Maximum iterations or time limit exceeded

[jarosz]

Hi,

I've got a following problem, when running my script I got an error:

ans =
Maximum iterations or time limit exceeded (SDPA-M)
ans =
maximum iterations or time limit exceeded

I tried to deal with the problem using 'sdpa.maxit' in sdpsettings() but nothing happened. BTW, in my problem I use a semidefinite matrix constraint and I was thinking if a condition "X>=0" is equal to "X is positive semidefinite"?

Thanks for any help!

[Johan Löfberg]

So did you increase it enough (time and iteration limit, which ever hit the limit)

Works here

X = sdpvar(10);
optimize([X>=0, X(1,3) == 9],[],sdpsettings('solver','sdpa','sdpa.maxit',1))

SDPA start at [Wed Jan 28 08:47:55 2015]
NumThreads  is set as 2
Schur computation : DENSE
   mu      thetaP  thetaD  objP      objD      alphaP  alphaD  beta
 0 1.0e+004 1.0e+000 1.0e+000 -0.00e+000 -0.00e+000 8.9e-001 8.6e-001 2.00e-001
maxIteration is reached 1 1.8e+003 1.1e-001 1.4e-001 -0.00e+000 +6.98e+001 8.9e-001 8.6e-001 2.00e-001

and yes X>=0 means psd
http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Tutorials.Basics

[jarosz]

Thanks for the swift reply! I changed my code a little bit and there's a tiny progress as all the script compiles now. However, it seems as an sdp condition (condition C47 to be specific) is totally ignored by a solver beacuse when I delete one, a solution doesn't change.

[Johan Löfberg]

So is C47 violated when you don't include it? If it is not violated, than it should make no difference if you add it, i.e., it is redundant.

I cannot run the code

solvesdp([C43, C44, C45, C46, C47, C48, C49], -objective)%,sdpsettings('solver','sdpa','sdpa.maxiter',1000))
Undefined function 'funkcja_pom' for input arguments of
type 'sdpvar'.
 

[jarosz]

Oh, I'm really sorry, here's the missing file. I'm not sure about redundancy as I checked semidifinity manually in the end of the script and one is not met.

[Johan Löfberg]

Are you sure you sent the correct models. It is claimed infeasible with all solvers I tried

[jarosz]

Yeah I know... Can I share with you a math formulation of the problem so you can check if my matlab statements (syntax) are correct?

[Johan Löfberg]

So why did you ask the original question then if you knew the model is infeasible to begin with?