Acceptance Correction • • • Use g3leps with known SDM’s in the event generators to generate ntuples. Filtering the ntuples with standard analysis codes used for real data and put into the weights of acceptance correction. Check whether we are able to extract the input SDM’s. Experience: 1. ML fitting framework is robust with the variations of starting values of SDM’s in fitting if there is no acceptance effect. 2. ML fitting suffers less from the statistics than from the acceptance. 3. MC tests only succeed for Egamma>2.2 and t>-0.2 where the experiment acceptance is close to be of full coverage. 4. The weighting scheme in ML fitting is not effective as expected. Need to be checked.