CODEWITHSUNDEEP
pythonarraysnumpyfloating-point
ekl
ekl

Reputation: 1051

Really strange numpy array floating point behavior

I am working with numpy for the first time and am experiencing some very strange issues with floating-point arrays.

This is extremely basic and I am probably missing something very obvious here -- can anyone tell me what the problem is?

These two lines of code

arr1 = numpy.ndarray([1.0, 2.0])
print "arr1: ", arr1

produce this output:

arr1:  [[  1.49166815e-154  -1.32750723e-315]]

That's... not right. What am I doing wrong?

Thank you for any help!

Upvotes: 3

Views: 434

Answers (2)

Ashwini Chaudhary
Ashwini Chaudhary

Reputation: 250951

You should use numpy.array to create an array not numpy.ndarray. numpy.ndarray is a low-level interface, and in most cases numpy.array should be used to create an array.

In [5]: arr1 = numpy.array([1.0, 2.0])                                                            

In [6]: arr1
Out[6]: array([ 1.,  2.])

Signature of numpy.ndarray:

ndarray(shape, dtype=float, buffer=None, offset=0, strides=None, order=None)

So, the first argument is shape not the array. So, numpy filled your array with some random data.

From the docstring of numpy.ndarray:

Arrays should be constructed using array, zeros or empty.

Upvotes: 4

Jayanth Koushik
Jayanth Koushik

Reputation: 9894

The argument that you are specifying is the shape. To fill with data you need to specify the buffer argument.

np.ndarray(shape=(1,2), buffer=np.array([1,2]), dtype=float)

Upvotes: 2

Related Questions