Sam Isaacson

This should work for `VariableRateJump`s. Have you tried it?

I updated my earlier comment to an example that has an explicit time dependence in the rate. Hopefully that makes it more clear that this should work for `VariableRateJump` problems.

It can be closed in the sense that there is a work around, but there isn’t yet a nice interface for this case.

The generic `VariableRateJump` solvers via ODE timesteppers require floating point values (since ODE solvers don't work with integers). They use the ODE solvers to integrate the intensities (hence why they...

@TorkelE getting into the various saving options and controls would probably make a nice tutorial too, especially combining with `EnsembleProblem` to handle saving for each trajectory.

Sounds good to me! Our current setup is just historical, based on the older testing style. I agree it would be nice to have a self-contained testing environment one can...

One issue that has to be considered is if the random parameter is only supposed to be sampled once at the start of a simulation, and then it is fixed,...

I don't see why a user might not want to have such terms in rates, or coupled equations either potentially. And having a random stoichiometric coefficient fixed at the start...

There are also updating QR factorization versions of Broyden I believe (instead of updating the inverse Jacobian approximation), but I'm not sure they can be made to work for sparse...

https://epubs.siam.org/doi/book/10.1137/1.9781611971200 Section 8.3 references the updating QR I believe. (However, I doubt this is the best reference for immediately seeing a pseudo-code, there are probably better ones.) I had searched...