""
125
126	Parameters
127	----------
128	plot : bool
129	if true, plot the output.
130	species : list
131	List of species on which to operate.
132	data_range : int
133	Data range to use in the analysis.
134	correlation_time : int
135	Correlation time to use in the window sampling.
136	atom_selection : np.s_
137	Selection of atoms to use within the HDF5 database.
138	molecules : bool
139	If true, molecules are used instead of atoms.
140	tau_values : Union[int, list, np.s_]
141	Selection of tau values to use in the window sliding.
142
143	Returns
144	-------
145	None
146	"""
147	if species is None:	1!
148	if molecules:	1✔
149	species = list(self.experiment.molecules)	1✔
150	else:
151	species = list(self.experiment.species)	1✔
152
153	if fit_range == -1:	1!
154	fit_range = int(data_range - 1)	1✔
155	# set args that will affect the computation result
156	self.args = Args(	1✔
157	data_range=data_range,
158	correlation_time=correlation_time,
159	atom_selection=atom_selection,
160	tau_values=tau_values,
161	molecules=molecules,
162	species=species,
163	fit_range=fit_range,
164	)
165	self.plot = plot	1✔
166	self.system_property = False	1✔
167
168	def ensemble_operation(self, ensemble):	1✔
169	"""
170	Calculate and return the msd.
171
172	Parameters
173	----------
174	ensemble : tf.Tensor
175	An ensemble of data to be operated on.
176
177	Returns
178	-------
179	MSD of the tensor_values.
180	"""
181	msd = tf.math.squared_difference(	1✔
182	tf.gather(ensemble, self.args.tau_values, axis=1), ensemble[:, None, 0]
183	)
184	self.count += msd.shape[0]	1✔
185	# average over particles, sum over dimensions
186	# msd = tf.reduce_sum(tf.reduce_mean(msd, axis=0), axis=-1)
187	msd = tf.reduce_sum(tf.reduce_sum(msd, axis=0), axis=-1)	1✔
188
189	# sum up ensembles to average in post processing
190	return np.array(msd)	1✔
191
192	def fit_diff_coeff(self):	1✔
193	"""Apply unit conversion, fit line to the data, prepare for database storage."""
194	# self.msd_array /= int(self.n_batches) * self.ensemble_loop
195	self.msd_array /= self.count	1✔
196
197	self.msd_array = self.experiment.units.length*2	1✔
198	self.time *= self.experiment.units.time	1✔
199
200	fit_values, covariance, gradients, gradient_errors = fit_einstein_curve(	1✔
201	x_data=self.time, y_data=self.msd_array, fit_max_index=self.args.fit_range
202	)
203	error = np.sqrt(np.diag(covariance))[0]	1✔
204
205	data = {	1✔
206	self.result_keys[0]: 1 / 6.0 * fit_values[0],
207	self.result_keys[1]: 1 / 6.0 * error,
208	self.result_keys[2]: fit_values[0],
209	self.result_keys[3]: fit_values[1],
210	self.result_series_keys[0]: self.time.tolist(),
211	self.result_series_keys[1]: self.msd_array.tolist(),
212	self.result_series_keys[2]: (np.array(gradients) / 6).tolist(),
213	self.result_series_keys[3]: (np.array(gradient_errors) / 6).tolist(),
214	}
215	return data	1✔
216
217	def run_calculator(self):	1✔
218	"""Run analysis."""
219	self._run_dependency_check()	1✔
220	for species in self.args.species:	1✔
221	# Here for now to avoid issues. Should be moved out when calculators become
222	# species-wise
223	self.time = None	1✔
224	self.time = self._handle_tau_values()	1✔
225	dict_ref = str.encode("/".join([species, self.loaded_property.name]))	1✔
226	batch_ds = self.get_batch_dataset([species])	1✔
227	self.msd_array = np.zeros(self.data_resolution)	1✔
228	self.count = 0	1✔
229	# loop over batches to get MSD
230	for i, batch in tqdm(	1✔
231	enumerate(batch_ds),
232	ncols=70,
233	desc=species,
234	total=self.n_batches,
235	disable=self.memory_manager.minibatch,
236	):
237	ensemble_ds = self.get_ensemble_dataset(batch, species)	1✔
238
239	for ensemble in ensemble_ds:	1✔
240	if not ensemble[dict_ref].shape[1] == self.args.data_range:	1!
UNCOV 241	continue	×
242	else:
243	self.msd_array += self.ensemble_operation(ensemble[dict_ref])	1✔
244	self.count += 1	1✔
245
246	# self.msd_array = np.array(tf.reduce_sum(self.msd_array, axis=0))
247	fit_results = self.fit_diff_coeff()	1✔
248	self.queue_data(data=fit_results, subjects=[species])	1✔
249
250	def plot_data(self, data):	1✔
251	"""
252	Plot the Einstein fits.
253
254	Parameters
255	----------
256	data
257
258	Returns
259	-------
260
261	"""
262	for selected_species, val in data.items():	1✔
263	fig = figure(x_axis_label=self.x_label, y_axis_label=self.y_label)	1✔
264
265	gradients = np.array(val[self.result_series_keys[2]])	1✔
266	gradient_errors = np.array(val[self.result_series_keys[3]])	1✔
267
268	time = np.array(val[self.result_series_keys[0]])	1✔
269	msd = np.array(val[self.result_series_keys[1]])	1✔
270
271	fig.y_range = Range1d(-0.0, 1.1 * max(msd))	1✔
272
273	# Compute the span
274	span = Span(	1✔
275	location=time[self.args.fit_range],
276	dimension="height",
277	line_dash="dashed",
278	)
279	# Compute msd and fit lines
280	fig.line(	1✔
281	time,
282	msd,
283	color=utils.Colour.ORANGE,
284	legend_label=(
285	f"{selected_species}: {val[self.result_keys[0]]: 0.3E} +-"
286	f" {val[self.result_keys[1]]: 0.3E}"
287	),
288	)
289	fit_data = val[self.result_keys[2]] * time + val[self.result_keys[3]]	1✔
290	fig.line(time, fit_data, color=utils.Colour.PAUA, legend_label="Curve fit.")	1✔
291	fig.extra_y_ranges = {	1✔
292	"Diff_range": Range1d(
293	start=0.999 * min(gradients), end=1.001 * max(gradients)
294	)
295	}
296
297	fig.add_layout(	1✔
298	LinearAxis(
299	y_range_name="Diff_range",
300	axis_label=r"$$\text{Diffusion Coefficient} / m^{2}s^{-1}$$",
301	),
302	"right",
303	)
304	grad_time = time[-len(gradients) :]	1✔
305	fig.line(	1✔
306	grad_time,
307	gradients,
308	y_range_name="Diff_range",
309	color=utils.Colour.MULBERRY,
310	)
311	fig.varea(	1✔
312	grad_time,
313	gradients - gradient_errors,
314	gradients + gradient_errors,
315	alpha=0.3,
316	color=utils.Colour.ORANGE,
317	y_range_name="Diff_range",
318	)
319
320	fig.add_tools(HoverTool())	1✔
321	fig.add_layout(span)	1✔
322	self.plot_array.append(fig)	1✔

zincware / MDSuite / 8594096068

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous