L0App: a very simple configurable application for emulating L0

The "L0 Trigger" is a hardware and firmware based trigger, but it can be emulated to surprizing accuracy in software.

(the result of task #42380 )

....... keeping-it-simple-project-plan-from-point-a-to-point-b

What is it?

  • L0App() is a very simple configurable which has only one use, to emulate the L0 and write out a file with L0 Added to it.
  • It is able to do this to any number of MDF files, RAW files, or DSTs, and employs a combination of:
    • InputCopyStream
    • L0Conf
    • MDFWriter

What does it add?

  • L0App adds or replaces L0 banks in the raw event. It will do this for any file which still has a raw event on it, such as:
    • YES: DST, DIGI, XDIGI, XDST, RAW, MDF
    • NO: SIM, XSIM, MDST

  • On real data it behaves the same as on simulated data in that it replaces all the L0 banks with emulated banks no matter what.

Disclaimer: on real data think about what you're doing first!

When re-emulating L0 on real data, as with any trigger re-emulation consider that the data that you are running on has already been triggered. The LHCb trigger is intended to bias events towards bigger events with higher chance of finding a B-hadron for your analysis. The performance of the re-emulation on already triggered data is then biased by the output of the previous trigger and/or selection, whatever that may be. You are performing a logical AND of your two trigger conditions Trigger1&&Trigger2. In an extreme case, for example, re-running the same L0 on the same file should ideally not reject a single event. So, when using real data remember:

  • Either: use truly random triggers, NoBias data
  • Or: do not trust your naive rates, efficiencies, and performance calculations.

How to use it?

  • Setup the project SetupProject Moore

  • Set up your input
    • any gaudi-card dataset will work, in Ganga, from the book-keeping, or from the TestFileDB... here is an example with a file stored in the TestFileDB
#imagine this is in a file called mydata.py
from GaudiConf import IOExtension
IOExtension().inputFiles(["root://eoslhcb.cern.ch//eos/lhcb/grid/prod/lhcb/swtest/lhcb/swtest/2011_smallfiles_EW/00010197_00000049_1.radiative.dst"])

#imagine this is in a file called L044.py
from Configurables import L0App
L0App().TCK='0x0044'
L0App().outputFile="radiative_plus_l044.dst"
L0App().ReplaceL0Banks=True #only needed if the input file already contains L0

  • e.g.: gaudirun.py mydata.py L044.py

Resources

Options and settings

  • L0App is about the simplest possible configurable application it's possible to write in LHCb software.
  • L0App has very few options, and all of them are very obvious...
class L0App(LHCbConfigurableUser):
    ## Possible used Configurables
    __used_configurables__ = [ LHCbApp
                             , L0Conf
                             , DecodeRawEvent ]
    
    __slots__ = {
          "EvtMax":            -1    # Maximum number of events to process
        , "SkipEvents":        0
        , "Simulation":        True # True implies use SimCond
        , "DataType":          '2012' # Data type, can be [ '2012','2011'... ]
        , "DDDBtag" :          'default' # default as set in DDDBConf for DataType
        , "CondDBtag" :        'default' # default as set in DDDBConf for DataType
        , "ReplaceL0Banks" : False # remove existing banks first?
        , 'TCK' : '' #"Can be a L0TCK or a full TCK, it doesn't matter"
        , "outputFile" :       '' # output filename
        , 'WriteFSR'    :  True #copy FSRs as required
          
        }

Input Requirements

  • Simulation: Ideally L0App runs after digitization but before reconstruction or HLT/Moore. This is equivalent of when the real L0 runs in the pit. So, ideally supply a DIGI or XDIGI
  • Real Data: Similarly the L0 happened before the HLT/Moore and reconstruction, and so it is best to run it on raw files whose HLT has been cleaned off and whose reconstruction has not happened.
  • Juggling: unfortunately the raw event is split up and moved around by current Brunel and so if you're running on reconstructed data you will need to think about -
  • DAQ/RawEvent:
    • just like Moore, L0App requires only one raw event location, and this should be DAQ/RawEvent
    • However, Brunel if run on your input file will most probably have split this up into different locations, and you need to recombine them.
    • Use RawEventJuggler twiki to understand what you need to do here and why.
    • You only need to juggle files which have already been reconstructed!
  • How to tell what my input file contains?
from Gaudi.Configuration import *
from Configurables import LHCbApp
from Configurables import RawEventDump
from Configurables import StoreExplorerAlg

StoreExplorerAlg().Load=True
RawEventDump().DumpData=False
LHCbApp().EvtMax=10

ApplicationMgr().TopAlg+=[StoreExplorerAlg(), RawEventDump()]

#add your input files..., example here from Eric
from GaudiConf import IOExtension
IOExtension().inputFiles(['root://castorlhcb.cern.ch//castor/cern.ch/user/e/evh/bwdivision/13774002-juggled.dst'],clear=True)

Common use-cases:

  • Emulating L0 on fully simulated data.
  • Re-emulating a new L0 on real data in order to record a TCK from a threshold setting (use NoBias data)
  • Re-emulating L0 with changed thresholds to perform the bandwidth division (either NoBias data [background] or simulation [signal])
  • Re-emulating L0 on real data for the purposes of testing in the nightlies (ideally use NoBias data, but this depends on the test)
  • Re-emulating in order to calculate the overlap in efficiencies between two TCKs.

Feature discussion

  • Immutable: Any L0App version should be able to emulate any/all previous L0 TCKs
    • Each night this is tested, if there is any problem emulating our firmware into the future, the L0 team need to know immediately.
  • Simple: L0App is meant to be as simple as possible, and only to do one task.


-- RobLambert - 24 Mar 2014

Edit | Attach | Watch | Print version | History: r6 < r5 < r4 < r3 < r2 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r6 - 2016-04-01 - unknown
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2018 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback