No abstract

The tting of data that contains problem measurements and the possible improvement in the ts in situations in which the measured quantities satisfy auxiliary di erential constraints is discussed. Problem data is de ned to include 1 cases in which the data may contain poorly measured quantities but exact constraints must be satis ed and 2 cases in which random external perturbations, such a s m ultiple scattering, may locally violate the constraint equations. Algorithms with a direct physical interpretation are developed that treat these situations. Applications to the measurement of magnetic elds and to detector surveying by using track data will be presented.