CODEWITHSUNDEEP
pythonarraysnumpy
Cupitor
Cupitor

Reputation: 11657

Correct way of passing arguments which is an array of arrays

Suppose you have an array of two arrays. How can you make a function correctly separate the parts and apply the lambda function here on each of them?

import numpy as np
a=np.asarray([[1]])
b=np.asarray([[1,2]])
r=np.asarray([a,b])
f=lambda x,y:x[:,0]+y
print f(*r)

This code will rise: IndexError: too many indices. However simple change of a two a (1,2) array will change everything since numpy makes r an ndarray and correctly apply function to each of a and b. I understand this is because that what f gets for its first argument is [array([[1]])] but not [[1]]. Is there any way to generate the same behaviour in the first case as well?

import numpy as np
a=np.asarray([[1,2]])
b=np.asarray([[1,3]])
r=np.asarray([a,b])
f=lambda x,y:x[:,0]+y
print f(*r)

With output:

[[2 4]]

EDIT: Just to clarify since f will be called in my real code for million times and its more complex its important for me to keep this procedure as efficient as possible. Thanks.

Upvotes: 0

Views: 60

Answers (1)

HYRY
HYRY

Reputation: 97331

You need to make a asobjectarray() function:

import numpy as np

def asobjectarray(alist):
    r = np.ndarray(len(alist), object)
    r[:] = alist
    return r

a = np.asarray([[1]])
b = np.asarray([[1,2]])
r = asobjectarray([a,b])
f=lambda x,y:x[:,0]+y
print f(*r)

Upvotes: 1

Related Questions