Mateusz Baran

Manopt currently lacks a solver structure that fits here. State is supposed to be constructed when you have at least a bit of information about the problem, and the only...

> I do agree that the `solver_name´ (high-level) functions were always meant to be the ones used by end users, but could you elaborate on > > > Manopt currently...

Ideally we should have manifold in only one place, either objective or solver. You suggested you prefer manifold to be in the objective. So we can't use Manopt states to...

Out of curiosity, how is one supposed to supply the function for the group exponential?

Nice. I'll try integrating it with the Lie group functionality of Manifolds.jl.

Sure, I'll let you know if I encounter any problems.

Could you always use calls like `expv(dt, L, u)` when performing a step for Lie group integrators?Converting the operator to matrix for all non-Krylov methods seems restrictive. Similarly I'd like...

So the only consequence of setting `krylov` to `true` here: https://github.com/SciML/OrdinaryDiffEq.jl/pull/1090/files#diff-9d28f892ddd67f73e3e8d9cc88442486R50 is that `expv` will be used? Does it have any other effects? I'm not particularly familiar with `DifferentialEquations.jl` yet.

Thank you for this work, I will try to help finish this. I need to clarify a couple of things first. If I understand correctly, adjoint_action for the left action...

Thanks for clarification, I should have some time tomorrow to work on this PR.