Next: Understanding control file structure
Up: ana10 program
Previous: Create your own analysis
ana10 is able to perform automatic normalisation of MC simulation. It is quite complicated, so it is worth for users to understand how it is working in ordered to tackle possible problems.
The following assumptions are made beforehand:
- There are few sets of REAL data (usually 1, 2 at most). Main (large) set and additional sets (e.g. most recent data)
- These data sets have no common runs and more then 100 runs each.
- For each set of REAL data there is a set of MTCA simulations, reconstructed with corresponding runlist. There are MTCA of all background and signal components for main data set, and some crucial background and signal simulation for additional sets. For other background main set MTCA is used for efficiency estimation for data sets where MTCA is missing.
The normalisation procedure:
- For each MTCA event nemor stores the following information in h10 tree: ``run''= run number used for apparatus efficiency simulation during this event (- added to distinguish from REAL data); ``time''=NEVGA, number of events generated, ``date'' = total useful time for runlist in hours.
- During slimming, for each MTCA event slim2 updates the ``time'' variable to be equal to (-total NEVGA)/1000. Total NEVGA is sum of generated events in all MTCA files put into slim , reconstructed with the same runlist. Divider 1000 introduced to overcome 4000000000 limit for Int_t type, minus sign introduced to distinguish slimmed (divided by 1000) h10 and nemor h10.
- ana10 loops through each event and if it passed all selection criteria, fills histogram with the weight equal to (total useful time)/(total NEVGA). Total usefull time is coming from ``date'' variable, and total NEVGA from the ``time'' variable. See GetEventWeight() function for details.
- at the end, when setnorm() is called for bana10 class, each MC histogram is divided by total MC runlist time. It is calculated by MCTime() as a time of all runs between first and last run encounetered during this MC component loop. At this point each MC histogram is already normaliased to the MC efficiency.
- Next, MC histogram is multiplied by the running time of the analysed period.
- Finally MC histogram is multiplied by the component activity and become normalised to the number of expected MC events in the NEMO3 during the period analysed.
Note that this algorith is quite flexible, and allows to analyse subset of the runs during the period, periods longer than ones used during MC reconstruction, combine MC reconstructedfor different periods. The latter oprtion had not been much tested, so one should be carefull.
Next: Understanding control file structure
Up: ana10 program
Previous: Create your own analysis
Vladimir Vasiliev
2008-12-02