[Convex.jl] Objective function of matrix factorization with missing data

Thanks for the update! The memory consumption is unfortunate, but I’m not terribly surprised as I’ve seen similar reports on the issue tracker before. Can you open an issue on Convex.jl with the full code (and fake data, eg via rand)? I probably can’t help anytime soon but I have been thinking about an internal restructuring to improve Convex.jl’s performance and it would be great to have another test case.

In the meantime, one option is to just use PyCall or MATLAB.jl to use those libraries from Julia, or to try JuMP.jl.

At the time I forgot to add a link here to the Github issue, in case someone wants to read the follow up.

1 Like