we'll satisfy ourselves with a finite difference approximation of helm_resid_deldeldel for the moment, and move on with the saturation curve problem.