let’s assume I have multiple 2D arrays (maps, tables…), each row of those belongs to a single design in a Pareto set (objective & parameter values).

I want to optimize on the combinations of those, therefore (I assume) I want to be able to index their rows with a discrete JuMP variable, so I can refer to them in the constraints & objective.

Do you think it’s possible/a reasonable thing to do?

model = Model()
@variable(model, x)
data = [1, 2, 3]
@objective(model, data[x]) # Can't index with x.

Instead, you should do something like the following:

model = Model()
data = [1, 2, 3]
@variable(model, x[1:length(data)], Bin)
@constraint(model, sum(x) == 1)
@objective(model, sum(x[i] * data[i] for i in 1:length(data))

Here, the binary variables x choose each element out of data, and there is a constraint that you can only choose one element.