The function piecewise(x, condlist, funclist) returns an output which has the same dtype as the input x. However, this input is processed by one of the functions in funclist and thus it should return an array which dtype same as those function's output.

for example

def func1(v):
    return v + 0.1

def func2(v):
    return v + 0.2

funclist = [func1, func2]
condlist = [[True, False], [False, True]]

x = np.array([1, 2])
np.piecewise(x, condlist, funclist)

# array([1, 2])

However, the actual expected answer should be array([1.1, 2.2]), which can be attained, but we must make sure the input is specified with float dtype, e.g.: x = array([1.0, 2.0]).

Since we have functions here operating on each element of x, we expect the output to be same as functions outputs.

NumPy/Python version information:

1.19.5 3.7.9 (default, Aug 31 2020, 12:42:55)
[GCC 7.3.0]