finding P value solving a set of LMI

[juseung Lee]

hello,

i have some questions about finding  a value of P,

i have some difficulty to use YALMIP tool because i am not familiar with this tool.

idon't know why P value (when i use 'sedumi') is different from that when i use 'lmilab' as a solver  

this is for 'sedumi' i 've programmed

A=[0 1; -1 -1];
n = length(A);

P=sdpvar(n,n);

constr=set(P>0)+set(A'*P+P*A<0);

opts=sdpsettings('solver','sedumi');
yalmipdiagnostics=solvesdp(constr,[],opts);

yalmipdiagnostics;
Psol=double(P);

this is for 'lmilab' i 've programmed

A=[0 1; -1 -1];
n = length(A);

P=sdpvar(n,n);

constr=set(P>0)+set(A'*P+P*A<0);

opts=sdpsettings('solver','lmilab');
yalmipdiagnostics=solvesdp(constr,[],opts);

yalmipdiagnostics;
Psol=double(P);

as you see, i just changed tag 'sedumi' to 'lmilab'

what should i do something for obtaining right value of P ?

Thanks in advance

[Johan Löfberg]

First, don't use SET
http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Commands.set

Second, don't use strict inequalities
http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Blog.Prepare-your-code

Third, why would you expect that you would get the same solution? There are infinitely many solutions to the problem

Fourth: Don't use LMILAB
http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Solvers.LMILAB

Fifth, Your problem is ill-posed. 0 is a feasible solution for a solver which works with finite precision and numerical tolerances