Numpy arrays, fancy indexing, complex numbers

Question

The following code multiplies a part of the array by a number

def mul_by_num(a,b):
    a[0:2] *= b


import numpy as np
a = np.ones(5,dtype=np.float64)
mul_by_num(a,1.0)
mul_by_num(a,1j) #Generates a warning (and casts to float!)

The second call generates a warning

-c:2: ComplexWarning: Casting complex values to real discards the imaginary part

The question is, what is the most pythonic way to multiply parts of numpy arrays by complex/real numbers without messing with dtypes? I do not really want to convert an array to complex from the beginning, but the program in principle can get a complex input.

EDIT:

I do not care about copying the full array, casting it to complex; But I want to avoid checking dtypes (i.e., np.float32, np.float64, np.complex, np.int etc.)

@IgnacioVazquez-Abrams I added a comment to the question, that I just want to avoid the full check of possible dtypes. For example, a + b works just fine, but assigning a part of the array does not. — Ivan Oseledets
– Ivan Oseledets, Commented Aug 15, 2012 at 6:59

ecatmur · Accepted Answer · 2012-08-15 09:20:43Z

5

You're going to need to convert the array to complex at some point, otherwise it won't be able to hold complex numbers.

The easiest way to convert an array to complex is to add 0j:

if (np.iscomplexobj(b)):
    a = a + 0j
a[0:2] *= b

note: not a += 0j as that will attempt to modify the array inplace, which won't work if it isn't complex already.

edited Aug 15, 2012 at 9:20

answered Aug 15, 2012 at 7:22

ecatmur

158k28 gold badges311 silver badges387 bronze badges

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

2 Comments

Ivan Oseledets Over a year ago

That is ok, but what if b is real, and I do not want to convert to complex in this case?

ecatmur Over a year ago

@IvanOseledets in that case check with np.iscomplexobj(b).

conjohn668 · Accepted Answer · 2012-08-15 07:31:28Z

0

Since increasing the speed of calculating, numpy array makes sure that having same type. May be you can try the python list or casting it.

answered Aug 15, 2012 at 7:31

conjohn668

233 bronze badges

Collectives™ on Stack Overflow

Numpy arrays, fancy indexing, complex numbers

2 Answers 2

2 Comments

Comments

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Related