Deleting multiple slices from a numpy array

Question

I have a given numpy array and a list containing a number of slice objects (alternatively containing (start, end) tuples). I am looking to remove the slice object positions from the original array and get a second array with the remaining values.

Toy example:

myarray = np.arange(20)

array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16,
       17, 18, 19])

mylist=(slice(2,4),slice(15,19))

Do something and result should be

array([0, 1, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14])

The array can be in a few hundred thousand large, the list of slice objects can contain a few thousand elements and I need to run the operation often, so speed is somewhat important.

Numpy delete does not take a list of slices as far I can see?

For now I am generating the complement of my slice object list and slicing that, but generating the complement is a somewhat awkward process where I am sorting my slice list then iterating through it, creating the complement slice objects as needed. I am hoping there is a more elegant way I have not figured!

Answer 1

You can use set() to identify which positions will be kept and np.take() to obtain the corresponding values, doing something like:

ind = np.indices(myarray.shape)[0]
rm = np.hstack([ind[i] for i in mylist])

ans = np.take(myarray, sorted(set(ind)-set(rm)))

Note that np.hstack() is used to obtain a single array with all the indices that will be removed. This takes about half the time of @HYRY's solution.