Skip to content
GitLab

Draft: Vectorize inputs of functions in euclidean.py

Closed Luka Rancic requested to merge euclidean-vectorized into develop

Changed the input of circle_fct from float to numpy.ndarray. This allows to parameterize a whole array at once rather then a single point. This is useful, because otherwise you would need to generate a list or use numpy.fromfunction to get a whole circle.

Essentially that means you can do:

>>> t = np.linspace(0, 2*np.pi, 10, endpoint=False)
>>> circle_fct(t, (0,0,0), radius=1, normals=(0,0,1))
array([[ 1.00000000e+00,  0.00000000e+00,  0.00000000e+00],
       [ 8.09016994e-01,  5.87785252e-01,  0.00000000e+00],
       [ 3.09016994e-01,  9.51056516e-01,  0.00000000e+00],
       [-3.09016994e-01,  9.51056516e-01,  0.00000000e+00],
       [-8.09016994e-01,  5.87785252e-01,  0.00000000e+00],
       [-1.00000000e+00,  1.22464680e-16,  0.00000000e+00],
       [-8.09016994e-01, -5.87785252e-01,  0.00000000e+00],
       [-3.09016994e-01, -9.51056516e-01,  0.00000000e+00],
       [ 3.09016994e-01, -9.51056516e-01,  0.00000000e+00],
       [ 8.09016994e-01, -5.87785252e-01,  0.00000000e+00]])

This is a breaking change, no tests or similar have been changed to account for it yet, I just want to get some feedback for now. I personally don't see a single reason why not to do it this way. The motivation to make this change was because in math.geometric_objects.py the function circle_fct is used very ineffectively.

There are more function in math.euclidean.py that could be vectorized like this.

Edited by Luka Rancic