genieclust.tools¶

Auxiliary functions

genieclust.tools.argkmin(x, k)¶

Finds the position of an order statistic. It holds argkmin(x, 0) == argmin(x), or, more generally, argkmin(x, k) == np.argsort(x)[k].

Run time is \(O(nk)\) and working memory is \(O(k)\). An insertion sort-like scheme is used to locate the order statistic. In practice, the function is very fast for small k and randomly ordered or almost sorted (increasingly) data.

Parameters:

xndarray: A vector with n elements of type int, float, or double.
kint: An integer between 0 and n - 1, preferably small.

Returns:

int: The index where the (k-1)-th smallest value in x is located.

Examples

>>> x = np.r_[2, 3, 6, 5, 1, 4]
>>> genieclust.tools.argkmin(x, 0) # index of the smallest value
4
>>> genieclust.tools.argkmin(x, 1) # index of the 2nd smallest value
0

genieclust.tools.cummax(x)¶

Cumulative maximum

Parameters:

xndarray: A vector with n elements of type int, float, or double.

Returns:

ndarray: A vector of length n whose i-th element is equal to max(x[:i]).

See also

genieclust.tools.cummin: Cumulative minimum

Examples

>>> genieclust.tools.cummax(np.r_[3, 2, 1, 4, 6, 5])
array([3, 3, 3, 4, 6, 6])

genieclust.tools.cummin(x)¶

Cumulative minimum

Parameters:

xndarray: A vector with n elements of type int, float, or double.

Returns:

ndarray: A vector of length n whose i-th element is equal to min(x[:i]).