How can I define the precision of the array elements in python?

Question

There is an array with large input elements named A. I want to define another array, named B, that its elements are obtained by the exponential of the negative of elements in A. I use the following code: B=np.exp(-A) The resulted elements in B are very small that in python they are shown equal to zero. How I can define the precision so that I don't get zero for the elements?

Thanks for your guidance in advance, Zahra

If there is more than 17-18 digits they're not going to be accurate anyway and this is due to how they are stored under the IEEE-754 “double precision” EDIT: link for python3.7 instead of 3.2 — Plopp
– Plopp, Commented Jan 10, 2020 at 13:33

Robert King · Accepted Answer · 2020-01-10 13:37:07Z

2

You can try specifying the data type.

a = np.array([1000], dtype='f8') # 64-bit float
b = np.exp(-a)
b
>>> array([0.])

a = np.array([1000], dtype='f16') # 128 bit float
b = np.exp(-a)
b
>>> array([5.0759589e-435], dtype=float128)

Note the comment of @Plopp, depending on how large your values of a are this may not be enough

answered Jan 10, 2020 at 13:37

Robert King

9946 silver badges16 bronze badges

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

1 Comment

Zahra Over a year ago

Thank you for your guide. I tried it but the elements' values are too large and based on @Ploppa's comment, putting dtype='f16' didn't work. Is there any other way?

Collectives™ on Stack Overflow

How can I define the precision of the array elements in python?

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related