The optimization model caused the problem? Or the solvers caused the problem?

[Kris]

Dear Prof Lofberg:

Hi, thanks for giving me advices about why Gurobi and Cplex gave me totally different ans last time.

After that, I checked my code several times to make sure it is right, however, I found no thing wrong.

Last night, when I was browsing Yalmip forum, one of your ans interested me " Gurobi needed long time to solve a problem, offering a good ans. Cplex solved the problem quickly, offering a unreasonable ans, Mosek solved the problem quickly and perfectly ,in terms of same Soc optimization ".

Above all, I guess the main reason caused my Soc optimization can not be solved is the performance of the solver.

I am not at school, therefore I  can not obtain academic Mosek immediately. Therefore, can you run my code with your Mosek and tell me the performance compared to Gurobi and Cplex?

I would like be appreciated it if you can give me a hand.

Best wishes

Kris

[Johan Löfberg]

Your model is simply unbounded, and that is easily detected by all solvers I tested

% We can generate arbitrarily good revenue, and it does so by letting beita(1) tend to -infinity

optimize([constraint,Revenue >= -1e9],Revenue);value(Revenue)

>> value(Revenue)

ans =

  -1.0000e+09

>> value(beita)

ans =

   1.0e+09 *

   -3.1085

    0.3066

    0.2243

    0.1372

    0.1269

    0.1806

    0.0023

    0.0020

    0.0023

    0.0021

    0.0023

    0.0021

[Kris]

Thanks Prof! I will check more.