Difference: RunningTheHlt (1 vs. 5)

Revision 52011-10-12 - RoelAaij

Line: 1 to 1
 
META TOPICPARENT name="LHCbTrigger"

Poor mans manual to run the HLT

Line: 36 to 36
  If no explicit filename is given to the Writer, it doesn't do anything special.
Deleted:
<
<

Running the Hlt in DaVinci, 2010

In order to run the Hlt in DaVinci, it is at a minimum necessary to specify a threshold settings which configure the Hlt. These can be found in the HltSettings package and each setting also specifies which L0 configuration is associated to the Hlt setting in question.

     from Configurables import DaVinci
     DaVinci().Hlt = True
     DaVinci().HltThresholdSettings = 'Physics_LoadsaEventsThankYouVeryMuch'
     
If you wish to re-run the L0, it is necessary to specify that you are Overruling the value specified in the Hlt settings and that you wish to Replace the existing L0 banks on your datafile with the output of the offline L0 emulation.
     DaVinci().OverruleL0TCK = '0xBlaBla' 
     DaVinci().ReplaceL0BanksWithEmulated = True
     
The list of L0 TCKs which can be entered in place of "0xBlaBla" above can be found in the L0DU package in the options folder. Not all the L0 TCKs are compatible with all the Hlt threshold settings. For running in 2010 you will be mostly interested in L0 TCK "0x1710" which is the default L0 TCK for physics data taking in the pit.

Each Hlt setting assumes a single corresponding L0 TCK. Sometimes, however, you do not want to reemulate the L0 TCK but you wish to rerun the Hlt on some data collected with more than one L0 TCK -- typically if you are running on real data taken early in 2010. By default, the Hlt will exit the job if it finds that an event was recorded with an L0 TCK different from the one specified in the Hlt settings. However, it is possible to tell the Hlt to ignore this

     from Configurables import HltConf
     HltConf().ForceSingleL0Configuration = False
     
Note that this may or may not work; some L0 TCK combinations are genuinely incompatible with each other and cannot be processed with the same Hlt setting. You will have to know in advance which L0 TCKs will be found on your input files and verify that they are compatible with the Hlt setting which you wish to use.

It is useful to turn on the monitoring of the Hlt algorithms, like so.

     HltConf().EnableMonitoring = True
     HltConf().HistogrammingLevel = 'Everything'
     
 

Running the Hlt with Moore

Changed:
<
<
Running Moore is the only way to be sure that you reproduce what ran in the pit
>
>
Running Moore is the only way to be sure that you reproduce what ran in the pit and is the only supported way of running the trigger on MC. Before you start, there are several things you need to know:
  • Was Moore already run as part of the production
  • Which TCK do you want to use (use TCKsh to list available configurations)
  • Which version of Moore? This is coupled to the TCK, but always take the latest patch release for a given major and minor number.
  • In case you have special requirements which are not available in any TCK, follow the instructions in PrivateTCK.
 
Changed:
<
<
Use RunMoorev10rXMC.py script in the python directory of the package Hlt/Hlt2CommissioningScripts
>
>
Several example scripts named RunMoorevXrYMC.py are available in python directory of the package Hlt/Hlt2CommissioningScripts. Use the one that best matches your required Moore version.
  See also Johannes' talk "Running Moore" at the November 2010 Software week.

Revision 42010-12-13 - UlrichKerzel

Line: 1 to 1
 
META TOPICPARENT name="LHCbTrigger"
Changed:
<
<

Poor mans manual to run the HLT

>
>

Poor mans manual to run the HLT

 
Changed:
<
<

Running the Hlt in DaVinci, 2010

>
>

Running Moore in Production (for simulated events)

Setting up TCKS

The TCKs used during central production of MC are placed in files in the directory AppConfig/options/Conditions. To edit, use

getpack AppConfig HEAD

E.g.

$ cat TCK-0x002a002a.py
from Configurables import Moore
Moore().InitialTCK = '0x002a002a'

Running in pass-through or reject mode

Whether or not the HLT is run in pass-through mode is configured via the Writer in Moore: Either no HLT bank is required, then all events are written out:

Moore().WriterRequires = []

or to run in rejection mode:

Moore().WriterRequires = [ 'HltDecisionSequence']

If no explicit filename is given to the Writer, it doesn't do anything special.

Running the Hlt in DaVinci, 2010

  In order to run the Hlt in DaVinci, it is at a minimum necessary to specify a threshold settings which configure the Hlt. These can be found in the HltSettings package and each setting also specifies which L0 configuration is associated to the
Line: 40 to 75
  HltConf().HistogrammingLevel = 'Everything'
Changed:
<
<

Running the Hlt with Moore

>
>

Running the Hlt with Moore

  Running Moore is the only way to be sure that you reproduce what ran in the pit

Revision 32010-12-06 - unknown

Line: 1 to 1
 
META TOPICPARENT name="LHCbTrigger"

Poor mans manual to run the HLT

Line: 40 to 40
  HltConf().HistogrammingLevel = 'Everything'
Added:
>
>

Running the Hlt with Moore

Running Moore is the only way to be sure that you reproduce what ran in the pit

Use RunMoorev10rXMC.py script in the python directory of the package Hlt/Hlt2CommissioningScripts

See also Johannes' talk "Running Moore" at the November 2010 Software week.

 -- VladimirGligorov - 16-Jun-2010

Revision 22010-06-16 - JohannesAlbrecht

Line: 1 to 1
 
META TOPICPARENT name="LHCbTrigger"
Added:
>
>

Poor mans manual to run the HLT

 

Running the Hlt in DaVinci, 2010

In order to run the Hlt in DaVinci, it is at a minimum necessary to specify a threshold settings which configure the Hlt.

Revision 12010-06-16 - VladimirGligorov

Line: 1 to 1
Added:
>
>
META TOPICPARENT name="LHCbTrigger"

Running the Hlt in DaVinci, 2010

In order to run the Hlt in DaVinci, it is at a minimum necessary to specify a threshold settings which configure the Hlt. These can be found in the HltSettings package and each setting also specifies which L0 configuration is associated to the Hlt setting in question.

     from Configurables import DaVinci
     DaVinci().Hlt = True
     DaVinci().HltThresholdSettings = 'Physics_LoadsaEventsThankYouVeryMuch'
     
If you wish to re-run the L0, it is necessary to specify that you are Overruling the value specified in the Hlt settings and that you wish to Replace the existing L0 banks on your datafile with the output of the offline L0 emulation.
     DaVinci().OverruleL0TCK = '0xBlaBla' 
     DaVinci().ReplaceL0BanksWithEmulated = True
     
The list of L0 TCKs which can be entered in place of "0xBlaBla" above can be found in the L0DU package in the options folder. Not all the L0 TCKs are compatible with all the Hlt threshold settings. For running in 2010 you will be mostly interested in L0 TCK "0x1710" which is the default L0 TCK for physics data taking in the pit.

Each Hlt setting assumes a single corresponding L0 TCK. Sometimes, however, you do not want to reemulate the L0 TCK but you wish to rerun the Hlt on some data collected with more than one L0 TCK -- typically if you are running on real data taken early in 2010. By default, the Hlt will exit the job if it finds that an event was recorded with an L0 TCK different from the one specified in the Hlt settings. However, it is possible to tell the Hlt to ignore this

     from Configurables import HltConf
     HltConf().ForceSingleL0Configuration = False
     
Note that this may or may not work; some L0 TCK combinations are genuinely incompatible with each other and cannot be processed with the same Hlt setting. You will have to know in advance which L0 TCKs will be found on your input files and verify that they are compatible with the Hlt setting which you wish to use.

It is useful to turn on the monitoring of the Hlt algorithms, like so.

     HltConf().EnableMonitoring = True
     HltConf().HistogrammingLevel = 'Everything'
     

-- VladimirGligorov - 16-Jun-2010

 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback