Closed 4er4er4er closed 1 year ago
When gurobi is given a model with no objective, it reports optimal solution but says nothing about the objective value:
optimal solution
ampl: model diet.mod; ampl: data diet.dat; ampl: drop Total_Cost; ampl: option solver gurobi; ampl: solve; Gurobi 10.0.0: optimal solution 3 simplex iterations ampl:
Perhaps the user should be alerted that there was no objective, and the solver just looked for a feasible point. The amplasl output sort of does that:
ampl: model diet.mod; ampl: data diet.dat; ampl: drop Total_Cost; ampl: option solver gurobiasl; ampl: solve; Gurobi 10.0.0: optimal solution; objective 0 3 simplex iterations Objective = find a feasible point. ampl:
(Also when the solution is found without having to iterate at all, either Gurobi version neglects to report 0 simplex iterations.)
0 simplex iterations
These observations might apply to other MP-based solvers, but I haven't tested them.
'Find a feasible point' is printed by AMPL based on the objno in .sol, which is currently wrong in MP, unless solved with objno=0.
There is a lack of acknowledgement of multiobjective solving, see AMPL's #92.
When gurobi is given a model with no objective, it reports
optimal solution
but says nothing about the objective value:Perhaps the user should be alerted that there was no objective, and the solver just looked for a feasible point. The amplasl output sort of does that:
(Also when the solution is found without having to iterate at all, either Gurobi version neglects to report
0 simplex iterations
.)These observations might apply to other MP-based solvers, but I haven't tested them.