This is a mixedinteger convex programming (MICP). I want to know how the sum in the optimization problem is represented, and should my optimization variable be set to a matrix? I hope you can give me some advice, thank you very much！

the xknt is binary variable.

You can declare something such as:

`variable r(I,N,T)`

or whatever the appropriate dimensions are. Then it’s just a matter of getting the indices correct for the summations, for which you can use `sum`

, the same as it would be in “regular” MATLAB Of course this presumes that the \sigma^2 difference terms evaluate to nonnegative.

I write the left-hand side of 4b and 4c as follows

But it doesn’t work

Could you give me some suggestions for modification. Thank you very much!

A model that has nonlinear equalities cannot be convex.

In this case, those are assignments, not constraints. So there are no nonlinear equalities.

I believe the cause of the error must be that not all of the sum of \sigma^2 difference terms are nonnegative. Hence those elements of `ps`

are concave. And some of those sum of \sigma^2 difference terms are positive, which makes those elements of `ps`

convex. You will need all those terms to be convex. Note the statement in my previous post “Of course this presumes that the \sigma^2 difference terms evaluate to nonnegative.”.

So you will need to change the input data so that all of the sum of \sigma^2 difference terms are nonnegative.

Thank you very much! I solved it!