cossio
cossio
Even if the implementation is not top performance, I think it still makes sense to add it now, and then later on tackle performance improvements.
@jonas-eschle One of the two implementations has gradient? I think the gradient is pretty crucial (if I didn't want gradient I'd just call out scipy or something)
Indeed the gradient is given by the same Faddeeva function:  Hopefully it's not difficult to add.
I would really like to have this feature as well. Unfortunately it seems that the mentioned PR https://github.com/zulip/zulip-desktop/pull/1059 is stale. Are there updates happening somewhere else?
@ararslan I'm now seeing the memory spike in the solve phase instead of the formulation. I'll close this unless something similar comes up again. I'll open an issue at ECOS...
@ararslan I just tested with ECOS and SCS, both consume > 4GB RAM (at which point I cancelled to avoid having to reboot my laptop again)
@ararslan Problem is my laptop dies before I can complete the profiling. Any suggestions?
I just tried limiting to 10GB ram at a server and it also crashes. Both SCS and ECOS. I don't know how to do the `Task` thing, and I'm really...
@ararslan I started julia with a mem constrain of 10 GB using: `timeout -m 10000000 julia --track-allocation=user` This tracked memory allocations across all of Convex.jl as the RAM usage increased,...