[Numpy-discussion] Concatenating string arrays

[Francesc Alted]

A Wednesday 18 March 2009, Sturla Molden escrigué:
> On 3/18/2009 7:30 PM, Thomas Robitaille wrote:
> > import numpy as np
> > arr1 = np.array(['a','b','c'])
> > arr2 = np.array(['d','e','f'])
> >
> > I would like to produce a third array that would contain
> > ['ad','be','cf']. Is there an efficient way to do this? I could do
> > this element by element, but I need a faster method, as I need to
> > do this on arrays with several million elements.
> >
>  >>> arr1 = np.array(['a','b','c'])
>  >>> arr2 = np.array(['d','e','f'])
>  >>> arr3 = np.zeros(6, dtype='|S1')
>  >>> arr3[::2] = arr1
>  >>> arr3[1::2] = arr2
>  >>> arr3.view(dtype='|S2')
>
> array(['ad', 'be', 'cf'],
>        dtype='|S2')

Nice example.  After looking at this, it is apparent how beneficial can 
be the mutable types provided by NumPy.

-- 
Francesc Alted