@Carl Love 

I went back to my old expreesion, which should be right. This is derived by hand but still, its just a few terms with some coefficents. After solveing for the coefficents I try the same thing again but now I get an error message using GlobalOptima and Search.

I Execute;

GlobalOptima(L(A, B, C, D, E, F, G, H), contstr);
Error, (in DirectSearch:-Search) number of parameters in procedure uneval(D) must coincides with number of variables 8

where L is just an the quotient of two polynomials in 8 variables.

@Carl Love 

I started to use it and it gives me a negative value even tho my expression comes from a norm. Im gonna have to go over my simplification of the this norm again and retry tomorrow. This looks like a really nice "add-on" tho. Thanks alot!