non problem is unclear

I don’t understand all that, and don’t know what the dimensions are. But I don’t see how (13e) is convex. `W`

is input data in (13e) per the SCA, despite being listed as an optimization variable. But that still leaves terms of the form `matrix_expression*matrix_expression'`

, which would seem to be a (non-convex) matrix expression of optimization variable `M`

. And all that stuff <= some other matrix. is that element-wise? I have no idea what that is really supposed to be, but you need to know exactly what it all is. If that mess is convex, there appears to be stuff going on which I don’t understand. it’s obvious that the author deviated from normal conventions by listing `W`

as an optimization variable in (13) when it is actually input data in (13); so I have no idea what other non-standard assumptions, conventions, and notation the author used, which perhaps somehow make (13e) convex.

Bottom line, it is incumbent on you to understand exactly what the convex optimization model you want to solve in CVX is. Forum readers can’t be expected to do so for complicated models.

It needs to be convex, which per my understanding of the problem it is not. Or is it a semidefinite constraint?

But as I wrote, i don’ necessarily understand what the author intended. but you need to.