Quadratic Convex Constraints and Valid decomposition

[Sarmad Munir]

Hi,

I obtain sos solution by solving:

[sol,v,Q,res] =solvesos(constraints,objective,ops,params);

If the minimum eigenvalue of Q is smaller than the res, is it valid decomposition?

My second question is about solving quadratic convex constraints. I applied schur complement to the quadratic constraint, which is also a sos constraint. Sedumi and Mosek both give error when I use such a constraint. Can it be solved by solvesos?

[Johan Löfberg]

Read reference here

https://yalmip.github.io/example/moresos/

For second question, not enough info. What does "give error" mean? If it is a simple elementwise convex quadratic constraint, YALMIP will do the reformulations for you

[Sarmad Munir]

Thank you for your reply. 

I am optimizing over the coefficients of a polynomial and the constraint is quadratic in that polynomial. Sedumi says 'No sensible solution found' and mosek gives 'Infeasible Primal', which means I am not handling the quadratic constraint properly. It is a convex quadratic constraint so I was wondering if it is possible to solve such a problem with this approach.

[Johan Löfberg]

If you think it should be feasible, you must show us the code, as it currently appears clearly infeasible

[Sarmad Munir]

Thank you. The problem is sorted out.