Declaring a numpy array and c-pointer in cython

Question

In my code, I usually use numpy arrays to interact between methods and classes. Optimizing the main parts of my program, I use cython with c pointers to these numpy arrays. Unfortunately, the way I'm declaring arrays right now is pretty long.

For example, let's say I have a method that should return a numpy array someArrayNumpy, but inside the function pointers * someArrayPointers should be used for speed. Here is how I usually declare it:

cdef: numpy.ndarray someArrayNumpy = numpy.zeros(someArraySize) numpy.ndarray[numpy.double_t, ndim=1] someArrayBuff = someArrayNumpy double *someArrayPointers = <double *> someArrayBuff.data [... some Code ...] return someArrayNumpy 

As you can see, this takes 3 lines of code for basically one array, and often I have to declare more of these arrays.

Is there a more compact / smart way to do this? I think something is missing.

EDIT:

So, because it was set by J. Martinot-Lagarde, I timed C-pointers and "numpy pointers". The code was basically

 for ii in range(someArraySize): someArrayPointers[ii] += 1 

and

 for ii in range(someArraySize): someArrayBuff[ii] += 1 

with the definitions above, but I added "ndim = 1, mode = 'c" to make sure. Results for someArraySize = 1e8 (time in ms):

 testMartinot("cPointers") 531.276941299 testMartinot("numpyPointers") 498.730182648 

What I roughly remember from previous / different tests.

Answer 1

You actually declare two numpy arrays here, the first of them is common, and the second has a specific type of dtype. You can skip the first line, someArrayBuff is ndarray.

This gives:

 numpy.ndarray[numpy.double_t] someArrayNumpy = numpy.zeros(someArraySize) double *someArrayPointers = <double *> someArrayNumpy.data 

You need at least two lines because you use someArrayPointers and return someArrayNumpy, so you need to declare them.

---

As you noticed, are you sure pointers are faster than ndarrays if you declare the type and number of dimensions of an array?

 numpy.ndarray[numpy.double_t, ndim=2] someArrayNumpy = numpy.zeros(someArraySize)