[R-sig-dyn-mod] Hybrid simulation of nonlinear ode and discrete state machine

[Eike Petersen]

Dear fellow R users,

I am currently interested in simulating a nonlinear ODE system which has 
an input function that is described by a finite state machine (FSM). 
Conversely, the state transitions of the FSM are depending on the state 
of the dynamical system. This is hence a hybrid, discrete-continuous 
problem. I have the impression that for a linear ODE system this should 
be possible using e.g. the 'radau' solver from the 'deSolve' package and 
using the solver's root-finding capability to detect discrete state 
change events. However, none of the solvers that support root-finding 
(see deSolve manual, page 34) appears to be capable of solving nonlinear 
ODEs.

Am I missing something here, or is this currently not possible? If the 
latter is true, how hard would it be to implement root-finding 
capability for e.g. the 'daspk' solver? We might be willing to spend 
some effort on this problem. I am also open to suggestions for 
completely different approaches to this problem.

Apologies in advance if this has been asked before; I could not find any 
previous questions on this matter but am unsure whether I searched in 
the right corners.

Kind regards,
Eike