Scaling the objective by a positive constant changes the result a lot when using solvesos

[Pan Zhao]

Dear Johan,

Hope you are doing well. Recently, I created an SOS program to find the incremental gain of a nonlinear system, which is solved using solvesos. I noticed that scaling the objective by a positive constant changes the result a lot, which does not make sense to me. I attached my code. Can you help take a look?

Thank you. 

Best,

Pan

[Michal Adamaszek]

Mosek has convergence issues, so it is likely that the solution is inaccurate, just each time in a different way.

Michal

[Pan Zhao]

Thank you, Michal. How did you typically mitigate the convergence issues?

Pan

[Pan Zhao]

Johan, can you share your insights about this issue? Thank you.