# -*- coding: utf-8 -*-
"""
qspec.tools
===========
Created on 05.05.2020
@author: Patrick Mueller
Module including mathematical and technical methods.
"""
import os
import sys
import ctypes
from pathlib import Path
import ctypes.wintypes
import shutil
import sqlite3
import numpy as np
from qspec._types import *
__all__ = ['ROMAN_NUMERALS', 'COLORS', 'get_rgb_print_command', 'print_colored', 'printh', 'printw', 'printf',
'map_corr_coeff_to_color', 'print_cov', 'get_default_path', 'create_doc_link', 'create_data_dir',
'get_config_dict', 'fraction', 'check_half_integer', 'half_integer_to_fraction', 'half_integer_to_str',
'get_val_with_unc', 'roman_to_int', 'odd', 'even', 'get_decimals', 'floor_log2', 'floor_log10', 'round_to_n',
'factorial', 'asarray_optional', 'in_nested', 'check_iterable', 'make_str_iterable_unique',
'check_dimension', 'check_shape_like', 'check_shape', 'nan_helper', 'dict_to_list', 'list_to_dict',
'list_to_excel', 'add_nested_key', 'merge_dicts', 'convolve_dict', 'combine_dicts', 'merge_intervals',
'absolute', 'absolute_complex', 'angle', 'angle_d', 'transform', 'unit_vector', 'e_r', 'e_theta', 'e_phi',
'orthonormal_rtp', 'orthonormal', 'rotation_matrix', 'Rotation', 'rotation_to_vector',
'import_iso_shifts_tilda']
ROMAN_NUMERALS = {'M': 1000, 'D': 500, 'C': 100, 'L': 50, 'X': 10, 'V': 5, 'I': 1}
[docs]class COLORS:
HEADER = '\033[95m'
OKBLUE = '\033[94m'
OKCYAN = '\033[96m'
OKGREEN = '\033[92m'
WARNING = '\033[93m'
FAIL = '\033[91m'
ENDC = '\033[0m'
BOLD = '\033[1m'
UNDERLINE = '\033[4m'
PYPLOT = ['C{}'.format(i) for i in range(10)]
[docs]def get_rgb_print_command(r, g, b):
"""
:param r: The fraction of red (0-255).
:param g: The fraction of green (0-255).
:param b: The fraction of blue (0-255).
:returns: The command str to print rgb colors in the console.
"""
return '\033[38;2;{};{};{}m'.format(r, g, b)
[docs]def print_colored(specifier, *values, returned=False, **kwargs):
"""
Print with the specified color.
:param specifier: str of the color name defined in the COLORS class or an RGB-value (0-255, 0-255, 0-255).
:param values: The values to print.
:param returned: Return the str instead of printing it.
:param kwargs: See print().
:returns: None.
"""
if isinstance(specifier, str):
_c = eval('COLORS.{}'.format(specifier.upper()))
else:
_c = get_rgb_print_command(*specifier)
_values = (_c, ) + values + (COLORS.ENDC, )
if returned:
return '{}{}{}'.format(*_values)
print('{}{}{}'.format(*_values), **kwargs)
[docs]def printh(*values, **kwargs):
"""
Print with the HEADER color (pink).
:param values: The values to print.
:param kwargs: See print().
:returns: None.
"""
_values = (COLORS.HEADER, ) + values + (COLORS.ENDC, )
print('{}{}{}'.format(*_values), **kwargs)
[docs]def printw(*values, **kwargs):
"""
Print with the WARNING color (yellow).
:param values: The values to print.
:param kwargs: See print().
:returns: None.
"""
_values = (COLORS.WARNING, ) + values + (COLORS.ENDC, )
print('{}{}{}'.format(*_values), **kwargs)
[docs]def printf(*values, **kwargs):
"""
Print with the FAIL color (red).
:param values: The values to print.
:param kwargs: See print().
:returns: None.
"""
_values = (COLORS.FAIL, ) + values + (COLORS.ENDC, )
print('{}{}{}'.format(*_values), **kwargs)
[docs]def map_corr_coeff_to_color(val, clip=True):
"""
:param val: A value between -1 and 1.
:param clip: Whether to clip vals outside -1 and 1. If False, raise ValueError.
:returns: The RGB values in the range 0-255.
"""
if clip:
if np.isnan(val):
val = 0.
elif val < -1:
val = -1
elif val > 1:
val = 1
if np.isnan(val) or val < -1 or val > 1:
raise ValueError('The correlation coefficient must be in [-1, 1].')
g = int(round(val * 127 + 127, 0))
return 255 - g, g, 0
[docs]def print_cov(cov, normalize=False, decimals=2):
"""
Print a covariance "as is" or as color-coded Pearson correlation coefficients.
:param cov: A covariance matrix.
:param normalize: Whether to normalize the covariance to be the Pearson correlation coefficient.
:param decimals: The number of decimal places to be printed.
:returns: None.
"""
cov = np.array(cov, dtype=float)
if normalize:
norm = np.sqrt(np.diag(cov)[:, None] * np.diag(cov))
nonzero = norm != 0
cov[nonzero] /= norm[nonzero]
cov = np.around(cov + 0., decimals=decimals)
digits = int(np.floor(np.log10(np.abs(cov.shape[0])))) + 1
for i, row in enumerate(cov):
print('{}: {}'.format(str(i).zfill(digits), ' '.join(
'{}{}{}'.format(get_rgb_print_command(*map_corr_coeff_to_color(val)),
'{:1.2f}'.format(val).rjust(decimals + 3), COLORS.ENDC) for val in row)))
""" System operations """
[docs]def get_default_path():
"""
:returns: The absolute path to the "PyCLS user folder."
"""
csidl_personal = 5 # My Documents
shgfp_type_current = 0 # Get current, not default value
buf = ctypes.create_unicode_buffer(ctypes.wintypes.MAX_PATH)
ctypes.windll.shell32.SHGetFolderPathW(None, csidl_personal, None, shgfp_type_current, buf)
return os.path.join(buf.value, 'PyCLS')
[docs]def create_doc_link(env_path: str = None):
"""
:param env_path: The path to the virtual environment where the 'Lib' folder resides in.
If None, the path is inherited from the currently running Python-executable.
:returns: None. Creates an html document redirecting to the API-Documentation inside the PyCLS installation.
"""
if env_path is None:
env_path = os.path.dirname(sys.executable)
if os.path.split(env_path)[1].lower() == 'scripts':
env_path = os.path.split(env_path)[0]
path = get_default_path()
html_link = os.path.join(env_path, 'Lib\\site-packages\\PyCLS\\docs\\build\\html_link.txt')
with open(html_link, 'r') as file:
s = file.read()
s = s.replace('ENV-PATH\\Lib', os.path.join(env_path, 'Lib'))
with open(os.path.join(path, 'API-Documentation.html'), 'w') as file:
file.write(s)
[docs]def create_data_dir(path: str = None, overwrite: bool = False):
"""
:param path: The path were to create the user folder "PyCLS" (do not include "PyCLS" in 'path').
If None, the user folder is created in the "Documents" Windows library.
:param overwrite: Whether to overwrite the current "PyCLS" user folder if it exists.
:returns: None. Creates the user folder "PyCLS" at the specified 'path'
or in the "Documents" Windows library if no 'path' is specified.
"""
path = get_default_path() if path is None else os.path.join(path, 'PyCLS')
if overwrite or not Path(os.path.join(path, 'PyCLS')).exists():
Path(path).mkdir(parents=True, exist_ok=True)
Path(os.path.join(path, 'databases')).mkdir(parents=True, exist_ok=True)
env_path = os.path.dirname(sys.executable)
if os.path.split(env_path)[1].lower() == 'scripts':
env_path = os.path.split(env_path)[0]
with open(os.path.join(path, 'config.cfg'), 'w') as config:
config.write('# PyCLS Config file\n\n')
config.write('Environment path = {}\n'.format(env_path))
config.write('Data path = {}\n'.format(path))
config.write('ffmpeg = {}\n'.format('C:\\Program Files\\ffmpeg\\bin\\ffmpeg.exe'))
src_dir = os.path.join(env_path, 'Lib\\site-packages\\PyCLS\\databases')
dst_dir = os.path.join(path, 'databases')
for file in os.listdir(src_dir):
shutil.copyfile(os.path.join(src_dir, file), os.path.join(dst_dir, file))
create_doc_link(env_path)
else:
raise ValueError('The user folder "...\\Documents\\PyCLS" already exists and \'overwrite\' is set to False. '
'If you would like to replace your user folder with the default user folder, '
'set \'overwrite=True\'.')
[docs]def get_config_dict() -> dict:
"""
:returns: A dictionary containing the content of the currently used config file.
"""
path = get_default_path()
ret = {}
with open(os.path.join(path, 'config.cfg'), 'r') as config:
for s in config:
if '=' in s:
ret[s[:s.find('=')].strip()] = s[(s.find('=')+1):].strip()
return ret
""" literal -> numeral operations """
[docs]def fraction(r: Union[Rational, str]) -> (int, int):
"""
:param r: A sympy.Rational or a str with the signature "'num'/'denom'".
:returns: the numerator and denominator of 'r'.
"""
if not isinstance(r, Rational) and not isinstance(r, str):
raise TypeError('Argument must be a sympy.Rational or str, but is {}: {}.'.format(type(r), r))
string = str(r)
slash = string.find('/')
if slash == -1:
num = int(string)
denom = 1
else:
num = int(string[:slash])
denom = int(string[(slash+1):])
return num, denom
[docs]def check_half_integer(*args: scalar):
"""
:param args: Scalar arguments.
:returns: None. Checks whether the given arguments are multiples of 1/2.
:raises ValueError: If any argument is not a multiple of 1/2.
"""
if sum([abs(arg % 0.5) for arg in args]) > 0:
raise ValueError('The numbers must be multiples of 1/2.')
[docs]def half_integer_to_fraction(val: scalar) -> (int, int):
"""
:param val: A scalar value.
:returns: The numerator and denominator of the half-integer.
"""
check_half_integer(val)
if abs(val % 1) > 0:
return int(2 * val), 2
return int(val), 1
[docs]def half_integer_to_str(val: scalar, symbol: str = '_'):
"""
:param val: A scalar value.
:param symbol: The symbol to use for the fraction.
:returns: A rational str-representation of a half-integer.
"""
num, denom = half_integer_to_fraction(val)
if denom == 1:
return str(num)
return '{}{}2'.format(num, symbol)
[docs]def get_val_with_unc(val_string: str) -> (float, float):
"""
:param val_string: The str representation of a number with uncertainty of the format '1.234(56)'.
Decimal separators can be used or spared arbitrarily.
:returns: The value and its uncertainty.
"""
err, decimals = 0., 0
if val_string == '':
return err, err
lbrace = val_string.find('(')
rbrace = val_string.find(')')
val = float(val_string) if lbrace == -1 else float(val_string[:lbrace])
err_string = val_string[(lbrace + 1):rbrace]
if -1 not in [lbrace, rbrace, len(err_string) - 1]:
err = float(err_string)
if '.' in val_string[:lbrace]:
dot = val_string.find('.')
decimals = len(val_string[(dot+1):lbrace])
if '.' not in val_string[lbrace:rbrace]:
err *= eval('1e-{}'.format(decimals))
else:
dot = val_string[lbrace:].find('.')
decimals = max([decimals, len(val_string[(dot + 1):rbrace])])
return round(val, decimals), round(err, decimals)
[docs]def roman_to_int(roman: str) -> int:
"""
Convert from Roman numerals to an integer
[jonrsharpe, https://codereview.stackexchange.com/questions/68297/convert-roman-to-int].
:param roman: The str representation of a roman number.
"""
numbers = []
for char in roman.upper():
numbers.append(ROMAN_NUMERALS[char])
total = 0
num2 = numbers[-1]
for num1, num2 in zip(numbers, numbers[1:]):
if num1 >= num2:
total += num1
else:
total -= num1
return total + num2
""" numeral -> numeral operations """
[docs]def odd(x: float) -> int:
"""
:param x: A float value.
:returns: The closest odd integer value.
"""
return int(x) if int(x) % 2 else int(x) + 1
[docs]def even(x: float) -> int:
"""
:param x: A float value.
:returns: The closest even integer value.
"""
return int(x) if not int(x) % 2 else int(x) + 1
[docs]def get_decimals(x: float) -> int:
"""
:param x: A float value.
:return: The number of displayed decimals of the float value 'x'.
"""
s = str(x)
dot = s.find('.')
return len(s[(dot+1):])
[docs]def floor_log2(x: array_like) -> ndarray:
"""
:param x: Scalar values.
:returns: The closest integer values below the logarithm with basis 10 of the absolute value of 'x'.
"""
x = np.asarray(x)
if len(x.shape) > 0:
nonzero = x.nonzero()
n = np.zeros_like(x)
n[nonzero] = np.floor(np.log2(np.abs(x[nonzero]))).astype(int)
return n
if x == 0:
return np.array(0)
return np.floor(np.log2(np.abs(x))).astype(int)
[docs]def floor_log10(x: array_like) -> ndarray:
"""
:param x: Scalar values.
:returns: The closest integer values below the logarithm with basis 10 of the absolute value of 'x'.
"""
x = np.asarray(x)
if len(x.shape) > 0:
nonzero = x.nonzero()
n = np.zeros_like(x)
n[nonzero] = np.floor(np.log10(np.abs(x[nonzero]))).astype(int)
return n
if x == 0:
return np.array(0)
return np.floor(np.log10(np.abs(x))).astype(int)
[docs]def round_to_n(x: array_like, n: int) -> (scalar, int):
"""
:param x: The input data.
:param n: The number of significant decimal places to round to.
:returns: x rounded to 'n' significant decimal places and
the corresponding number of decimal places after the decimal point.
"""
if isinstance(x, ndarray):
x = x.squeeze().tolist()
if isinstance(x, list):
raise TypeError('Only size-1 arrays are supported.')
if x == 0.:
return type(x)(0), 0
else:
decimals = -int(np.floor(np.log10(np.abs(x)))) + n - 1
return type(x)(np.around(x, decimals=decimals)), decimals
[docs]def factorial(n: array_like):
"""
:param n: The integer number.
:returns: n! (array compatible).
"""
n = np.asarray(n, dtype=int)
mask = n > 1
ret = np.ones_like(n)
if np.any(mask):
ret[mask] = n[mask] * factorial(n[mask] - 1)
return ret
""" Iterable operations """
[docs]def asarray_optional(a: Optional[array_like], **kwargs):
"""
:param a: Input data, see numpy docs.
:param kwargs: The keyword arguments are passed to numpy.asarray.
:returns: None if 'a' is None else 'numpy.asarray(a, \*\*kwargs)'.
"""
return None if a is None else np.asarray(a, **kwargs)
[docs]def in_nested(a, nested: Iterable) -> bool:
"""
:param a: The element to look for.
:param nested: The nested list.
:returns: Whether a is inside the 'nested' list.
"""
if not isinstance(nested, Iterable):
return False
if a in nested:
return True
ret = False
for el in nested:
ret += in_nested(a, el)
if ret:
return True
return ret
[docs]def check_iterable(arg: Any, dtype: type = str) -> list:
"""
:param arg: The argument that is checked.
:param dtype: The type that has to be matched by the elements of arg or arg itself.
:returns: 'arg' as a list if it is a 'list', 'tuple' or 'set' of values of type 'dtype',
a list of the single element 'arg' if it is of type 'dtype' and else an empty list.
"""
if isinstance(arg, list) or isinstance(arg, tuple) or isinstance(arg, set):
if all(isinstance(s, dtype) for s in arg):
return list(arg)
return []
elif isinstance(arg, dtype):
return [arg, ]
else:
return []
[docs]def make_str_iterable_unique(a: Iterable[str], identifier: Iterable[str] = None):
"""
:param a: An Iterable of str values.
:param identifier: An Iterable of str values to append to the values of 'a' if they appear more than once.
If None, '_i' is used as the identifier for the ith appearance of a value in 'a'.
:returns: 'a', but the elements which appear more than once are numerated/attached with the identifiers and bunched.
For example: ['a', 'b', 'c', 'b', 'd'] -> ['a', 'b_0', 'b_1', 'c', 'd'] or
['a', 'b', 'c', 'b', 'd'] -> ['a', 'b' + identifier[0], 'b' + identifier[1], 'c', 'd'].
"""
temp_1 = list(a)
temp_2 = []
if identifier is not None:
identifier = list(identifier)
for p in temp_1:
if p not in temp_2:
temp_2.append(p)
ret = []
for p in temp_2:
m = temp_1.count(p)
for i in range(m):
if m == 1:
ret.append(p)
else:
if identifier is None:
ret.append(p + '_{}'.format(i))
else:
ret.append(p + '_{}'.format(identifier[i % len(identifier)]))
return ret
[docs]def check_dimension(dim: int, axis: int, *args: ndarray) -> None:
"""
:param dim: The number of components the axis 'axis' must have.
:param axis: The axis which must have 'dim' components.
:param args: The arguments which must have 'dim' components along axis 'axis'.
:returns: None
:raises ValueError: All specified arguments must have 'dim' components along axis 'axis'.
"""
err_list = []
for arg in args:
arg = np.asarray(arg)
if arg.shape[axis] != dim:
err_list.append(str(arg.shape[axis]))
if err_list:
err_msg = 'One argument has {} component(s) instead of {} along axis {}.'\
.format(', '.join(err_list), dim, axis)
if len(err_list) > 1:
err_msg = '{} arguments have {} component(s) instead of {} along axis {}.'\
.format(len(err_list), ', '.join(err_list), dim, axis)
raise ValueError(err_msg)
[docs]def check_shape_like(*args: ndarray, allow_scalar=True) -> None:
"""
:param args: The arguments which must have shapes that can be broadcast together.
:param allow_scalar: Whether scalar values (shape=()) are seen as compatible with arbitrary shapes.
:returns: None
:raises ValueError: All specified arguments must have shapes that can be broadcast together.
"""
err_list = []
shape = args[-1].shape
allowed = [len(shape)]
if allow_scalar:
allowed.append(0)
for arg in args[:-1]:
if len(arg.shape) in allowed:
if arg.shape != () and any(i != j and i != 1 and j != 1 for i, j in zip(arg.shape, shape)):
err_list.append(str(arg.shape))
else:
err_list.append(str(arg.shape))
if err_list:
err_msg = 'Operands could not be broadcast together with shapes {} and {}.'.format(', '.join(err_list), shape)
raise ValueError(err_msg)
[docs]def check_shape(shape: tuple, *args: ndarray, allow_scalar=True, return_mode=False) -> Union[bool, None]:
"""
:param shape: The shape which must be matched by the specified arguments.
:param args: The arguments which must match the specified shape.
:param allow_scalar: Whether scalar values (shape=()) are seen as compatible with arbitrary shapes.
:param return_mode: Whether to raise a ValueError or return a boolean if any argument does not have shape 'shape'.
:returns: whether the arguments match the specified shape.
:raises ValueError: All specified arguments must match the specified shape.
"""
err_list = []
allowed = [shape]
if allow_scalar:
allowed.append(())
for arg in args:
if arg.shape not in allowed:
err_list.append(str(arg.shape))
if err_list:
if return_mode:
return False
else:
err_msg = 'Operand(s) with shape(s) {} do not match shape {}.'.format(', '.join(err_list), shape)
raise ValueError(err_msg)
return True
[docs]def nan_helper(y: array_like) -> (ndarray, Callable):
"""
:param y: An array like value.
:returns: The mask where y is NaN and a function which returns the nonzero indices of an array.
"""
return np.isnan(y), lambda z: z.nonzero()[0]
[docs]def dict_to_list(a: dict) -> (list, list):
"""
:param a: A dictionary.
:returns: Two lists with the keys and values of 'a', respectively.
The lists are in the same order, however, they are not sorted.
"""
kv = [[k, v] for k, v in a.items()]
return [k[0] for k in kv], [v[1] for v in kv]
[docs]def list_to_dict(values: array_like, keys: array_like = None) -> dict:
"""
:param values: The values of the dictionary.
:param keys: The keys of the dictionary. If omitted, the keys will be the indices of the values.
:returns: A dictionary with the given 'values' and 'keys'.
"""
v_list = list(np.asarray(values).flatten())
if keys is None:
k_list = list(range(len(values)))
else:
k_list = list(np.asarray(keys).flatten())
return {k: v for k, v in zip(k_list, v_list)}
[docs]def list_to_excel(*args: array_iter, save: str = None, delimiter: str = '\t', header: str = '', align: str = 'top'):
"""
:param args: 1- or 2-dimensional iterables to print or save in an excel-compatible format.
:param save: The filepath, either absolute or relative to the working directory. If None, the output is printed.
:param delimiter: The delimiter between two columns.
:param header: Add a header above the content to the output.
:param align: How to align the columns that are smaller than the largest column.
Supported alignments are {"top", "bottom", else == "top"}.
:returns: None. Prints or saves the 'args' in an excel-compatible way.
"""
columns = []
s = ''
for arg in args:
arg = np.asarray(arg)
if len(arg.shape) not in [1, 2]:
raise ValueError('All arguments must be either 1- or 2-dimensional iterables.')
elif len(arg.shape) == 2:
for c in arg.T:
columns.append(list(c))
s += '{}' + delimiter
else:
columns.append(list(arg))
s += '{}' + delimiter
if header != '':
header += '\n'
c_sizes = [len(c) for c in columns]
c_max = max(c_sizes)
rows = []
if align == 'bottom':
d = -1
for i in range(c_max):
rows.append(s[:-len(delimiter)].format(*['' if i >= n else c[n - i - 1] for c, n in zip(columns, c_sizes)]))
else:
d = 1
for i in range(c_max):
rows.append(s[:-len(delimiter)].format(*['' if i >= n else c[i] for c, n in zip(columns, c_sizes)]))
if save is None:
print(header + '\n'.join(rows[::d]))
return
with open(save, mode='w') as file:
file.write(header)
file.write('\n'.join(rows[::d]))
[docs]def add_nested_key(a: dict, key_list: list, val: array_like) -> dict:
"""
:param a: The target dictionary.
:param key_list: A list of nested keys of a.
:param val: The value that is assigned to the last key in key_list.
:returns: The dictionary a with val assigned to the last key in key_list.
"""
if not isinstance(a, dict):
raise TypeError('Cannot assign another key ({}) since a is not a dictionary.'.format(key_list[0]))
if len(key_list) == 1:
a[key_list[0]] = val
else:
if key_list[0] not in a.keys():
a[key_list[0]] = add_nested_key({}, key_list[1:], val)
else:
a[key_list[0]] = add_nested_key(a[key_list[0]], key_list[1:], val)
return a
[docs]def merge_dicts(a: dict, b: dict, path=None) -> dict:
"""
:param a: The first dictionary.
:param b: The second dictionary.
:param path: _.
:returns: b merged into a. Function copied from andrew cooke
https://stackoverflow.com/questions/7204805/how-to-merge-dictionaries-of-dictionaries.
"""
if path is None:
path = []
for key in b:
if key in a:
if isinstance(a[key], dict) and isinstance(b[key], dict):
a[key] = merge_dicts(a[key], b[key], path + [str(key)])
else:
a[key] = b[key]
else:
a[key] = b[key]
return a
[docs]def convolve_dict(a: dict, *key_lists: Iterable, operator: str = '+', short_keys: bool = True) \
-> Union[dict, ValueError]:
"""
:param a: A dictionary to be convolved.
:param key_lists: An arbitrary number of Iterables of the same length which include keys from 'a'.
:param operator: The operator which is used to combine the key_lists.
:param short_keys: Whether to just use the keys of the first key_list
or representations of the entire operations as the new keys.
:returns: A dictionary of the shape {key_lists[0] + key_lists[1] + ... :
a[key_lists[0]] + a[key_lists[1]] + ... } if operator == '+'.
"""
operators = ['+', '-', '*', '/', '%', 'gauss']
if operator not in operators:
raise ValueError('Operator \'{}\' not available try one of {}.'.format(operator, operators))
if operator.lower() == 'gauss':
return {keys[0] if short_keys else 'Gauss(' + ', '.format(operator).join([str(k) for k in keys]) + ')':
eval('np.sqrt(' + ' + '.join(['a[{}] ** 2'.format(k) for k in keys]) + ')', {'a': a, 'np': np})
for keys in zip(*key_lists)}
return {keys[0] if short_keys else ' {} '.format(operator).join([str(k) for k in keys]):
eval(' {} '.format(operator).join(['a[{}]'.format(k) for k in keys]), {'a': a})
for keys in zip(*key_lists)}
[docs]def combine_dicts(dicts: list, key_lists: Iterable, operator: str = '+', short_keys: bool = True) \
-> Union[dict, ValueError]:
"""
:param dicts: A list of dictionaries to be combined.
:param key_lists: A list of Iterables of the same lengths which include keys from 'a'.
Must have the same length as 'dicts'.
:param operator: The operator which is used to combine the key_lists.
:param short_keys: Whether to just use the keys of the first key_list
or representations of the entire operations as the new keys.
:returns: A dictionary of the shape {key_lists[0] + key_lists[1] + ... :
dicts[0][key_lists[0]] + dicts[1][key_lists[1]] + ... } if operator == '+'.
"""
operators = ['+', '-', '*', '/', '%', 'gauss']
if operator not in operators:
raise ValueError('Operator \'{}\' not available try one of {}.'.format(operator, operators))
if operator.lower() == 'gauss':
return {keys[0] if short_keys else 'Gauss(' + ', '.format(operator).join([str(k) for k in keys]) + ')':
eval('np.sqrt(' + ' + '.join(['dicts[{}][{}] ** 2'.format(i, k)
for i, k in enumerate(keys)]) + ')', {'dicts': dicts, 'np': np})
for keys in zip(*key_lists)}
return {keys[0] if short_keys else ' {} '.format(operator).join([str(k) for k in keys]):
eval(' {} '.format(operator).join(['dicts[{}][{}]'.format(i, k)
for i, k in enumerate(keys)]), {'dicts': dicts})
for keys in zip(*key_lists)}
[docs]def merge_intervals(intervals: array_iter) -> ndarray:
"""
:param intervals: An iterable of intervals.
An interval i is itself an iterable of two scalar values. If i[1] < i[0], the interval is reversed.
:returns: An iterable of non-overlapping intervals.
"""
inter = np.asarray(intervals)
check_shape_like(inter, np.zeros((1, 2)))
inter = np.array([i if i[0] <= i[1] else i[::-1] for i in inter])
sort = np.argsort(inter[:, 0])
inter = inter[sort, :]
new_inter = []
n = 0
m = inter.shape[0]
while n < inter.shape[0] - 1:
end = inter[n, 1]
for m, i in enumerate(inter[(n+1):, :]):
if not end < i[0]:
end = np.max([end, i[1]])
else:
break
new_inter.append([inter[n, 0], end])
n += m + 1
if inter.shape[0] == 1 or inter[-1, 0] > inter[-2, 1]:
new_inter.append(inter[-1, :])
new_inter = np.asarray(new_inter)
return new_inter
def get_subarray(a, i, axis):
"""
:param a: The array.
:param i: The indexes to select along the 'axis'
:param axis: The axis along which the indexes 'i' are selected.
:returns: The subarray selected with the indexes 'i' along the 'axis'.
"""
axis += a.ndim if axis < 0 else 0
index = ((slice(None),) * axis + (i,) + (slice(None),) * (a.ndim - axis - 1))
return a[index]
""" Vector math """
[docs]def absolute(x: array_like, axis=-1) -> array_like:
"""
:param x: A real vector or an array of real vectors.
:param axis: The axis along which the vector components are aligned.
:returns: The length(s) of the vector(s) x.
"""
x = np.asarray(x)
return np.sqrt(np.sum(x ** 2, axis=axis))
[docs]def absolute_complex(x: array_like, axis=-1):
"""
:param x: A complex vector or an array of complex vectors.
:param axis: The axis along which the vector components are aligned.
:returns: The length(s) of the vector(s) x.
"""
x = np.asarray(x)
return np.sqrt(np.sum(np.abs(x) ** 2, axis=axis))
[docs]def angle(x: array_like, y: array_like, axis=-1) -> array_like:
"""
:param x: The first vectors (arb. units).
:param y: The second vectors ([x]).
:param axis: The axis along which the vector components are aligned.
:returns: The angle between two vectors x and y (rad).
:raises ValueError: The shapes of x and y must be compatible.
"""
x, y = np.asarray(x), np.asarray(y)
check_shape_like(x, y)
return np.arccos(np.sum(x * y, axis=axis) / np.sqrt(np.sum(x ** 2, axis=axis) * np.sum(y ** 2, axis=axis)))
[docs]def angle_d(x: array_like, x_d: array_like, y: array_like, y_d: array_like, axis=-1):
"""
:param x: The first vectors (arb. units).
:param x_d: The uncertainties of the first vectors ([x]).
:param y: The second vectors ([x]).
:param y_d: The uncertainties of the second vectors ([x]).
:param axis: The axis along which the vector components are aligned.
:returns: The angle between two vectors x and y (rad).
:raises ValueError: The shapes of x and y must be compatible.
"""
z = np.sum(x * y, axis=axis)
n = np.sum(x ** 2, axis=axis) * np.sum(y ** 2, axis=axis)
arg = z / np.sqrt(n)
dx = np.sum(((n * y - z * x * np.sum(y ** 2, axis=axis)) / n ** 1.5 * x_d) ** 2, axis=axis)
dy = np.sum(((n * x - z * y * np.sum(x ** 2, axis=axis)) / n ** 1.5 * y_d) ** 2, axis=axis)
return np.sqrt((dx + dy) / (1 - arg ** 2))
[docs]def unit_vector(indexes: array_like, dim: int, axis: int = -1, dtype: type = float):
"""
:param indexes: The indexes of the vectors that are filled with a 1.
:param dim: The dimension of the unit vectors.
:param axis: The axis with the dimension 'dim'.
:param dtype: The data type of the unit vectors.
:returns: Unit vectors in the specified 'indexes' with dimension 'dim' along the specified 'axis'.
"""
index = np.asarray(indexes, dtype=int)
return np.moveaxis(np.choose(np.expand_dims(index, axis=-1), np.identity(dim, dtype=dtype)), -1, axis)
def vector_to_diag_matrix(a, axis: int = -1):
"""
:param a: The vector to be transformed into a diagonal matrix.
:param axis: The axis of the vector elements to be aligned along the diagonal of the matrix.
:returns: A diagonal matrix with one extra axis compared with 'a' at 'axis' + 1.
"""
a = np.asarray(a)
if axis < 0:
axis += len(a.shape)
shape = tuple(x1 if i > axis else x0 for i, (x0, x1) in enumerate(zip(a.shape + (1, ), (1, ) + a.shape)))
b = np.broadcast_to(np.expand_dims(a, axis=axis + 1), shape)
ident = np.identity(shape[axis], dtype=a.dtype)
if axis > 0:
ident = np.expand_dims(ident, axis=tuple(range(axis)))
if len(a.shape) - 1 > axis:
ident = np.expand_dims(ident, axis=tuple(range(axis + 2, len(b.shape))))
return b * ident
[docs]def e_r(theta: array_like, phi: array_like, axis=-1) -> array_like:
"""
:param theta: The angle theta.
:param phi: The angle phi.
:param axis: The axis along which the vector components are aligned.
:returns: The unit vector 'e_r'. Part of an orthonormal system defined by:
e_x = e_r(pi/2, .)
e_y = e_r(0, pi/2)
e_z = e_r(0, 0)
"""
x = np.expand_dims(np.sin(theta), axis=axis)
y = np.expand_dims(np.cos(theta) * np.sin(phi), axis=axis)
z = np.expand_dims(np.cos(theta) * np.cos(phi), axis=axis)
return np.concatenate([x, y, z], axis=axis)
[docs]def e_theta(theta: array_like, phi: array_like, axis=-1) -> array_like:
"""
:param theta: The angle theta.
:param phi: The angle phi.
:param axis: The axis along which the vector components are aligned.
:returns: The unit vector 'e_theta'. Part of an orthonormal system defined by:
e_x = e_r(pi/2, .)
e_y = e_r(0, pi/2)
e_z = e_r(0, 0)
"""
x = np.expand_dims(np.cos(theta) * np.ones_like(phi), axis=axis)
y = np.expand_dims(-np.sin(theta) * np.sin(phi), axis=axis)
z = np.expand_dims(-np.sin(theta) * np.cos(phi), axis=axis)
return np.concatenate([x, y, z], axis=axis)
[docs]def e_phi(theta: array_like, phi: array_like, axis=-1) -> array_like:
"""
:param theta: The angle theta.
:param phi: The angle phi.
:param axis: The axis along which the vector components are aligned.
:returns: The unit vector 'e_phi'. Part of an orthonormal system defined by:
e_x = e_r(pi/2, .)
e_y = e_r(0, pi/2)
e_z = e_r(0, 0)
"""
x = np.expand_dims(np.zeros_like(phi) * np.ones_like(theta), axis=axis)
y = np.expand_dims(np.cos(phi) * np.ones_like(theta), axis=axis)
z = np.expand_dims(-np.sin(phi) * np.ones_like(theta), axis=axis)
return np.concatenate([x, y, z], axis=axis)
[docs]def orthonormal_rtp(r: array_like, axis=-1) -> (array_like, array_like, array_like):
"""
:param r: The first vector of the orthonormal system. Does not have to be normalized.
:param axis: The axis along which the vector components are aligned.
:returns: The three orthonormal vectors 'e_r', 'e_theta' and 'e_phi'
given the vector 'r' pointing in the 'e_r' direction.
"""
r = np.asarray(r)
check_dimension(3, axis, r)
r_abs = np.expand_dims(absolute(r, axis=axis), axis=axis)
e_r0 = r / r_abs
theta = np.arcsin(np.take(e_r0, 0, axis=axis))
valid = np.abs(np.abs(theta) - np.pi / 2.) > 1e-5
phi = np.full_like(theta, 0.)
if valid.size != 0:
phi[valid] = np.arcsin(np.take(e_r0, 1, axis=axis)[valid] / np.cos(theta[valid]))
return e_r0, e_theta(theta, phi, axis=axis), e_phi(theta, phi, axis=axis)
[docs]def orthonormal(r: array_like, axis=-1) -> (array_like, array_like, array_like):
"""
:param r: The first vector of the orthonormal system. Does not have to be normalized.
:param axis: The axis along which the vector components are aligned.
:returns: Three orthonormal vectors, given the first vector 'r'.
"""
r1 = np.asarray(r, dtype=float)
check_dimension(3, axis, r1)
r1 = np.moveaxis(r1, axis, -1)
i_seed = np.argmin(np.abs(r1), axis=-1)
w = unit_vector(i_seed, 3)
r2 = np.cross(r1, w, axisa=-1, axisb=-1)
r3 = np.cross(r1, r2, axisa=-1, axisb=-1)
r1 /= np.expand_dims(absolute(r1, axis=-1), axis=-1)
r2 /= np.expand_dims(absolute(r2, axis=-1), axis=-1)
r3 /= np.expand_dims(absolute(r3, axis=-1), axis=-1)
r1 = np.moveaxis(r1, -1, axis)
r2 = np.moveaxis(r2, -1, axis)
r3 = np.moveaxis(r3, -1, axis)
return r1, r2, r3
[docs]def rotation_matrix(alpha: array_like, dr: array_iter):
"""
:param alpha: The angle to rotate.
:param dr: The vector to rotate about.
:returns: The rotation matrix defined by the angle alpha and the vector dr.
"""
alpha, dr = np.asarray(alpha), np.asarray(dr)
dr /= absolute(dr)
r = np.identity(3) * np.cos(alpha)
r += np.array([[0., -dr[2], dr[1]], [dr[2], 0., -dr[0]], [-dr[1], dr[0], 0.]]) * np.sin(alpha)
r += np.einsum('i,j->ij', dr, dr) * (1. - np.cos(alpha))
return r
[docs]class Rotation:
"""
An object specifying a rotation in 3d-space. The rotation is defined
through an angle 'alpha' and an rotational axis 'dr' by the user.
Additional instance attributes are the angle in degree 'alpha_deg' and the rotational matrix 'R'.
"""
def __init__(self, alpha: scalar = 0., dr: array_iter = None):
"""
:param alpha: The angle of the rotation (rad).
:param dr: The rotational axis of the rotation.
"""
self.alpha = alpha
self.alpha_deg = self.alpha * 180. / np.pi
if dr is None:
dr = np.array([0., 0., 1.])
self.dr = np.asarray(dr, dtype=float)
check_shape((3,), self.dr, allow_scalar=False)
self.dr /= absolute(self.dr)
self.R = rotation_matrix(self.alpha, self.dr)
[docs]def rotation_to_vector(x: array_iter, y: array_iter):
"""
:param x: The first vector.
:param y: The second vector.
:returns: A Rotation object which rotates the first vector onto the second.
"""
_x, _y = np.asarray(x), np.asarray(y)
check_shape((3, ), _x, _y, allow_scalar=False)
_x /= absolute(_x)
_y /= absolute(_y)
diff = _y - _x
if np.all(diff == 0.):
return Rotation()
elif np.all(diff == 2. * _y):
_x, dr, _ = orthonormal(_x)
return Rotation(alpha=np.pi, dr=dr)
alpha = angle(x, y)
dr = np.cross(x, y)
return Rotation(alpha=alpha, dr=dr)
""" External program helpers """
[docs]def import_iso_shifts_tilda(db: str, iso_shifts: dict):
"""
:param db: Location of a Tilda-compatible database.
:param iso_shifts: A dictionary of isotope shifts
with the structure {iso_str: {line_str: [val, stat_err, syst_err]}}.
:returns: None. Writes the entries of the given dict to the specified Tilda-compatible database.
"""
con = sqlite3.connect(db)
cur = con.cursor()
for iso, line_dict in iso_shifts.items():
for line, val in line_dict.items():
print(val)
cur.execute('INSERT OR REPLACE INTO Combined (iso, parname, run) VALUES (?, ?, ?)',
(iso, 'shift', line))
con.commit()
cur.execute('UPDATE Combined SET config=?, final=?, '
'val = ?, statErr = ?, statErrForm=?, systErr = ?, systErrForm=? '
'WHERE iso = ? AND parname = ? AND run = ?',
('[]', 0, val[0], val[1], 'err', val[2] if len(val) > 2 else 0., 0, iso, 'shift', line))
con.commit()
con.close()