The two underlying assumptions for this procedure are the ergodicity and a self-averaging disorder, which imply essentially that the computed quantity 〈O〉¯ should match the measurement of the observable O on an equilibrated and sufficiently large experimental sample (here with quenched disorder)....