Upload
adam-england
View
32
Download
1
Tags:
Embed Size (px)
DESCRIPTION
Status of MEG Software. Fabrizio Cei INFN and University of Pisa PSI Review Meeting PSI, 14 February 2007. Outline. MEG Software organization Status of Monte Carlo simulation Short remind of analysis framework Status of analysis codes Status of MEG computing power @PSI - PowerPoint PPT Presentation
Citation preview
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 11
Fabrizio CeiFabrizio Cei
INFNINFN and and UniversityUniversity of of PisaPisa
PSI Review MeetingPSI Review Meeting
PSI, 14 February 2007PSI, 14 February 2007
Status of MEG SoftwareStatus of MEG Software
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 22
OutlineOutline
MEG Software MEG Software organizationorganization Status of Status of Monte Carlo simulationMonte Carlo simulation Short remind of Short remind of analysis frameworkanalysis framework Status of Status of analysis codesanalysis codes Status of Status of MEG computing powerMEG computing power @PSI@PSI ProblemProblem ofof overall data sizeoverall data size ConclusionsConclusions
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 33
MEG Software MEG Software OrganizationOrganization
MCMC
BartendBartenderer
(ROME)(ROME)AnalyzerAnalyzer
(ROME)(ROME)
DAQDAQ
ZEBRAZEBRA
ROOTROOT
MIDASMIDAS
ROOTROOT
SimulationSimulation
Real DataReal Data
WFM andWFM and pile-up simulationsimulation
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 44
Status of Monte CarloStatus of Monte Carlo
MEGMC programMEGMC program- written in - written in Geant3.21Geant3.21;;- data output in - data output in ZEBRA banksZEBRA banks, automatically converted , automatically converted
to to C++ structuresC++ structures (readable from analysis codes); (readable from analysis codes);- it simulates - it simulates pairpair (e.g. (e.g. ee) or ) or singlesingle (e.g. (e.g. MichelMichel positronspositrons) ) events;events;- - Full simulation of detector & calibration devices; Full simulation of detector & calibration devices;
- specific modules (- specific modules (tbeamtbeam & & tbtctbtc) for ) for LXe/TC beam testsLXe/TC beam tests;; - - Next release soonNext release soon;; - - Updates since last meetingUpdates since last meeting
– Common magnetic field with MEGAnalyzerCommon magnetic field with MEGAnalyzer– Different geometrical configurationsDifferent geometrical configurations ( (finalfinal or or run 2006run 2006););
- - To be doneTo be done;;- Few event types missing, e.g. LED;Few event types missing, e.g. LED;- Revision for radiative decay and AIF.Revision for radiative decay and AIF.
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 55
Examples of MC Examples of MC eventsevents
ee event event Michel positronMichel positron
RUN 2006 configurationRUN 2006 configuration::No LXe, no TICZ, 8 DCH, No LXe, no TICZ, 8 DCH, TICP displaced by TICP displaced by ± 12 cm± 12 cm
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 66
MC mass productionMC mass production Expected to Expected to start very soonstart very soon;; Event Event typestypes and and numbersnumbers to be decided. to be decided.
Typical CPU & DISK Typical CPU & DISK requirements for requirements for
gemgem
CPU time / 10 k CPU time / 10 k eventsevents
disk size / 10 k disk size / 10 k eventsevents
SignalSignal 24 hrs24 hrs 2.5 GB2.5 GB
alphaalpha 1.3 hrs1.3 hrs 400 MB400 MB
Michel (> 40 MeV) Michel (> 40 MeV) 10 min10 min 700 MB700 MB
Michel (> 5 MeV)Michel (> 5 MeV) 5 min5 min 550 MB550 MB
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 77
Software framework: Software framework: ROMEROME
ROMEROME ( (RRoot basedoot based OObjectbject OrientedOriented MMidasidas EEnvironmentnvironment) ) is a is a framework generatorframework generator.. It uses only It uses only 6 different C++ objects6 different C++ objects.. ROMEROME makes the makes the “dirty job”“dirty job”: : creating the structure, defining C++ classes, writing many creating the structure, defining C++ classes, writing many include include files, creating the dependences and the hierarchy …; files, creating the dependences and the hierarchy …; the users and detector expertsthe users and detector experts perform the perform the “smart job”“smart job”: : writing the analysis methods (writing the analysis methods (taskstasks) and the related ) and the related foldersfolders (data stored on memory) and (data stored on memory) and treestrees (data stored on disks). (data stored on disks). The most important feature is the modularityThe most important feature is the modularity: : the tasks can be exchanged at runtime.the tasks can be exchanged at runtime.
Main developer: M. Schneebeli (PSI)Main developer: M. Schneebeli (PSI)
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 88
ROME InterconnectionsROME Interconnections
Folders
TasksTasksTasksFill
Read
TreesTreesTreesFill
Flag
HistogramsHistogramsHistograms
HistogramsHistogramsHistograms
HistogramsHistogramsHistograms
Fill
Fill
Disk (Output)Write (ROOT)
Disk (Input)
Read (Format: ZEBRA, MIDAS & ROOT)
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 99
ROME Event Display ROME Event Display (ARGUS)(ARGUS)
Display includes tracks Display includes tracks and energy depositsand energy deposits
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1010
Waveform/track Waveform/track displaydisplay
Liquid XenonLiquid Xenon
Drift ChamberDrift Chamber
Both used in on-line Both used in on-line and analysis tooand analysis too
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1111
Trigger/TICP displayTrigger/TICP displayRate of individual Rate of individual
channelschannels
TC barsTC bars
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1212
Status of Status of MEGBartenderMEGBartender
MEGBartender runs stable.MEGBartender runs stable. Event mixing Event mixing withwith multiple formatsmultiple formats
planned planned (presently only ZEBRA);(presently only ZEBRA); Refinement of simulation parametersRefinement of simulation parameters on on
the basis of the basis of run 2006 datarun 2006 data;; Waveform simulation completedWaveform simulation completed for for LXeLXe, ,
TICP/TICZ, DCH wiresTICP/TICZ, DCH wires; work needed ; work needed for pad simulation;for pad simulation; Several trigger simulationsSeveral trigger simulations included. included.
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1313
Status of MEGAnalyzer Status of MEGAnalyzer 1)1)
MEGAnalyzer modified for use in online & MEGAnalyzer modified for use in online & offlineoffline
LXeLXe: ((R. Sawada (Tokyo)R. Sawada (Tokyo), , G. Signorelli (Pisa)G. Signorelli (Pisa), Y. , Y.
Uchiyama (Tokyo), S. Yamada (UCI), F. Cei (Pisa))Uchiyama (Tokyo), S. Yamada (UCI), F. Cei (Pisa))-- Waveform decodingWaveform decoding implemented; implemented;- Charge-based reconstruction algorithms implementedCharge-based reconstruction algorithms implemented; ; most of them tested;most of them tested;- Timing reconstruction & calibration algorithmsTiming reconstruction & calibration algorithms under implementation and testing;under implementation and testing;- Peak findingPeak finding and and pattern recognition taskspattern recognition tasks existing. existing.
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1414
Status of MEGAnalyzer Status of MEGAnalyzer 2)2)
TICP/Z TICP/Z ((P. Cattaneo (Pavia)P. Cattaneo (Pavia), Y. Uchiyama, D. Zanello (Rome),, Y. Uchiyama, D. Zanello (Rome), F. Xiao (UCI), A. Barchiesi (Rome), S. Dussoni (Genova))F. Xiao (UCI), A. Barchiesi (Rome), S. Dussoni (Genova))
-- waveform analysis implementedwaveform analysis implemented;;- preliminary hit reconstruction implemented- preliminary hit reconstruction implemented (Q,t(Q,tLL,t,tRR) ) (z,<t>) (z,<t>);;
-- still missingstill missing: : correlation between adjacent bars & with DCH.correlation between adjacent bars & with DCH.
DCH DCH ((H. Nishiguchi (Tokyo)H. Nishiguchi (Tokyo), , M. SchneebeliM. Schneebeli, Y. Hisamatsu, , Y. Hisamatsu, V. Tumakov (UCI))V. Tumakov (UCI)) - 3D-map of magnetic field- 3D-map of magnetic field;; - tracking by (preliminary) Kalman filter implemented- tracking by (preliminary) Kalman filter implemented;; - waveform decoding existing- waveform decoding existing;;
- extraction of - extraction of z-coordinatez-coordinate from cathode pad information in progress from cathode pad information in progress.. - still missing: full hit reconstruction (3D information);- still missing: full hit reconstruction (3D information);
pattern recognition at the beginning.pattern recognition at the beginning.
Software changing very fast on the basis of 2006 dataSoftware changing very fast on the basis of 2006 data
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1515
Status of MEGAnalyzer Status of MEGAnalyzer 3)3)
Trigger Trigger (G. Signorelli, (G. Signorelli, D. NicolD. Nicolòò (Pisa) (Pisa))) - trigger info/banks coded in MIDAS:- trigger info/banks coded in MIDAS: run #run #, , event #event #, , triggertrigger code & WFMscode & WFMs, , scalersscalers (useful for determining run/live time);(useful for determining run/live time); - - cchargeharge and and timing reconstruction algorithmstiming reconstruction algorithms implemented implemented and under testing.and under testing.
Database Database ((R. SawadaR. Sawada)) - two databases:- two databases: MySQLMySQL and and sqlite3sqlite3; easy conversion; ; easy conversion; - MySQL needs network; sqlite3 for stand-alone environment in - MySQL needs network; sqlite3 for stand-alone environment in a separate svn module (a separate svn module (megdbmegdb); ); - included:- included: geometry, trigger/hardware configuration, run geometry, trigger/hardware configuration, run table, physical constants, reconstruction coefficients …table, physical constants, reconstruction coefficients …
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1616
DRS in MEGAnalyzerDRS in MEGAnalyzer
Data structure establishedData structure established- - It is possible to write It is possible to write limited time regionslimited time regions of of
waveforms;waveforms;- - Data sizeData size (including header) (including header) can be zerocan be zero if the if the
channel is not interesting;channel is not interesting;- Each chip has - Each chip has clockclock and and trigger signaltrigger signal for for calibration.calibration.
Calibration procedures under Calibration procedures under development development
and to be studied in detail.and to be studied in detail.
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1717
MEG computing @PSIMEG computing @PSI
Sun Fire x4100 quad core 4 GB
Sun Fire x4100 quad core 4 GB
Sun Fire x4100 quad core 4 GB
Sun Fire x4100 quad core 4 GB
Sun Fire x4100 quad core 4 GB
Fib
er
Ch
an
nel Sw
itch
15 x 500 GB SATA
GB
it E
thern
et
Offline cluster for MEG (LCMEG)Offline cluster for MEG (LCMEG)
• Presently available : Presently available : 20 CPU cores + 30 TB disk + Sun Grid 20 CPU cores + 30 TB disk + Sun Grid EngineEngine
• Final situation : Final situation : 64 CPU cores + 100 TB disk in total64 CPU cores + 100 TB disk in total• Easily extensibleEasily extensible; planned within 2007; planned within 2007
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1818
Overall Data RateOverall Data Rate
• In 2006 run we had In 2006 run we had 2.8 MB events2.8 MB events (50% DC + TC, no LXe; (50% DC + TC, no LXe; 1/3 of the final configuration) and could run at 1/3 of the final configuration) and could run at ~ 10 Hz~ 10 Hz
• Full detector estimates Full detector estimates without any reductionwithout any reduction: : 9 9 MB/eventMB/event, ,
5 Hz5 Hz, , 101077 sec sec, , 450 TB/year450 TB/year• Possible strategies for reducing data size:Possible strategies for reducing data size:
– Zero suppressionZero suppression (50% on LXe, 80% on DC) (50% on LXe, 80% on DC)– ADC/TDC values for non-signal-like eventsADC/TDC values for non-signal-like events– Partial waveform readoutPartial waveform readout (reduced window size) (reduced window size)– Keep timing information in SQL databaseKeep timing information in SQL database– 33rdrd level trigger in online cluster level trigger in online cluster (e.g. linear fit, fast tracking …)(e.g. linear fit, fast tracking …)
Effects on data Effects on data to be evaluated to be evaluated
Effects on data Effects on data to be evaluated to be evaluated
30 Tb/year30 Tb/year
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 1919
ConclusionsConclusions The MEG software is in an advancedThe MEG software is in an advanced state ofstate of preparation:preparation:
– MEGMC + MEGBartenderMEGMC + MEGBartender: : almost finishedalmost finished– MEGAnalyzerMEGAnalyzer: :
LXeLXe: close to completion;: close to completion; TICP/ZTICP/Z & & DCHDCH: some parts missing, but fast : some parts missing, but fast
evolution and significant effort; evolution and significant effort; Trigger/DatabaseTrigger/Database: ~ ok.: ~ ok.
First part (~ 1/3) of MEG offline clusterFirst part (~ 1/3) of MEG offline cluster in operation; in operation; Data size to be reducedData size to be reduced; ; discussion under discussion under
wayway..
14 February 20014 February 20077
Fabrizio CeiFabrizio Cei 2121
MC Man PowerMC Man Power
All persons @ 10 All persons @ 10 50 % of their time 50 % of their time:
Coordination: Coordination: S. Yamada (UCI)S. Yamada (UCI), , F. Cei (Pisa)F. Cei (Pisa);; SVN repository: SVN repository: S. YamadaS. Yamada;; Event generation:Event generation:F. Cei, S. YamadaF. Cei, S. Yamada, , Y. Hisamatsu (Tokyo)Y. Hisamatsu (Tokyo);; LXe: LXe: S. Yamada, F. CeiS. Yamada, F. Cei, , G. Signorelli (Pisa);G. Signorelli (Pisa); TICP/TICZ: TICP/TICZ: P. Cattaneo,P. Cattaneo, Y. Uchiyama (Tokyo)Y. Uchiyama (Tokyo);; DCH: DCH: H. Nishiguchi (Tokyo)H. Nishiguchi (Tokyo), , M. Hillebrandt M. Hillebrandt
(PSI)(PSI); ; Beam & Magnet: Beam & Magnet: W. Ootani (Tokyo)W. Ootani (Tokyo);; Target: Target: V. Tumakov (UCI);V. Tumakov (UCI); NaI: NaI: Y.Y. Nishimura (Tokyo);Nishimura (Tokyo); Calibrations: Calibrations: F. CeiF. Cei