[ODE] Running ODE backwards in time?

[Marc Toussaint]

Thank you for your comments!

Actually I thought, friction-less physics should be time-invertable -- CTP  
(charge-time-parity) symmetry..? And friction is only an external  
(collision-based) feature of ODE.

-- Jon, what kind of singularities might occur with more than two body  
interactions? It would help me a lot being aware of something like that -  
Thanks for your help!

-- Adam, your matlab example uses a different code for forward and  
backward simulation only because of the finite time steps and plain Euler  
integration, right? Actually, I thought, that ODE should make something  
smarter than Euler integration, maybe Runge-Kutta 4th order or  
something...?

Anyway, since I desparately need to do some inverse simulations (and my  
system is, I hope, simple enough to have no singularities) I'll just make  
a hack to invert velocities/accelerations and hope for the best... (Maybe,  
I'll try also a special collision handling to mimic (as good as possible)  
inverse friction...)

Cheers!
Marc.

-- 
http://homepages.inf.ed.ac.uk/mtoussai/