Search code examples
pythonnumpydtype

Which dtype would be correct to prevent numpy.arange() from getting the wrong length?

I am trying to get a shifting array containing 200 values in a range with a difference of 40.

Therefore i am using numpy.arange(a, b, 0.2) with starting values a=0 and b=40 and going upwards (a=0.2 b=40.2, a=0.4 b=40.4 and so on).

When I reach numpy.arange(25.4, 65.4, 0.2) however I suddenly get an array with a length 201 values:

x = numpy.arange(25.2, 65.2, 0.2)
print(len(x))

Returns 200

x = numpy.arange(25.4, 65.4, 0.2)
print(len(x))

Returns 201

I got so far to notice that this happens probably due to rounding issues because of the data type...

I know there is a option 'dtype' in numpy.arrange():

numpy.arange(star, stop, step, dtype)

The question is which data type would fit this problem and why? (I am not so confident with data types jet and https://docs.scipy.org/doc/numpy/reference/generated/numpy.dtype.html#numpy.dtype hasn't helped me to get this issue resolved. Please help!

Solution

  • np.arange is most useful when you want to precisely control the difference between adjacent elements. np.linspace, on the other hand, gives you precise control over the total number of elements. It sounds like you want to use np.linspace instead:

    import numpy as np

offset = 25.4

x = np.linspace(offset, offset + 40, 200)

print(x)
print(len(x))

    Here's the documentation page for np.linspace: https://docs.scipy.org/doc/numpy/reference/generated/numpy.linspace.html