Closed lentinj closed 10 months ago
Sounds reasonable to me. I'm not even sure where sparse matrices would be useful, unless we are dealing with huge datasets.
I can't remember if I added it as part of an attempt of likelihood observations, landing data, or "just because". Regardless, it was probably only used for a week and then lurked unused for the rest of the time.
The closest I can think of to needing sparse matrices is the parameter tables, but std::map
fills our needs much better.
We depend on Matrix for an R equivalent sparseMatrix support.
However, this has not been used for a long time, is not tested, and thus invariably broken.
Whilst it sounds useful, IIRC the performance hit wasn't worth it.