1. Modifications in Gaudi v5 release:
Access to SICB data
Modifications in Gaudi v5 release:
Pile up.
Other new features:
Pre-staging of several tapes
Access to new data base
Gonzalo Gracia Abril

2. Pile Up Implementation
Pile up infra-structure was already in Gaudi v4
But it uses internally SICB and it was not working.
Some more debugging was required.
Pile up is implemented as a separate algorithm which has to run before any other algorithm which access data:
JobOptions: ApplicationMgr.TopAlg = { "PileUpAlg",”myAlg1",”myAlg2" };
Number of pile up events is calculated as a function of the luminosity.
RELUMI re-implemented in C++:
Reads beam.cdf file to get input parameters.
Uses new random generator service (See later)
Two files read in, one for the main event and the second for pile up events.

3. SicbEventSelector ApplicationMgr taSvc icnbSicbPileupAlg ReMerge SICB
IProperty
IProperty “
EventSetector ”
SicbEventSelector
IEventSelector
SicbEventSelector
IEventSelector
ApplicationMgr
ApplicationMgr
IDataManagerService
IDataManagerSvc
ZEBRA “
EventDataSvc ”
taSvc
EventDataSvc “
PileupAlg ”
icnbSicbPileupAlg
SicbPileUpAlg
IDataProviderSvc
ReMerge
IDataProviderSvc
SICB
specific
IDataProviderSvc “
PileupSvc ”
SicbPileupSvc “
PileupDataSvc ”
EventDataSvc
SicbPileUpSvc
EventDataSvcvc
ZEBRA
IDataManagerService
IDataManagerSvc “
PileupSelector ”
SicbEventSelector
IEventSelector
SicbEventSelector
The Algorithm and
Service are merged
into one Algorithm for
the Sicb version
Not needed for
the SICB version
IEventSelector
The Algorithm and
Not needed for
Service are merged
the
Sicb
version
IProperty
into one Algorithm for
IProperty
the
Sibc
version

4. Random Numbers
Reminder : Two random numbers are generated for every event to get the number of pile up events:
First number generated following a flat distribution (0,1) to calculate the current luminosity (from the fill duration and beam decay time, the current time in the fill is the random number).
Second (number of interactions): Follows a Poisson distribution whose mean is equal to the average number of interactions for the current luminosity (extracted from the cross section and the bunch crossing rate)
If the number of iterations is 0 we repeat the process until is different than 0
Random number service is now available in Gaudi and we have chosen to use it.

5. Number of interactions.
Histogram: Gaudi
Points: SICBDST
500 Events
Number of interactions.

6. Average number of interactions per bunch crossing.
Histogram: Gaudi
Points: SICBDST
500 Events
Average number of interactions per bunch crossing.

7. Pile Up: Some comments The actual merging of events is done by REMERGE
ATMC, AVMC and raw hits banks are merged. (CALO raw banks are not converted)
When you retrieve, from your Gaudi Algorithm, data converted from one of these banks you will get the merged data.
In the conversion Gaudi follows every reference, we find always the missing or wrong references!!.
(It runs in 500 events, several data samples tried… but watch the DST2 problems!!)
Two instances of the event selector which are configured separately

8. Other new features Pre-staging of several tapes:
You wait for the first tape. The others are staged while you read the first one. If you finish reading the first file and the second tape is not staged you will get a message and will have to wait.
Access to tapes is not possible from NT
Access to new data base
Gaudi sends a request to the ORACLE web server
Executes the same procedure used by mcstagein to get the volume number, file sequence number and size.