Can someone shed light on the way coordinates errors are obtained in REMARK 3 using maximum of likelihood as compared to Luzzati or sigmaa? What article can be cited in this respect?