current community

your communities

more stack exchange communities

company blog
Stack Exchange Inbox Reputation and Badges
tour help
stack overflow careers
Sign up ×
Stack Overflow is a question and answer site for professional and enthusiast programmers. It's 100% free.
up vote 0 down vote favorite

I want to assign row-specific values to each row in a 2D numpy array. In particular, each row should contain the value of the reciprocal of its row number. In other words, all columns in row 1 should have values 1, all columns in row 2 should be 1/2, all in row 3 should be 1/3, and so on. I tried this:

m = 3
n = 10

train = np.empty([n,m], float)

for curr_n in range(n):
    train[curr_n,:] = 1/(curr_n+1)

print train

But the output is:

[[ 1.  1.  1.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]
 [ 0.  0.  0.]]

What am I doing wrong? I used "float" at the beginning, but I'm still getting solid 0 integers for all but the first row.

share|improve this question
2  
you need 1. to convert it to a float. – John Barça Apr 21 '14 at 12:19
    
np.true_divide – Constantine Apr 21 '14 at 13:30
1  
1/(curr_n+1) always return 1 or 0 because curr_n is a type of integer. You might want to do like 1.0/(curr_n+1) or 1/float(curr_n+1). – keimina Apr 21 '14 at 13:51

2 Answers 2

up vote 5 down vote accepted

Implicit type conversion has been added in Python 3, so your original code will then work, as is:

from __future__ import division

m = 3
n = 10

train = np.empty([n,m], float)

for curr_n in range(n):
   train[curr_n,:] = 1/(curr_n+1)

print train

Some information on the __future__ module can be seen in the official docs future module

Or, as sshashank124 put in his answer and I put in the comment, you can use 1. or 1.0 to force float behavior.

share|improve this answer
up vote 3 down vote

You have a simple type problem. Use 1.0 instead of 1 in your reciprocation. The following works:

m = 3
n = 10

train = np.empty([n,m])

for curr_n in range(n):
    train[curr_n,:] = 1.0/(curr_n+1)   #converted 1 to 1.0

print train

Explanation

Although you might think that numpy deals with floats by default, in this case, the numpy array is getting assigned the value after python has had the time to calculate the inverses. It is then that python truncates your floats to an int and numpy innocently converts that sneaky int to a float just as it is supposed to and ends up taking all the blame.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.