DC06 MC samples are documented here.
And some of the inclusive samples are described here.
Status of DC06 stripping
There is a table of the stripping progress here. Usual username/password.
An excel spreadsheet with a list of stripping retentions is here, correct as of March 2008 software week.
Read Me First! Using stripped/reprocessed data
Using stripped OR reprocessed data it is necessary to include the following line in your DaVinci.opts file.
The attached ascii files (provided on the last column of the tables above) can be used to generate a LHCbDataset in Ganga. Thus, there is no need to convert the lists into Gaudi cards. There are three ways of generating the dataset.
Tricky: For dCache Add dcap://pool1.epcc.ed.ac.uk:22125 or dcap://pool2.epcc.ed.ac.uk:22125 to the start of the pathname, for DPM add rfio:. This then becomes the full path to the DATASET which you can use in your DaVinci options. Make sure you check out the monitoring page.
Easy: An example on how to proceed is provided here. You'll need to append dcap://pool1.epcc.ed.ac.uk to the start of dcache filenames, and rfio: for DPM data in the dat files, if it doesn't already exist. This is simple to do on the command-line in ganga.
Even easier: Use Greig's Ganga Goodies to automatically create the dataset from the name of the above datafile. ganga_utils.dataset_from_twiki('filename'). You can see the list of files from within ganga by doing ganga_utils.list_local_datasets().
How much min bias is that?
To work out how much the sample you have run over corresponds to in real terms, like a length of time T, first you must know:
N How many events have I run over?
n How many evens of this sample correspond to 1 second in 4-pi?
e What is the stripping efficiency?
a What is the LHCb acceptance?
c What correction factors should be applied?
T = N*c / ( n * e * a)
The BEST way of calculating N is to run a simple event counting algorithm which creates a histogram which you can add together in a ganga merger. This is the simplest gaudi algorithm it is possible to write. An example for you to compile is given here EventCount.hEventCount.cpp. Run it at the start before any selections to get the total number of events.
You should calculate the cross-section * branching fraction of your decay to calculate n.
If you are using a stripped sample, the equivalent time is longer. You should see the table of stripping retention rates here to get e.
In DC06 decays with a DecProdCut the total acceptance of a generated event with all products inside the detector is a ~0.19. For a full list of efficiencies see gen-level_efficiencies.txt and on the DC06 stats pages for bb-samples here and here
The DC06 pythia has a MUCH higher bb-cross section in minimum bias than theory suggests, roughly 750 microbarns (Pythia), vs 500 microbarns (Theory). Hence for any min bias data you must take a correction factor of c ~1.5 into account.
According to this talk (footnote, page 8) there are no generator level cuts imposed upon the L0-yes sample, but it corresponds to a rate of 1.15MHz after L0. Or in detail 884360 Min-bias events are pased in 1 second of level zero, n* a* e =884360. The bb in this sample will be over estimated by a factor of 1.5 due to the Pythia cross-section as stated above.
MicroDST
Documentation
For testing purposes
I have created a sample of microDSTs from the Bs->Jpsi phi (DC06 phys-v2-lumi2 dataset) using the standard pre-selections for this channel. They are available at:
/Disk/lochnagar0/lhcb/microDST and their names start with microDST_Bs2JpsiPhi_*. There are 119 files (1.6 Gb) and there should be ~ 180k selected events.