MATPOWER’s MIPS solver is a very basic vanilla primal-dual interior point
method. On the other hand, fmincon provides a number of different algorithms,
including a primal-dual interior point method, which is the one chosen by
default in MATPOWER when selecting fmincon. I suspect that its implementation
includes numerous “enhancements” to the very vanilla implementation used by
MIPS. These may be helpful in some circumstances for a general nonlinear
solver, but in my experience with solving OPFs do not often improve either
performance or robustness over MIPS. I’m sure there are certain instances in
which they do make a positive difference. The Ipopt and Knitro solvers are also
based on interior point method, and as with fmincon, I’m sure they include
enhancements to improve performance and robustness in general, which may or may
not help in any given OPF problem.
I’m sorry I can’t give you the precise details of the differences because I do
not have access to the details of the implementations of these other solvers.
Ray
> On May 2, 2018, at 9:56 AM, Viswanath Hariharan <[email protected]> wrote:
>
> Hi
>
> I was wondering if anyone could explain the mathematical difference between
> MATLAB's fmincon solver to solve ACOPF and MATPOWER's runopf function/mips to
> solve the ACOPF problem. If both of them are going to find the local minima,
> what is the mathematical difference between the both?
> Thank you.
>
> Regards
> Viswanath
