Setting heuristic solution involving symmetric matrix in JuMP callback

Specific Domains Optimization (Mathematical)
3

Thanks @mike_k ! Yes, this would work for this simple example, but I am not sure if it will extend to other container types in https://jump.dev/JuMP.jl/stable/manual/containers/#Containers, as eachindex(.) seems like the unified way to loop over a generic container.

For example, consider the more complicated case (this particular code snippet runs fine, because I have used the vec operation for Z), where each variable is of a different type:

JuMP_variables = vcat(
     # λ is a dense axis array over a dictionary of custom types
     [BnB_PEP_model[:λ][i_j_λ] for i_j_λ in eachindex(BnB_PEP_model[:λ])],
     # ν is a scalar
     BnB_PEP_model[:ν],
     # Z is a symmetric matrix
     vec([BnB_PEP_model[:Z][i_j] for i_j in eachindex(BnB_PEP_model[:Z])]),
     # α is sparse axis array that ranges over (i = 1:5, j in 1:i-1)
     [BnB_PEP_model[:α][i_j] for i_j in eachindex(BnB_PEP_model[:α]) ]
)

In this particular case, your suggestion will not extend to \lambda and \alpha. I was wondering if there is a unified way to construct these vcat objects over different types of containers?