pychemstation.utils.num_utils

 1from typing import Tuple
 2
 3import numpy as np
 4
 5
 6def find_nearest_value_index(array, value) -> Tuple[float, int]:
 7    """Returns closest value and its index in a given array.
 8
 9    :param array: An array to search in.
10    :type array: np.array(float)
11    :param value: Target value.
12    :type value: float
13
14    :return: Nearest value in array and its index.
15    """
16
17    index_ = np.argmin(np.abs(array - value))
18    return array[index_], index_
19
20
21def interpolate_to_index(array, ids, precision: int = 100) -> np.ndarray:
22    """Find value in between arrays elements.
23
24    Constructs linspace of size "precision" between index+1 and index to
25    find approximate value for array[index], where index is float number.
26    Used for 2D data, where default scipy analysis occurs along one axis only,
27    e.g. signal.peak_width.
28
29    Rough equivalent of array[index], where index is float number.
30
31    :param array: Target array.
32    :type array: np.array(float)
33    :param ids: An array with "intermediate" indexes to interpolate to.
34    :type ids: np.array[float]
35    :param precision: Desired presion.
36
37    :return: New array with interpolated values according to provided indexes "ids".
38
39    Example:
40        >>> interpolate_to_index(np.array([1.5]), np.array([1, 2, 3], 100))
41            ... array([2.50505051])
42    """
43
44    # breaking ids into fractional and integral parts
45    prec, ids = np.modf(ids)
46
47    # rounding and switching type to int
48    prec = np.around(prec * precision).astype("int32")
49    ids = ids.astype("int32")
50
51    # linear interpolation for each data point
52    # as (n x m) matrix where n is precision and m is number of indexes
53    space = np.linspace(array[ids], array[ids + 1], precision)
54
55    # due to rounding error the index may become 100 in (100, ) array
56    # as a consequence raising IndexError when such array is indexed
57    # therefore index 100 will become the last (-1)
58    prec[prec == 100] = -1
59
60    # precise slicing
61    true_values = np.array(
62        [space[:, index[0]][value] for index, value in np.ndenumerate(prec)]
63    )
64
65    return true_values