I wonder if a similar corrrective action happens whenever a Newton-iteration exception occurs: in this case, does the solver simply reduce the stepsize for the following integration step?
Not for the following, but for the current because the current step has to be repeated.
Or is there any other specific action?
Reduction of the integration order of the integrator if variable order is possible (like for BDF solvers or some Runge-Kutta codes).
Does this also depend on the type of Newton-iteration exception? (depending on the solver used, "DAE" Newton Iteration exceptions or "ODE" Newton iteration exceptions could occur)
I don't think it will be possible for the solver to decide whether the failure was due to the algebraic or the differential part of the DAE system. So no.
