Why during slicing assignment in numpy array decimals disappears?

Question

from numpy import *
a=array([0.,0.001,0.002])
b=array([[1,11],[2,22],[3,33]])
b[:,1]=a
print b

I expected as a result:

array([[  1. , 0. ],[  2. ,  0.001 ],[  3. , 0.002 ]])

But I got:

array([[ 1 , 0 ], [ 2 , 0 ] , [ 3 , 0 ]])

To obtain desired result I had to type:

from numpy import *
a=array([0.,0.001,0.002])
b=array([[1,11],[2,22],[3,33]])
b=b.astype(float)
b[:,1]=a
print b

Is it a bug? Shouldn't assignment automatically make numpy array a float type?

It's not really possible to change the dtype of an array in-place. The internal memory layout of an int32 array isn't compatible with that of a float64 array. — user2357112
– user2357112, Commented Feb 2, 2014 at 0:05

Ashwini Chaudhary · Accepted Answer · 2014-02-02 00:04:10Z

3

No, it is not a bug. From docs:

Note that assignments may result in changes if assigning higher types to lower types (like floats to ints) or even exceptions (assigning complex to floats or ints)

answered Feb 2, 2014 at 0:04

Ashwini Chaudhary

252k60 gold badges478 silver badges519 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Aleksander_B Over a year ago

Thanks! I think this fact is not well known, I mean I was googling quite a lot, and in simple tutorials for beginners, and even for intermediate users that is not exposed :) Thank you once again :)

WGS Over a year ago

It is actually a very core concept in programming, IMHO, not only in Python. The differences are there across languages, but it's pretty much present. It's not often you'll see them in tutorials, but they are covered extensively in the beginning parts of most good books, regardless of language covered. :)

Aleksander_B Over a year ago

Yeah, syntax is not everything I guess, the basics of hardware handling are also important. Could you (BK201) mention some title of such a good book? :)

Akavall · Accepted Answer · 2014-02-02 01:17:58Z

0

The behavior seems intuitive to me, array b remains the same dtype before and after the assignment, so dtype of a needs to be changed to dtype of b.

>>> a.astype(b.dtype) # and when you convert a to dtype of b you get:
array([0, 0, 0])
>>> 
>>> b[:, 1] = a.astype(b.dtype) # I believe this is what is going on under the hood.
>>> b
array([[1, 0],
       [2, 0],
       [3, 0]])

answered Feb 2, 2014 at 1:17

Akavall

86.8k58 gold badges214 silver badges261 bronze badges

Collectives™ on Stack Overflow

Why during slicing assignment in numpy array decimals disappears?

2 Answers 2

3 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related