Can cvx resolve this issue properly?

Because I heard that cvx can not solve problems with larger constraints !!!

I’ve implemented this but I’m not responding

If cvx can not resolve, please guide

Thanks

# Because I heard that cvx can not solve problems with larger constraints!

**qeraatimohammd**(mohammad) #1

**Mark_L_Stone**(Mark L. Stone) #2

I don;t understand what the decision variables are, for instance, with respect to use of subscripts or not. If \tau_i and P_u are both decision variables, then the objective is non-convex and CVX can not be used on this problem.

If you have produced CVX code, please show us your code, and preferably provide a reproducible example (as Preformatted text), complete with all input data required for forum readers to copy, paste, and run it. Also show us all solver and CVX output from running it (do not use the quiet option on cvx_begin).

**qeraatimohammd**(mohammad) #3

Hello

Yes, it’s true that the problem is not convex

But, by fixing the two variables (u) P and (tau) p and updating them, we first compute the tau parameter and then calculate the two parameters (u) P and (tau) p

With this method, the problem is convex

But my problem is that cvx does not correctly calculate tau

And as far as I’m researching, cvx has errors in solving problems with more (larger) inequalities.

Now tell you that this is true or not?

Please contact us for more information

Thanks.

**Mark_L_Stone**(Mark L. Stone) #4

You can use an alternating optimization algorithm as you describe. To do so, you need a loop which includes successive cvx_begin … cvx_end blocks. In one of them, \tau is declared a CVX variable and P is not declared in CVX. In the other, P i is declared a CVX variable and \tau is not declared in CVX. In general, there is no guarantee that such an algorithm will converge to anything, let alone a local minimum, let alone a global minimum. But you are free to use CVX in such a manner, i.e., to solve convex subproblems within an overall non-convex optimization algorithm.

If you have more specific questions, you need to show your code, preferably with all input data, and show the solver and CVX output, plus any demonstration if you think CVX ius not performing properly. Keep in mind, that the alternating optimization algorithm may fail to find a global or local minimum, even if CVX correctly solves every problem presented to it.

You wrote:

And as far as I’m researching, cvx has errors in solving problems with more (larger) inequalities.

What specific research and errors are you referring to? Please note that CVX 3.0beta is known to have many bugs. But CVX 2.1 does not have many known bugs. if you find what you believe to be an error when using CVX 2.1, please provide a reproducible problem if possible, together with all solver and CVX output from running it, and any discussion and evidence you have as to what errors were made by CVX.

You wrote:

Because I heard that cvx can not solve problems with larger constraints !!!

What specifically did you hear from what source? Not all information you hear is correct, and in many cases, true statements are misinterpreted by other people, and pretty soon, people are told and believe things which are far removed from the truth (the game “telephone”).