amd.utils.random_cell() - Code Metrics - Inspection of "1.3.3a2 cleanup" - dwiddo/average-minimum-distance - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Push — master ( 9779af...19222a )

by Daniel

created 2022-11-04 01:02 UTC

amd.utils.random_cell() A

↳ Parent: amd.utils

Complexity

Conditions

Size

Total Lines	27
Code Lines	18

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	18
dl	0
loc	27
rs	9.0333
c	0
b	0
f	0
cc	5
nop	3

"""General utility functions."""

from typing import Tuple
import sys
import time
import datetime

import numpy as np
import numba
from scipy.spatial.distance import squareform


def diameter(cell):
    """Diameter of a unit cell (as a square matrix in Cartesian/Orthogonal form)
    in 3 or fewer dimensions."""

    dims = cell.shape[0]
    if dims == 1:
        return cell[0][0]
    if dims == 2:
        d = np.amax(np.linalg.norm(np.array([cell[0] + cell[1], cell[0] - cell[1]]), axis=-1))
    elif dims == 3:
        diams = np.array([
            cell[0] + cell[1] + cell[2],
            cell[0] + cell[1] - cell[2],
            cell[0] - cell[1] + cell[2],
            - cell[0] + cell[1] + cell[2]
        ])
        d = np.amax(np.linalg.norm(diams, axis=-1))
    else:
        msg = f'diameter() not implemented for dims > 3 (passed cell shape {cell.shape}).'
        raise NotImplementedError(msg)
    return d


@numba.njit()
def cellpar_to_cell(a, b, c, alpha, beta, gamma):

    """Simplified version of function from :mod:`ase.geometry` of the same name.
    3D unit cell parameters a,b,c,α,β,γ --> cell as 3x3 NumPy array.
    """

    eps = 2 * np.spacing(90.0)  # ~1.4e-14

    cos_alpha = 0. if abs(abs(alpha) - 90.) < eps else np.cos(alpha * np.pi / 180.)
    cos_beta = 0. if abs(abs(beta) - 90.) < eps else np.cos(beta * np.pi / 180.)
    cos_gamma = 0. if abs(abs(gamma) - 90.) < eps else np.cos(gamma * np.pi / 180.)

    if abs(gamma - 90) < eps:
        sin_gamma = 1.
    elif abs(gamma + 90) < eps:
        sin_gamma = -1.
    else:
        sin_gamma = np.sin(gamma * np.pi / 180.)

    cy = (cos_alpha - cos_beta * cos_gamma) / sin_gamma
    cz_sqr = 1. - cos_beta ** 2 - cy ** 2
    if cz_sqr < 0:
        raise RuntimeError('Could not create unit cell from given parameters.')

    cell = np.zeros((3, 3))
    cell[0, 0] = a
    cell[1, 0] = b * cos_gamma
    cell[1, 1] = b * sin_gamma
    cell[2, 0] = c * cos_beta
    cell[2, 1] = c * cy
    cell[2, 2] = c * np.sqrt(cz_sqr)

    return cell


@numba.njit()
def cellpar_to_cell_2D(a, b, alpha):
    """2D unit cell parameters a,b,α --> cell as 2x2 ndarray."""

    cell = np.zeros((2, 2))
    cell[0, 0] = a
    cell[1, 0] = b * np.cos(alpha * np.pi / 180.)
    cell[1, 1] = b * np.sin(alpha * np.pi / 180.)

    return cell


def cell_to_cellpar(cell):
    """Unit cell as a 3x3 NumPy array -> list of 6 lengths + angles."""
    lengths = np.linalg.norm(cell, axis=-1)
    angles = []
    for i, j in [(1, 2), (0, 2), (0, 1)]:
        ang_rad = np.arccos(np.dot(cell[i], cell[j]) / (lengths[i] * lengths[j]))
        angles.append(np.rad2deg(ang_rad))
    return np.concatenate((lengths, np.array(angles)))


def cell_to_cellpar_2D(cell):
    """Unit cell as a 2x2 NumPy array -> list of 2 lengths and an angle."""
    cellpar = np.zeros((3, ))
    lengths = np.linalg.norm(cell, axis=-1)
    ang_rad = np.arccos(np.dot(cell[0], cell[1]) / (lengths[0] * lengths[1]))
    cellpar[0] = lengths[0]
    cellpar[1] = lengths[1]
    cellpar[2] = np.rad2deg(ang_rad)
    return cellpar


def neighbours_from_distance_matrix(
        n: int,
        dm: np.ndarray
) -> Tuple[np.ndarray, np.ndarray]:
    """Given a distance matrix, find the n nearest neighbours of each item.

    Parameters
    ----------
    n : int
        Number of nearest neighbours to find for each item.
    dm : :class:`numpy.ndarray`
        2D distance matrix or 1D condensed distance matrix.

    Returns
    -------
    (nn_dm, inds) : Tuple[:class:`numpy.ndarray`, :class:`numpy.ndarray`]
        ``nn_dm[i][j]`` is the distance from item :math:`i` to its :math:`j+1` st
        nearest neighbour, and ``inds[i][j]`` is the index of this neighbour
        (:math:`j+1` since index 0 is the first nearest neighbour).
    """

    inds = None

    # 2D distance matrix
    if len(dm.shape) == 2:
        inds = np.array([np.argpartition(row, n)[:n] for row in dm])

    # 1D condensed distance vector
    elif len(dm.shape) == 1:
        dm = squareform(dm)
        inds = []
        for i, row in enumerate(dm):
            inds_row = np.argpartition(row, n+1)[:n+1]
            inds_row = inds_row[inds_row != i][:n]
            inds.append(inds_row)
        inds = np.array(inds)

    else:
        ValueError(
            'Input must be an ndarray, either a 2D distance matrix '
            'or a condensed distance matrix (returned by pdist).')

    # inds are the indexes of nns: inds[i,j] is the j-th nn to point i
    nn_dm = np.take_along_axis(dm, inds, axis=-1)
    sorted_inds = np.argsort(nn_dm, axis=-1)
    inds = np.take_along_axis(inds, sorted_inds, axis=-1)
    nn_dm = np.take_along_axis(nn_dm, sorted_inds, axis=-1)
    return nn_dm, inds


def random_cell(length_bounds=(1, 2), angle_bounds=(60, 120), dims=3):
    """Dimensions 2 and 3 only. Random unit cell with uniformally chosen length and
    angle parameters between bounds."""

    ll, lu = length_bounds
    al, au = angle_bounds

    if dims == 3:
        while True:
            lengths = [np.random.uniform(low=ll, high=lu) for _ in range(dims)]
            angles = [np.random.uniform(low=al, high=au) for _ in range(dims)]

            try:
                cell = cellpar_to_cell(*lengths, *angles)
                break
            except RuntimeError:
                continue

    elif dims == 2:
        lengths = [np.random.uniform(low=ll, high=lu) for _ in range(dims)]
        alpha = np.random.uniform(low=al, high=au)
        cell = cellpar_to_cell_2D(*lengths, alpha)

    else:
        raise ValueError(f'random_cell only implimented for dimensions 2 and 3 (passed {dims})')

    return cell


class _ETA:

    """Pass the number to items to process, then call .update() on every loop.
    """

    # epochtime_{n+1} = factor * epochtime + (1-factor) * epochtime_{n}
    _moving_av_factor = 0.3

    def __init__(self, to_do, update_rate=1000):
        self.to_do = to_do
        self.update_rate = update_rate
        self.counter = 0
        self.this_epoch = 0
        self.start_time = time.perf_counter()
        self.time_per_epoch = None
        self.done = False
        self.tic = self.start_time

    def update(self):
        """Call on each loop."""

        self.counter += 1

        if self.counter >= self.to_do:

            if not self.done:
                self.finish()
            return

        elif not self.counter % self.update_rate:
            self.end_epoch()

    def finish(self):
        """Called when done."""
        total = time.perf_counter() - self.start_time
        self.done = True
        msg = f'Time ({self.counter} items): {round(total, 2)}s, ' \
              f'{round(self.to_do/total, 2)} items/s\r\n'
        sys.stdout.write(msg)

    def end_epoch(self):
        """Called at the end of every epoch, printing the ETA to stdout."""
        toc = time.perf_counter()
        epoch_time = toc - self.tic
        if self.time_per_epoch is None:
            self.time_per_epoch = epoch_time
        else:
            self.time_per_epoch = _ETA._moving_av_factor * epoch_time + \
                                  (1 - _ETA._moving_av_factor) * self.time_per_epoch
        percent = round(100 * self.counter / self.to_do, 2)
        remaining = int(((self.to_do - self.counter) / self.update_rate) * self.time_per_epoch)
        eta = str(datetime.timedelta(seconds=remaining))
        self.tic = time.perf_counter()
        sys.stdout.write(f'{percent}%, ETA {eta}' + ' ' * 20 + '\r')


1			"""General utility functions."""
2
3			from typing import Tuple
4			import sys
5			import time
6			import datetime
7
8			import numpy as np
9			import numba
10			from scipy.spatial.distance import squareform
11
12
13			def diameter(cell):
14			"""Diameter of a unit cell (as a square matrix in Cartesian/Orthogonal form)
15			in 3 or fewer dimensions."""
16
17			dims = cell.shape[0]
18			if dims == 1:
19			return cell[0][0]
20			if dims == 2:
21			d = np.amax(np.linalg.norm(np.array([cell[0] + cell[1], cell[0] - cell[1]]), axis=-1))
22			elif dims == 3:
23			diams = np.array([
24			cell[0] + cell[1] + cell[2],
25			cell[0] + cell[1] - cell[2],
26			cell[0] - cell[1] + cell[2],
27			- cell[0] + cell[1] + cell[2]
28			])
29			d = np.amax(np.linalg.norm(diams, axis=-1))
30			else:
31			msg = f'diameter() not implemented for dims > 3 (passed cell shape {cell.shape}).'
32			raise NotImplementedError(msg)
33			return d
34
35
36			@numba.njit()
37			def cellpar_to_cell(a, b, c, alpha, beta, gamma):
			0 ignored issues – show best-practice introduced 2022-04-14 08:53 UTC by Report Bug Copy Issue Report Too many arguments (6/5) Loading history...
38			"""Simplified version of function from :mod:`ase.geometry` of the same name.
39			3D unit cell parameters a,b,c,α,β,γ --> cell as 3x3 NumPy array.
40			"""
41
42			eps = 2 * np.spacing(90.0) # ~1.4e-14
43
44			cos_alpha = 0. if abs(abs(alpha) - 90.) < eps else np.cos(alpha * np.pi / 180.)
45			cos_beta = 0. if abs(abs(beta) - 90.) < eps else np.cos(beta * np.pi / 180.)
46			cos_gamma = 0. if abs(abs(gamma) - 90.) < eps else np.cos(gamma * np.pi / 180.)
47
48			if abs(gamma - 90) < eps:
49			sin_gamma = 1.
50			elif abs(gamma + 90) < eps:
51			sin_gamma = -1.
52			else:
53			sin_gamma = np.sin(gamma * np.pi / 180.)
54
55			cy = (cos_alpha - cos_beta * cos_gamma) / sin_gamma
56			cz_sqr = 1. - cos_beta 2 - cy 2
57			if cz_sqr < 0:
58			raise RuntimeError('Could not create unit cell from given parameters.')
59
60			cell = np.zeros((3, 3))
61			cell[0, 0] = a
62			cell[1, 0] = b * cos_gamma
63			cell[1, 1] = b * sin_gamma
64			cell[2, 0] = c * cos_beta
65			cell[2, 1] = c * cy
66			cell[2, 2] = c * np.sqrt(cz_sqr)
67
68			return cell
69
70
71			@numba.njit()
72			def cellpar_to_cell_2D(a, b, alpha):
73			"""2D unit cell parameters a,b,α --> cell as 2x2 ndarray."""
74
75			cell = np.zeros((2, 2))
76			cell[0, 0] = a
77			cell[1, 0] = b * np.cos(alpha * np.pi / 180.)
78			cell[1, 1] = b * np.sin(alpha * np.pi / 180.)
79
80			return cell
81
82
83			def cell_to_cellpar(cell):
84			"""Unit cell as a 3x3 NumPy array -> list of 6 lengths + angles."""
85			lengths = np.linalg.norm(cell, axis=-1)
86			angles = []
87			for i, j in [(1, 2), (0, 2), (0, 1)]:
88			ang_rad = np.arccos(np.dot(cell[i], cell[j]) / (lengths[i] * lengths[j]))
89			angles.append(np.rad2deg(ang_rad))
90			return np.concatenate((lengths, np.array(angles)))
91
92
93			def cell_to_cellpar_2D(cell):
94			"""Unit cell as a 2x2 NumPy array -> list of 2 lengths and an angle."""
95			cellpar = np.zeros((3, ))
96			lengths = np.linalg.norm(cell, axis=-1)
97			ang_rad = np.arccos(np.dot(cell[0], cell[1]) / (lengths[0] * lengths[1]))
98			cellpar[0] = lengths[0]
99			cellpar[1] = lengths[1]
100			cellpar[2] = np.rad2deg(ang_rad)
101			return cellpar
102
103
104			def neighbours_from_distance_matrix(
105			n: int,
106			dm: np.ndarray
107			) -> Tuple[np.ndarray, np.ndarray]:
108			"""Given a distance matrix, find the n nearest neighbours of each item.
109
110			Parameters
111			----------
112			n : int
113			Number of nearest neighbours to find for each item.
114			dm : :class:`numpy.ndarray`
115			2D distance matrix or 1D condensed distance matrix.
116
117			Returns
118			-------
119			(nn_dm, inds) : Tuple[:class:`numpy.ndarray`, :class:`numpy.ndarray`]
120			``nn_dm[i][j]`` is the distance from item :math:`i` to its :math:`j+1` st
121			nearest neighbour, and ``inds[i][j]`` is the index of this neighbour
122			(:math:`j+1` since index 0 is the first nearest neighbour).
123			"""
124
125			inds = None
126
127			# 2D distance matrix
128			if len(dm.shape) == 2:
129			inds = np.array([np.argpartition(row, n)[:n] for row in dm])
130
131			# 1D condensed distance vector
132			elif len(dm.shape) == 1:
133			dm = squareform(dm)
134			inds = []
135			for i, row in enumerate(dm):
136			inds_row = np.argpartition(row, n+1)[:n+1]
137			inds_row = inds_row[inds_row != i][:n]
138			inds.append(inds_row)
139			inds = np.array(inds)
140
141			else:
142			ValueError(
143			'Input must be an ndarray, either a 2D distance matrix '
144			'or a condensed distance matrix (returned by pdist).')
145
146			# inds are the indexes of nns: inds[i,j] is the j-th nn to point i
147			nn_dm = np.take_along_axis(dm, inds, axis=-1)
148			sorted_inds = np.argsort(nn_dm, axis=-1)
149			inds = np.take_along_axis(inds, sorted_inds, axis=-1)
150			nn_dm = np.take_along_axis(nn_dm, sorted_inds, axis=-1)
151			return nn_dm, inds
152
153
154			def random_cell(length_bounds=(1, 2), angle_bounds=(60, 120), dims=3):
155			"""Dimensions 2 and 3 only. Random unit cell with uniformally chosen length and
156			angle parameters between bounds."""
157
158			ll, lu = length_bounds
159			al, au = angle_bounds
160
161			if dims == 3:
162			while True:
163			lengths = [np.random.uniform(low=ll, high=lu) for _ in range(dims)]
164			angles = [np.random.uniform(low=al, high=au) for _ in range(dims)]
165
166			try:
167			cell = cellpar_to_cell(lengths, angles)
168			break
169			except RuntimeError:
170			continue
171
172			elif dims == 2:
173			lengths = [np.random.uniform(low=ll, high=lu) for _ in range(dims)]
174			alpha = np.random.uniform(low=al, high=au)
175			cell = cellpar_to_cell_2D(*lengths, alpha)
176
177			else:
178			raise ValueError(f'random_cell only implimented for dimensions 2 and 3 (passed {dims})')
179
180			return cell
181
182
183			class _ETA:
			0 ignored issues – show best-practice introduced 2022-11-04 01:05 UTC by Report Bug Copy Issue Report Too many instance attributes (8/7) Loading history...
184			"""Pass the number to items to process, then call .update() on every loop.
185			"""
186
187			# epochtime_{n+1} = factor * epochtime + (1-factor) * epochtime_{n}
188			_moving_av_factor = 0.3
189
190			def __init__(self, to_do, update_rate=1000):
191			self.to_do = to_do
192			self.update_rate = update_rate
193			self.counter = 0
194			self.this_epoch = 0
195			self.start_time = time.perf_counter()
196			self.time_per_epoch = None
197			self.done = False
198			self.tic = self.start_time
199
200			def update(self):
201			"""Call on each loop."""
202
203			self.counter += 1
204
205			if self.counter >= self.to_do:
			0 ignored issues – show unused-code introduced 2022-04-14 08:53 UTC by Report Bug Copy Issue Report Unnecessary "elif" after "return" Loading history...
206			if not self.done:
207			self.finish()
208			return
209
210			elif not self.counter % self.update_rate:
211			self.end_epoch()
212
213			def finish(self):
214			"""Called when done."""
215			total = time.perf_counter() - self.start_time
216			self.done = True
217			msg = f'Time ({self.counter} items): {round(total, 2)}s, ' \
218			f'{round(self.to_do/total, 2)} items/s\r\n'
219			sys.stdout.write(msg)
220
221			def end_epoch(self):
222			"""Called at the end of every epoch, printing the ETA to stdout."""
223			toc = time.perf_counter()
224			epoch_time = toc - self.tic
225			if self.time_per_epoch is None:
226			self.time_per_epoch = epoch_time
227			else:
228			self.time_per_epoch = _ETA._moving_av_factor * epoch_time + \
229			(1 - _ETA._moving_av_factor) * self.time_per_epoch
230			percent = round(100 * self.counter / self.to_do, 2)
231			remaining = int(((self.to_do - self.counter) / self.update_rate) * self.time_per_epoch)
232			eta = str(datetime.timedelta(seconds=remaining))
233			self.tic = time.perf_counter()
234			sys.stdout.write(f'{percent}%, ETA {eta}' + ' ' * 20 + '\r')
235

dwiddo / average-minimum-distance

Push — master ( 9779af...19222a )

amd.utils.random_cell() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like