Daniel R. Reynolds

@ciropom: you will need to emulate the code from `cvLsDenseDQJac` ([here](https://github.com/LLNL/sundials/blob/main/src/cvode/cvode_ls.c#L1034)), but ensuring that each of the operations use device-resident vector data, and that these operations fill device-resident data within...

Your hypothesis seems reasonable to me. This might be simple enough to test (add `SUNDIALS_EXPORT` to these functions and try to rebuild on Windows).

Dear Bill, FCVODE (and all of the other F77 interfaces) are not threadsafe, nor can they be (the F77-C "glue" code relies on global variables). This was the main motivation...

The majority of memory used in SUNDIALS comes from the vectors themselves. As such, in collaborating with applications we have frequently implemented custom vectors that get their memory from a...

I think I understand. Currently, SUNDIALS allocates all of its Krylov subspace vectors at initialization to improve performance (as compared to repeated allocation/deallocation). You worry that this unnecessarily increases the...

Got it. I'll discuss this with the rest of the team. I don't recall us ever discussing this before, but I agree that it could certainly help in memory-limited environments...

Thanks for bringing this up @ianabel. @balos1 is on travel this week, but hopefully he'll chime in soon.

> > how much testing with various RK table options did you do before settling on these specific default tables? > > I specifically chose these method due to there...

> Overall, I this looks great and it will make creating a stepper significantly easier. I think most or all of my requests are very minor changes. Thanks for all...

Revised plan: 1. Once everyone is happy with this PR as-is, we can go ahead and merge. 2. In a very-easy-to-review subsequent PR, I'll adjust the order of the contents...