JuMP - NLopt - How to set local optimizer


I am formulating an NLP in JuMP and would like to try out different solvers using NLopt. Some of them require setting a local optimization algorithm. How do I do that in JuMP?

To be more specific, say I have:
m = Model(solver=NLoptSolver(algorithm=:AUG_LAG))

On the NLopt_algorithm page it says I should use ‘nlopt_set_local_optimizer’. However, I do not see any field in m.solver where I could specify a local optimizer.

Thanks already in advance!


It’s quite likely not implemented. Pull requests are welcome. Have a look here.