 |
 |
Hi all, I think I have to clarify this point once again. TS are done using 1) SemiExclUser to reconstruct the B. More than one B per event is allowed since we are not using the selection of the best B at Beta level. 2) CPFramework to process the output of SemiExclUser (hbook files). CPFramework put a tag on each reconstructed B (depending on the reconstructed B and D mode), and applies some additional cuts but above all selects the best B per event accordingly to the a-priori purity using the tables that are both in CPFramework (CPFramework/tables) and CompositionTools and they are IDENTICAL. We have TS for each seed (D*,Dc,D0,D*0). In principle you can have duplicates between different seeds but not in two different superblocks of the same seed just because of the way I implemented it in CPFramework. Any deviation from this shows a problem and could be fixed. Using TS we produce reduced collection. For Ibu side we do as following: 1) make kanga file from reduced collections 2) process them with SemiExclUser but this time with the selection of the best D and B switched on at Beta level. The tables used to select the best B are the same of CPFramework BUT the selection of the best D is not the same and it is done before selecting the best B. Then CPF and SemiExclUser with the selection of the best B ARE NOT EQUIVALENT. 3) then Ibu processes the event picking up the one and only one B coming from SemiExclUser and produces hbook files. Then CPF output and Ibu output could be different but the differences have to be small and you can have modes with purity lower than 80% even if you are processing superblock1 reduced collections. By the way if you notice any striking difference (for instance one mode with very low purity is present in superblock 1 root files in many cases) this could mean that the tables have a bug. Other comments: - about the duplicates, are you applying the selction of the best B after cutting on mes (mes > 5.27)? if so, you could bias the sample. It should be done before selecting on mes. - in the study of the contamination use as a reference the study done by Urs one month (or more) ago in which he showed the crossfeed running on reduced collection with a different seed. This could tell you which are the matrix elements with the higher expected crossfeed. - my optimization study on SemiExclusive reconstruction modes shows that we can cut on very low integrated purity. This implies that Alessio's study in very crucial in superblock 3. Daniele On Wed, 30 Jan 2002, Alessio Sarti wrote: > Hi all, > I've just ultimated my first check on timestamps with the new machinery. > This email is intented to let you now how is the full chain now working: > a) It is now possible to run anaRecoil/abcRecoil with the -ts '#' option > In this way timestamp of events falling in the deltaE/mes right box are > dumped to the ts_file.# sampled by integrated purity. > > The files with even # are the pure timestamp files (== to those that > CPFramework produces); odd # ones are the files needed in order to remove > the duplicates (they contains the purity information). > > For example there are: > 1) the file produced by CPFramework: > http://www.slac.stanford.edu/~asarti/recoil/ts_stu/ allDstar0_1 > 2) the file obtained running abcRecoil on the reduced collections > belongings to the CPFramework timestamp: > http://www.slac.stanford.edu/~asarti/recoil/ts_stu/ ts_file.26 > > There are some differences: mainly due to the fact that the mes/Delta box > is defined in two different ways in CPFramework and anaRecoil > > The strange thing that I observe is the following: > running on superblock1 events I got events with purities < 80% (either > integrated and per mode): How this can happen? > > The same is for the other superblocks. Can Daniele or Urs comment on this? > > b) obtained the files for D0 and Dstar0 superblock1 I could start the > equal candidates scan: > I found 64 duplicated events on top of > 4987 superblock 1 dstar0 eve & > 14152 superblock 1 d0 > > Those duplicated events are assigned to the events with the highest per > mode purity. > I'm going to start the same study for superblocks 1 & 2. > > Any question or comment is really welcomed. > Thanks, > Alessio > > ______________________________________________________ > Alessio Sarti > Universita' & I.N.F.N. Ferrara > tel +39-0532-781928 Ferrara > > "Quod non fecerunt barbari, fecerunt Berlusconi" > > "Che il bianco sia bianco, e che il nero sia nero > che uno e uno fanno due e che la scienza dice il vero.... > DIPENDE !" > >