The long integer when using NumPy

Question

I have the following simple codes for the sum of two vectors.

However, I get the wrong results when I use NumPy.

In [12]: %run -i test.py

The last 2 elements of the sum [7980015996L, 7992002000L]

The last 2 elements of the sum [-609918596 -597932592]

It's not a long integer, Why?

import numpy as np 
def numpysum(n):
  a = np.arange(n) ** 2 
  b = np.arange(n) ** 3 
  c = a + b

  return c

def pythonsum(n): 
  a = range(n) 
  b = range(n) 
  c = []

  for i in range(len(a)): 
      a[i] = i ** 2 
      b[i] = i ** 3 
      c.append(a[i] + b[i])

  return c

size = 2000

c = pythonsum(size)
print "The last 2 elements of the sum", c[-2:]
c = numpysum(size)
print "The last 2 elements of the sum", c[-2:]

BrenBarn · Accepted Answer · 2017-04-22 06:04:10Z

4

Plain Python integers can grow arbitrarily large. Numpy integers cannot; they are limited by the size of the data type. If they get too big, they will wrap around and become negative. It looks like your array dtype is probably int32, which overflows and results in negative results. You can get the correct results in this case by using int64:

a = np.arange(n, dtype=np.int64) ** 2 
b = np.arange(n, dtype=np.int64) ** 3

However, it will still overflow eventually (if you make size larger). You could also use float64, which allows even larger numbers, but then you will lose precision.

The cap on integer sizes is the price you pay for the speed numpy gives you.

answered Apr 22, 2017 at 6:04

BrenBarn

253k39 gold badges421 silver badges392 bronze badges

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

3 Comments

Stefan Pochmann Over a year ago

Could also use dtype=object.

BrenBarn Over a year ago

@StefanPochmann: If you do that, there's not much reason to be using numpy for your computations.

Stefan Pochmann Over a year ago

You still get all the cool functionality. Like simply using that ** 3 on the array or a + b of two arrays. Or matrix multiplication as a * b etc. In pure Python that takes more code, which is more work to write and read. NumPy isn't just about execution speed.

Collectives™ on Stack Overflow

The long integer when using NumPy

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related