models : change sensitivity of lambda approximation (claibration) to speed up the simulation (it really improved the speed !)