Real-Time Physics Simulation Forum

I am confused about what kind of complementarity problem that Bullet solved. I have read some thesis and posts in the forum. Someone stated that Bullet solved the MLCP for the contact, due to the frcition cone was simplified. But others said Bullet can solve the NCP directly with the PGS solver. I don't know which side to believe. Based on my understading, MLCP treatment is a simplification of the NCP due to the friction simplified treatment. But if PGS can solve the NCP directly, why we should simplify the friction cone? Does Bullet really treat the firciton cone as a polyhedral approximation?

Thank you at first!
Daye

Hi, I am running into the same problem trying to understand what is actually running in the bullet stimulator.
Have you found any answer yet?