Inverse subset: order 2 not yet implemented

[James Thorson]

Hi all,

I have a model that yields the following error using `framework = "TMBad"`:

but runs as expected using `framework = "CppAD"`.  

Does anyone have experience with a workaround to this in `TMBad`?

I'd be happy to provide an example if interested.

cheers,

Jim

[Kasper Kristensen]

A temporary fix is to disable this atomic, which is indirectly generated as the log determinant derivative:

TMB::config(tmbad.atomic_sparse_log_determinant = FALSE)