Add Cofunction as possible input to Projector

[colinjcotter]

I’m thinking of adding support for v to be a Cofunction in Projector. This would allow easy implementation of explicit DG schemes using the Slate option for the inverse mass matrix. I could then assemble the DG RHS into a cofunction and use the projector to get the result of the RK stage.

Rationale is that I would otherwise be repeating a lot of that code. Just sounding out, if there is not strong objections then I will implement.

[connorjward]

It might be good to flag this either on Slack or bring to the weekly meeting.

[dham]

Isn't this just duplicating the riesz_representation() method of Cofunction?

[colinjcotter]

It is, so maybe it is better to add SLATE options to riesz_representation(). The problem is that we end up duplicating code either way.

[dham]

Well the really nice thing to do is probably to add Cofunction to Projector then replace the implementation of riesz_representation with a call to project thereby removing all code duplication.