How to solve this problem with cvx

Uncategorized
Jun 24, 2014
J

My convex optimisation problem is of the form minimize z in min \,tr\left\lbrace (H_{M}^{\dagger}H_{M})^{-1}Z\right\rbrace
subject to the constraints:
tr\left\lbrace (H_{E}^{\dagger}H_{E})^{-1}Z\right\rbrace >=\gamma and
tr\left\lbrace Z^{-1}\right\rbrace <= P_{avg}
Here H_{M},H_{E},z are 2x2 matrices. \gamma and P_{avg} are scalars and are known… values of H_{M},H_{E} are also known…
It is also given z is positive definite…
Can someone suggest me how to solve this problem in cvx? Is it possible to solve this problem by using optimisation toolbox in matlab?
Thanks in advance

M

What have you tried so far? This forum is not a substitute for reading the manual and trying out the software. The trace and trace_inv commands are both relevant here, as is SDP mode. And it’s a support forum for CVX and TFOCS, not the optimization toolbox.

J

Sorry for my mistake. I am new to this field. Is it possible to solve this problem in cvx? I am confused because here my optimisation variable z is a 2x2 matrix. z is positive definite also.

M

Leaving aside other details of the problem, optimization variables which are matrices constrained to be positive definite are right up CVX’s alley.

M

Right. I suggest that you read the user guide, particularly the function reference, and look at some examples. I am sure you will find that CVX is a good tool for this task. But we’re not a model building service here; we expect you to do the work!