5.1.4 Array copies

Sometimes it is necessary to explicitly request that the data be copied. This is simply achieved with the NumPy function called array:

M = array([[1.,2.],[3.,4.]])
N = array(M.T) # copy of M.T

We can verify that the data has indeed been copied:

N.base is None # True

In this section, you saw the concept of array views. NumPy works with views instead of copies of a given array to save memory, which – especially for large arrays – can be crucial. On the other hand, unintentionally using views may cause programming errors that are hard to debug.