how to setup an optimization faster?

Hi,

I am using 'fmincon' to solve an optimization problem for feedback control. I need to solve it repeatedly at every sampling time. The optimization problems are the same except that only a few parameters' value change.

Currently, I use 'fmincon' to solve it. And I have to define the optimization problem completely at every sampling time. It is quite time consuming. I guess the time to build the optimization problem is even longer than the time to solve it.

I am wondering is there a smart method to program, such that the time spend on define the optimization problem is less.

I am also trying to write the code in CVX, a convex optimization package. So if anyone know how to do that in CVX, it would be great as well.

Any suggestions are welcome. Thanks a lot.

Cheers Elvin