sammccallum

Right, turns out this is a slightly more involved bug! The Lévy area term $H_{s, t}$ isn't correctly handled when solving backwards-in-time too. I believe this is due to VBT...

Yeah, that's what I thought and it semi-works: flipping the sign of `H` after it is returned from `diffusion.contr` gets all the additive noise methods (`SEA`, `SRA1`, `ShARK`) back to...

I've also added the `Reversible` RK solvers here which just subclasses `AbstractReversibleAdjoint`. Let me know what you think of this and I can add some documentation when it's good to...

I think I've now addressed all of your comments, so it should be ready for review 👍 Understanding how to rebase through multiple merges was an experience but I believe...

(No pressure to review anytime soon Patrick, I just marked this as "review requested" so it's easy for you to see across your sprawling jax empire)