Solved to optimlity but status is infeasible

Specific Domains Optimization (Mathematical)
1

Hello there

    println("=================================================")
    println(" Optimal Solution Status: ", termination_status(sub))
    println(" Optimal Value: ", objective_value(sub))
    println("=================================================\n")

: Optimal Solution Status: INFEASIBLE
: Optimal Value: 270.0

How comes that the problem is infeasible but it has an objective value?
Thanks

1 Like
2

You can usually still evaluate the objective of an NLP even if the constraints are violated.

3

In convex optimization, I’ve encountered situations where an optimal solution is found with high accuracy, while rounding errors lead to a false report that the problem is infeasible.

1 Like
4

This situation is not rare. If an inequality constraint is active, this can happen.

1 Like
5

Hi @ShGE, welcome to the forum.

You can find the JuMP documentation on how to query and interpret solutions at Solutions · JuMP.

A useful command is solution_summary(sub) which will provide more information about what happened.

It will likely tell you that the primal solution is an INFEASIBLE_POINT.

1 Like