Scraping for performance improvements without the use of C or CYTHON. Stretching an array for an image resolution matching issue.

stretch.py

def discrete_stretch(array, size):
    """ Expands 2d array by some factor.
        For example the following array is expanded by a factor of 3
  
        [ 1  2 ]
        [ 3  4 ]
       
        [ 1  1  1  2  2  2]
        [ 1  1  1  2  2  2]
        [ 1  1  1  2  2  2]
        [ 3  3  3  4  4  4]
        [ 3  3  3  4  4  4]
        [ 3  3  3  4  4  4]
        
        The code is not elegant or readable but timeit shows significant performance over np.kron
        for kroenecker products and marginal improvements over mixing np.tile with np.repeat
        
        Input:
            np.array - 2d array of values you want stretched
        Output:
            np.array - 2d np array of stretched values.  
    """
    
    return np.repeat(np.repeat(a, size, axis = 1), size, axis = 0).reshape( a.shape[0] * size, a.shape[1] * size )