Why LMI toolbox/ Yalmip get a semi-positive definite matrix as the answer?

I　recently was do something about optimizing control. In my simulations, I already set the constraints with P>0, but the LMI toolbox/Yalmip returned me a answer as following:

 P=[21.0930    0.0101  -21.0926   -0.0100;
    0.0101    0.1467   -0.0100   -0.1467;
  -21.0926   -0.0100   21.0922    0.0100;
   -0.0100   -0.1467    0.0100    0.1467]

while eig(P)=[ 0.0000 0.0000 0.2934 42.1852]. As you can see, there are two zero eigenvalues of this P, it's not positive definite!

Could anybody explain this to me??? Many thanks!!!

[iqra masroor]

Mam can u plz tell me,which versions of MATLAB support the LMI toolbox?

Answer 1

[Johan Löfberg]

Strict inequalities are not really supported

http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Blog.Prepare-your-code

BTW, you get much faster response on

https://groups.google.com/forum/?fromgroups#!forum/yalmip