1. SNiPER在LHAASO实验中的应用： LodeStar
李腾 黄性涛 张学尧
山东大学

2. Outline Overview of Lodestar Core software of Lodestar
Applications of Lodestar
Next to do

3. Overview of Lodestar LodeStar: Constituent of LodeStar:
LHAASO Offline Data Processing Software Framework
Constituent of LodeStar:
SNiPER: underlying framework
offline: specific to LHAASO Experiments
External Libraries: frequently used third-party software or tools

4. Lodestar Environments
Programming Language：C++、Python
very popular in HEP
most frequently used software implemented in C++
Supported OS：Linux
Scientific Linux 5, 6 are currently supported
More OS will be tested and supported according to needs
Software Management Tool：CMT
Automatically compile, build packages
Configuration Language：Python
Very flexible
Supported by SNiPER
Version Control Tool：SVN
keep history of codes developing
synchronization and sharing between developers

5. Functionalities of Lodestar
Management of the data processing procedures
Process/Sequence management
Based on Task, Algorithm, Incident etc. in Sniper
Management of the event data
Data Store
Data Model, Data I/O
Providing common services or tools during processing
HisogramSvc, RandomSvc, DatabaseSvc
Based on Service, Tool etc. in Sniper
Providing friendly user interfaces
Easy to develop user code
Configure jobs, based on SniperPython

6. Data Processing Procedures in Lodestar
Data Processing Workflow:
Generator
GenEvent
Each data processing is developed based on the Algorithm, Service and Tool in Sniper
Each data processing reads event data from previous step and produces its event data for the next one
Simulation
SimEvent
Reconstruction
RecEvent
Analysis
PhyEvent

7. Data Processing Procedures in Lodestar
Data Processing Workflow:
Generator
GenEvent
Simulation
The event data is saved in the Data Model objects, stored in the Data Store
Data Model objects can be converted into persistent form and saved into files
SimEvent
Reconstruction
RecEvent
Analysis
PhyEvent

8. Design of the DataModel
Design of LHAASO Event Data Model
Each process defines their Event Objects
EventObject inherits from TObject.
TRelations: Matching between Headers (optional)

9. Design of the DataModel
Each event data consists of two parts:
HeaderObject
EventObject
separate meta data from event data
Some “tag” data can be put into header
To speed up event selection
SmartRef: One new type of smart pointer
Support correlation of data objects
Support lazy-loading

10. XmlObjDesc: Define DataModel with XML
Traditionally writing C++ Code by hand
Many repeatable work such as Getters and Setters
Difficult to be maintained
New way to define EDM with XML file
Strong syntax (DTD, XML Schema)
More readable, easier to maintain
Automatically generate the Get-, Set- functions, Streamers
XmlObjDesc (XOD) is used as a tool to define EDM with XML
*.h
XOD
CINT
G++
*.xml
*Dict.h
*Dict.cc
lib*.so
*.cc
*LinkDef.h

11. Event Data Management Event Data Management in Lodestar:
Event data is stored in the Data Store, managed by DataStoreMgr
Data Store is a dynamically allocated place, shared by all the algorithms
All algorithms/services/tools can retrieve/put event data from/to the Data Store with data paths
Data Store is automatically configured with Data I/O Service
Data Store
(MEMORY)
Simulation
Reconstruction
Analysis
DISK
DataInputSvc
DataOutputSvc
Other Svc
DataStoreMgr
DataStoreMgr

12. Event Data Management Event Data Management in Lodestar:
In Lodestar, Data Input/Output is implemented with Data I/O Services:
Responsible for converting transient/persistent Data Model objects
Currently support:
Root Input/Output streams
Corsika Input stream
Data Store
(MEMORY)
Simulation
Reconstruction
Analysis
DISK
DataInputSvc
DataOutputSvc
Other Svc
DataStoreMgr
DataStoreMgr

13. Data I/O Service RootInputSvc/RootOutputSvc: CorsikaInputSvc
Writing/Reading official ROOT data files
Support lazy-loading mechanism
CorsikaInputSvc
Convert the raw corsika file to the DataModel objects
Support different types of input files (particles, Cherenkov photons, Longitudinal parameters)
Support event-splitting

14. Application: G4argo An example of G4-based detector simulation package
DetSimAlg: Common algorithm of Geant4 detector simulation
G4Svc: Interface to the Geant4 core
ArgoSimFactory: Build all the detector simulation options
Generator
Geometry
User-actions

15. Application: G4argo
Some comparison from the origin results of G4argo:

16. Application：KM2A FastSim
KM2A fast simulation
Implemented with KM2ADetSimAlg and KM2ARecAlg
Some comparison of the results:

17. Application：WFCTA Simulation
Implemented with WFCTADetSimAlg
Some comparison of the results:
Lodestar
Origin
Corex, corey , ,
Zenith, azimuth ,
Total photon
Photons arriving telescope
Photon after reflecting
Photon after ray tracing
220824

18. Summary and Next To Do Summary: We still have a large TODO list:
Lodestar has been built based on SNiPER
The G4-based DetSim software prototype is provided
Some of the code has been moved to Lodestar
We still have a large TODO list:
Further improvements of Lodestar needs more considerations and discussions
Many tools and services to be added
Lots of existing packages to be moved to Lodestar
Build our SVN/Trac server
Lots of implementation work to do… …

19. Thank you !