How to apply a forcing term ?

[ReHoss]

Hi, I would like to apply control on PDEs using your library.

To do so, I would like to apply a forcing term to the pde for a small dt.

A workaround I just found is to redefine at each control time step the evolution_rate method with the new computed forcing term in the RHS and then integrate again for a small dt ?

Thank you for your consideration,

EDIT: Sorry it might be better to move this issue into the discussion section...

[david-zwicker]

I'm not sure what you're trying to achieve. If you use the python/numpy interface, your evolution_rate method can of course depend on additional parameters. If you also want to use the numba interface things are more tricky since we currently do not support external parameters.

[ReHoss]

I'd like to use the python interface to implement d_x(t)/dt = F(x(t)) + u(t), with F = KPZ for instance. Where u(t) is recomputed every dt by a neural network for instance (feedback control).

[david-zwicker]

What prevents you from simply implementing $u(t)$ in evolution_rate?