lsqnonlin/fminunc

Hello! I want to minimize the (norm of the) function (f(x) = (f1(x) f2(x)), size(f1)=size(f2) I tried it with fminunc(g), with g = sum(f1.^2) + sum(f2.^2), but it doesnt work very well. I think by nature the lsqnonlin algorithm could handle this problem in a better way , but i dont know how to write it down. lsqnonlin(h), with h = f1 + f2 produces an even worse result work, i think because then matlab computates sum((f1+f2)^2) Anyone has got an idea? Thanks