--- Begin Message ---
> |Δε|*|σ| sounds straightforward, but with du and df, and non-linear
> elasticty on both normal and shear, and plasticity for fun, I have no
> initial guess...
dammit, I forgot stored elastic energy ;-)
> On the top of that, there is the case when plasticity==true at time
> "t", and plasticity==false at time "t+dt". You have to decompose |Δε|
> into plastic and elastic on one timestep.
If I don't miss something: plastic deformation changes discretely (by
you, in the functor), and the energy that is dissipated doesn't depend
on time. So that computation is stepwise, but exact. What dissipates, is
added to plastic work; what doesn't, stays stored in elastic energy.
Nothing left out.
--- End Message ---
