Closed jolars closed 2 years ago
I don't know if this is worth (in time) doing now. I think we could do a simulation study where we start from a previous starting value and see how it does there. What I mean is that we are comparing solvers, not path issue.
We should consider adding benchmarking for the full SLOPE path. We should also use the sequential strong screening rule strategy for SLOPE when doing so.