I have 3 vectors like the following:
a = np.ones(20) b = np.zeros(20) c = np.ones(20) I am trying to combine them into one matrix of dimension 20x3.
Currently I am doing:
n1 = np.vstack((a,b)) n2 = np.vstack((n1,c)).T This works, but isn't there a way to fill matrix with arrays in column-wise fashion?
There are a few different ways you could do this. Here are a some examples:
Using np.c_:
np.c_[a, b, c] Using np.dstack and np.squeeze:
np.dstack((a, b, c)).squeeze() Using np.vstack and transpose (similar to your method):
np.vstack((a,b,c)).T Using np.concatenate and reshape:
np.concatenate((a, b, c)).reshape((-1, 3), order='F') If efficiency matters here, the last method using np.concatenate appears to be by far the quickest on my computer:
>>> %timeit np.c_[a, b, c] 10000 loops, best of 3: 46.7 us per loop >>> %timeit np.dstack((a, b, c)).squeeze() 100000 loops, best of 3: 18.2 us per loop >>> %timeit np.vstack((a,b,c)).T 100000 loops, best of 3: 17.8 us per loop >>> %timeit np.concatenate((a, b, c)).reshape((-1, 3), order='F') 100000 loops, best of 3: 3.41 us per loop Yes, use column_stack:
np.column_stack((a,b,c)) That works for stacking general 1-d arrays. In your specific case where you want a 20x3 array such that each row is (1,0,1), I'd suggest:
np.tile([1.,0.,1.], (20,1)) 