def test_resample_from_linear_density__numerical_innacurracy():
"""
Previous implementation was using arange to calculate the new paths,
which led to the accumulation of a numerical error. As a result the
last path was numerically smaller that the total length in the resampling
function which led in the generation of an extra point.
"""
points = np.array([[-1.474797606, -12.873973846, 1.465831757],
[-1.268972993, -13.383681297, 1.386174798],
[-1.207281470, -13.681849480, 1.339448690],
[-0.772920549, -14.285424232, 1.751677036],
[-0.640972078, -14.575642586, 1.706354022],
[-0.157932609, -15.903386116, 1.498738170],
[0.108525813, -16.787338257, 1.856412053],
[0.603999615, -17.670700073, 1.718663454],
[1.443708062, -19.519208908, 1.925997615],
[1.577559590, -19.740325928, 1.891538978],
[2.688630342, -22.260030746, 2.004102230],
[3.322759628, -23.588979721, 2.292642832],
[4.462557793, -25.080135345, 2.556796074],
[4.665114880, -25.744903564, 2.948777676]])
total = np.linalg.norm(points[:1] - points[:-1], axis=1).sum()
ids, fractions = tested._resample_from_linear_density(points,
linear_density=1.)
new_points = tested._parametric_values(points, ids, fractions)
assert not np.allclose(new_points[-1], new_points[-2])
def test_resample_parametric_values():
ids = np.array([0, 0, 1])
fractions = np.array([0.5, 1.0, 0.5])
values = np.array([[0., 0., 0.], [1., 0., 0.], [2., 0., 0.]])
# Using and array of 3D points
new_values = tested._parametric_values(values, ids, fractions)
npt.assert_allclose(new_values,
[[0.0, 0., 0.], [0.5, 0., 0.], [1.0, 0., 0.],
[1.5, 0., 0.], [2.0, 0., 0.]])
# using an array of 1D values
values = np.array([1., 2., 3.])
new_values = tested._parametric_values(values, ids, fractions)
npt.assert_allclose(new_values, [1.0, 1.5, 2.0, 2.5, 3.0])