Nonlinear Polynomial Optimization

60 views
Skip to first unread message

Asad

unread,
Apr 12, 2014, 2:38:29 PM4/12/14
to yal...@googlegroups.com
I am trying to solve this constrained optimization problem using yalmip,


sdpvar x y gamma1;
p1 = (1+x*y)^2-x*y+(1-y)^2;
p2 =  x^2+y^2;
[S1,c1,v1] = polynomial([x y],2);
[S2,c2,v2] = polynomial([x y],2);
[S3,c3,v3] = polynomial([x y],2);


F =[sos(-S1*(gamma1-p1)- S2*(p2-10) - S3*(p2-100)*(gamma1-p1) - (p2-10)^2)];


solvesos(F,-gamma1,[],[gamma1;c1;c2;c3])
double(gamma1)





---I got this message when i ran it,
solvertime: 0
          info: 'No suitable solver'
       problem: -2
    yalmiptime: 0.7860
ans =
   NaN

After searching for the problem i came to know that i needed "SNOPT". I have downloaded it but i dont know how to use it in Yalmip?


asad

Johan Löfberg

unread,
Apr 12, 2014, 2:56:29 PM4/12/14
to
You are multiplying decision variables (gamma1 and c1, etc), so you have a nonconvex semidefinite program. It is typically intractable. Sure local solverpenlab, and/or the global solver bmibnb built-in YALMIP "solves" polynomial SDPs, but the probability of this to work is very low.

You have to derive a linear SOS problem, or devise your own solution strategy (bisection in gamma1 would be pretty simple and would probably work fine)
Reply all
Reply to author
Forward
0 new messages