从Numpy中的变换点构造分段函数

Question

I want to construct a piecewise function from changepoints in python. I expect my inputs and outputs to be large, so speed is important.

Input:

• int numpy array: A = [1,7, 1000, 1500]
• bool numpy array: B = [True, False, True, True, False, True, False, False] where the length of A is equal to the number of True in B

Output:

• int numpy array: C = [1, 1, 7, 1000, 1000, 1500, 1500, 1500] where the length of C is the same as the length of B

Essentially each element of A is repeated until the next True in B shows up in which case the next element of A is used.

Answer 1

In [1]: import numpy

In [2]: A = numpy.array([1, 7, 1000, 1500])

In [3]: B = numpy.array([True, False, True, True, False, True, False, False])

In [4]: A[B.cumsum() - 1]
Out[4]: array([   1,    1,    7, 1000, 1000, 1500, 1500, 1500])

B.cumsum() - 1 computes which element of A to use for each element of the output, and then A[B.cumsum() - 1] extracts those elements. You could probably also work out a way to use numpy.repeat to do this.