How to create a standalone `JuMP.Containers.SparseAxisArray{Float64, n, Tuple{Int, ...}}`?

Specific Domains Optimization (Mathematical)
,
1

I’m interested in the X per se, as a container of a set of Float64 numbers (not depending on some JuMP variable x built beforehand).

julia> import JuMP, Gurobi

julia> model = JuMP.direct_model(Gurobi.Optimizer());

julia> x = JuMP.@variable(model, [i = 1:3, j = i:3], lower_bound = i-j)
JuMP.Containers.SparseAxisArray{JuMP.VariableRef, 2, Tuple{Int64, Int64}} with 6 entries:
  [1, 1]  =  _[1]
  [1, 2]  =  _[2]
  [1, 3]  =  _[3]
  [2, 2]  =  _[4]
  [2, 3]  =  _[5]
  [3, 3]  =  _[6]

julia> JuMP.set_silent(model); JuMP.optimize!(model)

julia> X = JuMP.value.(x)
JuMP.Containers.SparseAxisArray{Float64, 2, Tuple{Int64, Int64}} with 6 entries:
  [1, 1]  =  0.0
  [1, 2]  =  -1.0
  [1, 3]  =  -2.0
  [2, 2]  =  0.0
  [2, 3]  =  -1.0
  [3, 3]  =  0.0

I wonder how could I build such an object easily, e.g. via some syntax resembling [i-j for i=1:3, j=i:3]? Or I’m also okay with standard instatiating procedure such as

X = Vector{Float64}(undef, 6) # instantiate
for i = 1:6 # fill in the entries with concrete numbers
   X[i] = rand()
end
2

Appears Dict((i, j) => i-j for i=1:3 for j=i:3) can do :sweat_smile:
I’m surprised