Visualising loss/objective function value with iterations using `Convex` + `SCS`

Specific Domains Optimization (Mathematical)
, ,
1

Anyone worked with Convex.jl + SCS.jl ?
How do I plot loss values/objective function values vs iteration during training process?

(Naive way, that comes to my mind is allowing verbose=true in the solver and copypasting first and second columns from printed training details…but not a good way if the colums are too -too long) Is there any better way?

2

Convex does not solve problems, it only formulates them. Whether you can do this depends on which solver you use, but for most solvers, and in general, there is no way to do this.

3

ahh yes, i meant using Convex.jl with SCS.jl as solver

4

I don’t think SCS supports logging the iterations.

1 Like