iblutil.numerical

Functions

between_sorted	Given a vector of sorted values, returns a boolean vector indices True when the value is between bounds. If multiple bounds are given, returns the equivalent OR of individual bounds tuple Especially useful for spike times indices = between_sorted(spike_times, [tstart, tstop]).
bincount2D	Computes a 2D histogram by aggregating values in a 2D array.
hash_uuids	Returns a hash of an array or list of UUID strings
intersect2d	Performs intersection on multiple columns arrays a0 and a1
ismember	equivalent of np.isin but returns indices as in the matlab ismember function returns an array containing logical 1 (true) where the data in A is B also returns the location of members in b such as a[lia] == b[locb]
ismember2d	Equivalent of np.isin but returns indices as in the matlab ismember function returns an array containing logical 1 (true) where the data in A is B also returns the location of members in b such as a[lia, :] == b[locb, :]
rcoeff	Computes pairwise Pearson correlation coefficients for matrices.
within_ranges	Detects which points of the input vector lie within one of the ranges specified in the ranges.

between_sorted(sorted_v, bounds=None)

Given a vector of sorted values, returns a boolean vector indices True when the value is between bounds. If multiple bounds are given, returns the equivalent OR of individual bounds tuple Especially useful for spike times

indices = between_sorted(spike_times, [tstart, tstop])

Parameters:

• sorted_v – vector containing sorted values (won’t check)

• bounds – minimum included value and maximum included value can be a list[tstart, tstop] or an array of dimension (n, 2)

Returns:

hash_uuids(uuids, algo='sha256')

Returns a hash of an array or list of UUID strings

Parameters:

• uuids – collection of UUID strings: list | np.ndarray | pd.Series | pd.Index

• algo – hashing algorithm to use, default is sha256

ismember(a, b)

equivalent of np.isin but returns indices as in the matlab ismember function returns an array containing logical 1 (true) where the data in A is B also returns the location of members in b such as a[lia] == b[locb]

Parameters:

• a – 1d - array

• b – 1d - array

Returns:

isin, locb

ismember2d(a, b)

Equivalent of np.isin but returns indices as in the matlab ismember function returns an array containing logical 1 (true) where the data in A is B also returns the location of members in b such as a[lia, :] == b[locb, :]

Parameters:

• a – 2d array

• b – 2d array

Returns:

isin, locb

intersect2d(a0, a1, assume_unique=False)

Performs intersection on multiple columns arrays a0 and a1

Parameters:

• a0

• a1

• assume_unique – If True, the input arrays are both assumed to be unique,

which can speed up the calculation. :return: intersection :return: index of a0 such as intersection = a0[ia, :] :return: index of b0 such as intersection = b0[ib, :]

bincount2D(x, y, xbin=0, ybin=0, xlim=None, ylim=None, weights=None)

Computes a 2D histogram by aggregating values in a 2D array.

Parameters:

• x – values to bin along the 2nd dimension (c-contiguous)

• y – values to bin along the 1st dimension

• xbin – scalar: bin size along 2nd dimension 0: aggregate according to unique values array: aggregate according to exact values (count reduce operation)

• ybin – scalar: bin size along 1st dimension 0: aggregate according to unique values array: aggregate according to exact values (count reduce operation)

• xlim – (optional) 2 values (array or list) that restrict range along 2nd dimension

• ylim – (optional) 2 values (array or list) that restrict range along 1st dimension

• weights – (optional) defaults to None, weights to apply to each value for aggregation

Returns:

3 numpy arrays MAP [ny,nx] image, xscale [nx], yscale [ny]

rcoeff(x, y)

Computes pairwise Pearson correlation coefficients for matrices.

That is for 2 matrices the same size, computes the row to row coefficients and outputs a vector corresponding to the number of rows of the first matrix. If the second array is a vector then computes the correlation coefficient for all rows.

Parameters:

• x – np array [nc, ns]

• y – np array [nc, ns] or [ns]

Returns:

r [nc]

within_ranges(x: ndarray, ranges: ndarray | Sequence, labels: ndarray | Sequence | None = None, mode: str = 'vector', dtype: Type[D] = 'int8') → ndarray

Detects which points of the input vector lie within one of the ranges specified in the ranges. Returns an array the size of x with a 1 if the corresponding point is within a range.

The function uses a stable sort algorithm (timsort) to find the edges within the input array. Edge behaviour is inclusive.

Ranges are [(start0, stop0), (start1, stop1), etc.] or n-by-2 numpy array. The ranges may be optionally assigned a row in ‘matrix’ mode or a numerical label in ‘vector’ mode. Labels must have a length of n. Overlapping ranges have a value that is the sum of the relevant range labels (ones in ‘matrix’ mode).

If mode is ‘vector’ (default) it will give a vector, specifying the range of each point. If mode is ‘matrix’ it will give a matrix output where each range is assigned a particular row index with 1 if the point belongs to that range label. Multiple ranges can be assigned to a particular row, e.g. [0, 0,1] would give a 2-by-N matrix with the first two ranges in the first row. Points within more than one range are given a value > 1

Parameters:

• x (array_like) – An array whose points are tested against the ranges. multi-dimensional arrays are flattened to 1D

• ranges (array_like) – A list of tuples or N-by-2 array of ranges to test, where N is the number of ranges, i.e. [[start0, stop0], [start1, stop1]]

• labels (vector, list) – If mode is ‘vector’; a list of integer labels to demarcate which points lie within each range. In ‘matrix’ mode; a list of column indices (ranges can share indices). The number of labels should match the number of ranges. If None, ones are used for all ranges.

• mode ({'matrix', 'vector'}) – The type of output to return. If ‘matrix’ (default), an N-by-M matrix is returned where N is the size of x and M corresponds to the max index in labels, e.g. with labels=[0,1,2], the output matrix would have 3 columns. If ‘vector’ a vector the size of x is returned.

• dtype (str, numeric or boolean type) – The data type of the returned array. If type is bool, the labels in vector mode will be ignored. Default is int8.

Returns:

• A vector of size like x where zeros indicate that the points do not lie within ranges (

• ‘vector’ mode) or a matrix where out.shape[0] == x.size and out.shape[1] == max(labels) + 1.

Examples

# Assert that points in ranges are mutually exclusive np.all(within_ranges(x, ranges) <= 1)

Tests

>>> import numpy as np
>>> within_ranges(np.arange(11), [(1, 2), (5, 8)])
array([0, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0], dtype=int8)
>>> ranges = np.array([[1, 2], [5, 8]])
>>> within_ranges(np.arange(10) + 1, ranges, labels=np.array([0,1]), mode='matrix')
array([[1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 1, 1, 1, 1, 0, 0]], dtype=int8)
>>> within_ranges(np.arange(11), [(1,2), (5,8), (4,6)], labels=[0,1,1], mode='matrix')
array([[0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 1, 2, 2, 1, 1, 0, 0]], dtype=int8)
>>> within_ranges(np.arange(10) + 1, ranges, np.array([3,1]), mode='vector')
array([3, 3, 0, 0, 1, 1, 1, 1, 0, 0], dtype=int8)
>>> within_ranges(np.arange(11), [(1,2), (5,8), (4,6)], dtype=bool)
array([False,  True,  True, False,  True,  True,  True,  True,  True,
       False, False])