If you mean that you just want to find a feasible point without actually minimizing anything, you can provide H=zeros(N), f=zeros(N,1). This makes the quadratic objective constant (i.e. zero) over all x and therefore any feasible solution will be optimal.
It may be simpler, however, to apply this technique with linprog, rather than quadprog.
