Too many nonzeroes for NLP in JuMP and Examodels

Specific Domains Optimization (Mathematical)
7

If ExaModels.jl does not converge im sure they would like a bug report.

And yes, JuMP is slower, which is why we wrote MathOptSymbolicAD.jl but the choice of AD backend should not change convergence.

You’ve probably already seen it, but see

1 Like