[ODE] Constraints: position-, velocity-, or acceleration-based?

[nlin@nlin.net]

Hello,

My previous question didn't get any response on the list, and another poster
emailed me privately with the same question, so I am reposting a more
concise query which hopefully someone can answer.

1. Does ODE use velocity-based constraints (where I understand
   "velocity-based" to mean "expressed in terms of the first derivative
   of the state vector")? Does ODE use a time-stepping method?
2. Baraff appears to use acceleration-based constraints. Why does ODE use
   velocity-based constraints? Is it because of friction problems with
   multiple/no solutions? Do these problems (at least, the "no-solution"
   but not necessarily the "multiple solutions" problem) disappear with
   a velocity-based time-stepping formulation?
3. Some other systems appear to use position-based constraints (I think
   Bart Barenburg's Dynamo library does this). What advantages or
   disadvantages would this have?

Basically the general question is why one would prefer one constraint
representation over another. In one paper Baraff briefly mentions that
position- and velocity- based constraints can also be expressed as
acceleration-based constraints, but doesn't indicate how (he has references
to other papers which I have not yet been able to obtain). 

Norman Lin