Closed andreadelprete closed 4 years ago
This looks good to me!
On Apr 9, 2020, at 01:23, Bilal Hammoud notifications@github.com wrote:
the way I could think of this is by computing normal and tangent dx to the contact surface, n is the surface normal direction
dx = x - x_start normal = (n.T dx) n tangent = dx - normal
for the velocity, v in global frame
normal_vel = (n.T dx) n tangent_vel = v - normal_vel
is my geometry correct ? is there more I have to look to in terms of exploiting cpu instructions ?
I will look into other cases as I come across them
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/andreadelprete/consim/issues/7#issuecomment-611243776, or unsubscribe https://github.com/notifications/unsubscribe-auth/AANW4OQG3GRFQPDL3EQIZATRLUBPDANCNFSM4LQ63L6A.
this was resolved, I will close it for the time being
I just noticed that often vector operations are performed through for loops. See for instance here. This is in general inefficient because vector operations can exploit custom CPU instructions, so we should try to fix that.