You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
sVC = area * amp.vert[amp.faces, 2].mean(axis=1) * amp.norm[:, 2]
if return_closed is True:
return sVC.sum(), amp
else:
return sVC.sum()
amp.norm[:, 2] can have negative values, giving a negative volume contributions. The .sum() contribution does not take this into account.
suggest replacing sVC.sum() with (np.abs(sVc)).sum()
Smaller issue more likely caused by poor meshing but I belive there is possible issue if a triangle face has a very small angle the two adjacent edges, may have very similar values and with rounding their cross product will be zero/the norm will end up with nan which then results in .sum() giving nan. Can be fixed by using np.nansum((np.abs(sVC))) instead of .sum()
The text was updated successfully, but these errors were encountered:
amp.norm[:, 2] can have negative values, giving a negative volume contributions. The .sum() contribution does not take this into account.
suggest replacing sVC.sum() with (np.abs(sVc)).sum()
Smaller issue more likely caused by poor meshing but I belive there is possible issue if a triangle face has a very small angle the two adjacent edges, may have very similar values and with rounding their cross product will be zero/the norm will end up with nan which then results in .sum() giving nan. Can be fixed by using np.nansum((np.abs(sVC))) instead of .sum()
The text was updated successfully, but these errors were encountered: