Warning: recomputing forward solution; please report this script on launchpad

[Martin Sandve Alnæs]

I get this warning:
  Warning: recomputing forward solution; please report this script on launchpad
the script is spread across a few python modules and a bit of work to reduce,
is there some particular information you need to understand this issue?

[Patrick Farrell (pefarrell)]

One of the most important/difficult/complicated features of libadjoint is the handling of forward state: you want to forget as much as you can as soon as you can, but no earlier. The algorithm that's there does a pretty good job, but I added that warning message to trap a case that I hadn't dealt with: where the computation of the adjoint solution doesn't require a particular forward variable (and so it's forgotten), but the computation of the \partial F/\partial m term does. If not solved, it will lead to a significant inefficiency, unfortunately, so I'd really like to deal with it. Dealing with it involves careful tracing of exactly when things are computed, stored, and forgotten, and requires incense, self-flagellation and the occasional sacrificial goat.

I'll be at Simula from 13 May so maybe you can show me then?

[Martin Sandve Alnæs (martinal)]

Ok, I'll see if I can find a goat, they're kinda hard to come by in Oslo.

[Martin Sandve Alnæs (martinal)]

Btw, I get the warning 5 times for each replay, and I also get 14 callbacks to eval_cb and derivative_cb within a single iteration, shouldn't that happen only once? I have 4 Functions/InitialConditionParameters and 5 ScalarParameters/Constants as controls.

[Martin Sandve Alnæs (martinal)]

Actually, I only get the warning if I do a superfluous solve before the timestep, which was a bug in my code.