Annoying loss of precision when transforming numpy array to list

Python Help
1

Dear Python experts,

I am seeing this:

In [15]: numpy.linspace(-1, +1, 10 + 1,).tolist()
Out[15]:
[-1.0,
 -0.8,
 -0.6,
 -0.3999999999999999,
 -0.19999999999999996,
 0.0,
 0.20000000000000018,
 0.40000000000000013,
 0.6000000000000001,
 0.8,
 1.0]

In [16]:

i.e. there is a loss of precision when going to a list. I will later convert the floats into strings like 0p4 and this will mess up things. Do you know how to prevent this?

Cheers.

2

There is no loss of precision, floating point numbers just aren’t precises: https://0.30000000000000004.com/

numpy might be default cut the numbers of earlier, but you can do the same with options for your string conversion.

3

If it is a choice that you can make, use 1, 2 and 2^n+1 as the number of elements: 1,2,3,5,9,17,33,65, 129, ....

4

See also 15. Floating Point Arithmetic: Issues and Limitations — Python 3.12.4 documentation

Not all numbers are representable as a (finite) binary (floating point) number. For example 1/5=0.2 is not, similar to how 1/3=0.3333… is not representable as a (finite) decimal number.

The values you get are probably the (exact) floating point values that are closest to the real numbers you want.