[ODE] solving penetrations "statically"

[Dirk Gregorius]

>> You can't do that if your nonpenetration constraint equations are expressed
>> in velocity like ODE. You always have some drift in position.


You can't do what? The method of Erin is first correct - see for example 
the thesis of Bradley Cline - and
secondly works perfectly in practice. Especially in the context of game 
development. Each position constraint 
imposes a (hidden) velocity and acceleration constraint to the system. 
If you have a velocity constraint you also
have geometric constraint as well (I must admit that I am not sure for 
friction constraints), but definitely for
non-penetration constraints and joints.


-Dirk