LHCb applications

Overview of Gaudi and of
LHCb applications
LHCb Simulation Tutorial
CERN, 21 st -22 nd February 2012
011010011101
10101000101
01010110100
B00le
1

Simulation Tutorial - Feb. 2012
2
LHCb software Organization
Applications built on top of
frameworks and implementing the
required algorithms.
One framework for basic services
+ various specialized frameworks:
detector description,
visualization, persistency,
interactivity, simulation, etc.
High level triggers
Reconstruction
Simulation
Frameworks
Toolkits
Analysis
A series of widely used basic
libraries: Boost, GSL, Root etc.
Foundation Libraries

Simulation Tutorial - Feb. 2012
3
LHCb applications
Event Model / Detector Description / Conditions
MicroDST
GenParticles
Raw Data
Simul.
Gauss
Digit.
Boole
HLT
Moore
Reco.
Brunel
Stripping
DaVinci
Analysis
DaVinci
Analysis
Bender
MCParticles
MCHits
DST
Ev.Disp.
Panoramix
Gaudi

Simulation Tutorial - Feb. 2012
4
Gaudi Framework
! GAUDI is an architecture and framework for eventprocessing
applications (simulation, reconstruction, etc.)
! Initially developed for LHCb, it has been adopted and extended by
ATLAS and adopted by several other experiments including Fermi
Space Telescope, HARP, Minerva, Daya Bay
! Main Design Choices best illustrated by looking at the
Gaudi Object diagram
! A snapshot of the components in memory, once the job is initialized
and running

Simulation Tutorial - Feb. 2012
5
Gaudi Object Diagram
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
6
Separation between “data” and “algorithms”
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
7
Three basic categories of data
characterized by their “lifetime” in the job
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
8
Separation between “transient” and “persistent”
representations of the data
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
Data store-centered (“whiteboard”) architectural
style
9
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
10
“User code” (Algorithms, Tools) encapsulated in few specific
places, callable and controllable from within the framework
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
11
“User code” (Algorithms, Tools) encapsulated in few specific
places, callable and controllable from within the framework
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Gauss
Event Generation
Algorithms & Tools
Detector Simulation
Algorithms & Tools
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
software
Simulation dedicated
Geant4 Service (GiGa)
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files
Generator Libraries

Simulation Tutorial - Feb. 2012
12
Well defined component “interfaces”
Application
Manager
Event
Selector
Converter
Converter
Converter
Message
Service
JobOptions
Service
Particle Prop.
Service
Algorithm
Algorithm
Event Data
Service
Detec. Data
Service
Transient
Event
Store
Transient
Detector
Store
Persistency
Service
Persistency
Service
Data
Files
Data
Files
Other
Services
Histogram
Service
Transient
Histogram
Store
Persistency
Service
Data
Files

Simulation Tutorial - Feb. 2012
13
Gaudi Interface model
ApplicationMgr
ISvcLocator
EventDataSvc
DetectorDataSvc
HistogramSvc
IDataProviderSvc
IDataProviderSvc
IHistogramSvc
IAlgorithm IProperty
Concrete
Algorithm
MessageSvc
IMessageSvc
Obj_A
Obj_B
ToolSvc
IToolSvc

Simulation Tutorial - Feb. 2012
14
Laptop Interface Model
IWiFiPort
IDVIConnector
IVGAConnector
IKeyboard
IUSBConnector
• Each interface is specialized in a
domain.
• Interfaces are independent of
concrete implementations.
• You can mix devices from several
constructors.
• Application built by composing.
• Standardizing on the interfaces
gives us big leverage.

Simulation Tutorial - Feb. 2012
15
Electrical plug interface model
Don’t define too many interfaces for the same job!

Simulation Tutorial - Feb. 2012
16
IMyInterface.h
Interfaces in Practice
class IMyInterface: virtual public IAlgTool {
virtual double doSomething( int a, double b ) = 0;
}
MyDoSomethingTool.h
#include “IMyInterface.h”
class MyDoSomethingTool : public GaudiTool,
virtual public IMyInterface {
inline double doSomething( int a, double b )
{ return b * (double)a; }
}

Simulation Tutorial - Feb. 2012
17
ClientAlgorithm.cpp
#include “IMyInterface.h”
ClientAlgorithm::myMethod() {
// Declare the interface
IMyInterface* myInterface;
// Get the interface from somewhere
myInterface = tool(“MyDoSomethingTool”);
// Use the interface
double result = myInterface->doSomething( 10, 100.5);
}

Simulation Tutorial - Feb. 2012
18
Algorithm & Transient Store
Data T1
Data T1
Data T2, T3
Algorithm
A
Transient Event
Data Store
Data T2
Data T4
Algorithm
B
Data T3, T4
Data T5
Algorithm
C
Apparent dataflow
Real dataflow
Data T5

Simulation Tutorial - Feb. 2012
19
Tools and Services
Algorithms have some limitations
Only called once per event, only share data via TES
What about private data?
Tools are components providing algorithmic functionality
Callable many times per event, through specific interface and with arguments
IBdlTool, IDistanceCalculator, IHltSummaryTool, IMassVertexFit,
IProtoParticleFilter, ISTReadoutTool, ITrackFitter….
Can be private: owned and accessible only by the component creating it
Or public: owned by the framework and accessible globally
Services:
Similar to public tools, created by framework to provide global functionality
Job Options Service, Message Service, Particle Properties Service, Event Data
Service, Histogram Service, Ntuple Service, Detector Data Service, Magnetic
Field Service, Random Number Service, Persistency Services…

Simulation Tutorial - Feb. 2012
20
Persistent datasets
! A special algorithm (“OutputStream”) runs at the end of
each event to select transient event data to be written to
the persistent output file
! Defines content of output file
! Configured by “Configurables” in GaudiConf package
SimConf.py, DigiConf.py, DstConf.py
! Called by main application Configurable
Gauss(), Boole(), Brunel(), DaVinci()
! Several pre-defined contents are available:
gen: Gauss().Phases = [“Generator”]
xgen: Gauss().Phases = [“Generator”, “GenToMCTree”]
sim: Gauss default
digi Boole default
xdigi Boole().DigiType = “Extended”
dst Brunel default
xdst Brunel().OutputType = “XDST” (needs xdigi as input)
(sdst) Brunel().OutputType = “SDST” (only for real data production)
(mdst) Stripping output, several streams defined, only real data

Simulation Tutorial - Feb. 2012
21
! gen
! Generator files: Files with only the /Event/Gen tree, hence only generator information and the HepMC
record. The complete hard interaction is available.
! xgen
! Extended generator files: Files with the /Event/MC/MCParticles and /Event/MC/MCVertices tree in
addition to /Event/Gen To produce these files only the generator phase of Gauss has been run, i.e. the
content in the MCtruth tree is a copy of the generator information after hadronization. In other words
quarks and strings do not appear in the MCParticles
! sim
! Simulation files: Files resulting from the detector simulation. The configuration of the detector (all, velo
open, calo only, dddb tag) is specified in the SimulationConditions. The MCtruth tree is filled with the
information from the generator for primary particles and with the secondaries produced in the detector
trough transport. Hits for all detectors are also present (MCHits, MCRichHits, MCCaloHits) and
additional simulation information as MCRichSegments. In case of simulations with spill-over the whole
information is also available for the the spill-over slots. The full /Event/Gen is copied.
! digi
! Digitization files: Files containing the digitization of the simulation and the raw event as from the DAQ.
They result from the Boole processing. After the Moore processing the trigger information is added. They
include the MCTruth, hence the /Event/Gen and the /Event/MC/Particles and Vertices.
! xdigi
! Extended digitization files: Files with the whole content of both sim and digi processing.
! dst
! Dst files: Files resulting from the reconstruction of MC samples or real data. They include the RawEvent
and the full reconstruction output. For MC they also include the MC truth, hence the /Event/Gen and
the /Event/MC/Particles and Vertices. After stripping, also include particles forming signals selected by
the lines that have fired
! xdst
! Extended dst files: Files with the whole content of sim, digi and dst (simulated data only)
! sdst
! dst files for input to real data stripping: reconstruction output only, no MC truth, no RawEvent
! mdst
! MicroDST files: the smallest format containing reconstruction information of only particles forming signal
selected by the stripping lines that have fired. MC truth not implemented