Bounded parameters w/ Optim

Specific Domains Optimization (Mathematical)
,
2

Using NaN is almost certainly going to cause issues (and I’m surprised that it worked), mainly because of this:

julia> 0 <= NaN
false

julia> NaN <= 0
false

julia> NaN == NaN
false

If you have constraints, you should checkout JuMP. Here’s an example of a NLP: https://jump.dev/JuMP.jl/stable/examples/mle/

1 Like