GRAPES-3 ROOT Framework

Pravata K Mohanty

Tata Institute of Fundamental ResearchOn behalf of the GRAPES-3 collaboration

Workshop on Astroparticle Physics, Bose Institute, Darjeeling,10 - 12 December 2009

Scintillator Detectors (ADC+TDC)

400 + ………721

Proportional Counters (muon hit+ pulse width)

3712 + ………7424

Scint. Count Rate

Monitoring

DAQ(1GB/day)

EASDAQ

(8GB/day)

PC Count Rate monitoring + muon angle

DAQ(5GB/day)

GRAPES-3 Data(14GB/day)

GRAPES-3 DATA

With expanded array the data size ~ 40GB/day or 15 Tera Bytes/year

Mandate● Large storage space and computing power● Efficient monitoring of detectors● Ease of accessing data ● Parallel approach to develop data analysis,

detector monitoring software with participation of bigger team

● Portability of data to wider collaboration

Object Orie

nted Approach

Object Oriented Approach to GRAPES-3 Data ● Adaptation of object oriented language C++

● Object oriented design of all analysis programs in form of classes under ROOT framework.

● Storage of event data in ROOT Tree structure which provides efficient access of data for analysis.

● ROOT provides excellent graphical connectivity to the data object

The biggest advantage of OO design is, ease in managing large codes and lot of scope for any future developments. Not so easy in a procedural oriented approach.

Our ApproachStep 1: Conversion of binary data to ROOT for various data streams like

scintillator and muon detector data for EAS, scintillator rate monitoring

data, muon monitoring data and weather data.

Step 2: Various monitoring plots to monitor the scintillator and muon

detectors using these ROOT files. Built intelligence in the program so

that program should pick out the abnormality behavior of the detectors.

Step 3: Make a table for abnormality based on monitoring output and the

calibration constants in more automated way.

Step4: Use this table and the root data to reconstruct various shower

parameters like core location (Xc, Yc), arrival direction (, ), shower

size Ne, shower age S, number of muons N. CORSIKA + GEANT

simulation to convert Ne to E. Store them in ROOT tree.

Huge amount of programming effort required to reach step 4.

root [2] scevtree->Show(0)======> EVENT:0 runno = 13528 eventno = 1430 trigger = 2 evdate = 20050201 evtime1 = 0 evtime2 = 32140000 evstatus = 1 ndet = 33 detno = 13, 29, 44, 49, 51, 58, 73, 86, 94, 107, 124, 134, 141, 149, 151, 153, 170, 180, 193, 237 adchh = 1590, 226, 219, 290, 323, 528, 814, 512, 512, 604, 575, 477, 772, 640, 771, 812, 458, 1487, 566, 669 adchl = 213, 31, 32, 41, 46, 71, 108, 67, 68, 78, 74, 61, 101, 83, 101, 106, 59, 196, 76, 90 adclh = -1, -1, 328, -1, -1, 205, -1, -1, 156, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 adcll = -1, -1, 40, -1, -1, 24, -1, -1, 17, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 tdc = 1034, 4095, 960, 4095, 4095, 4095, 832, 4095, 4095, 4095, 4095, 4095, 879, 826, 1069, 959, 4095, 4095, 4095, 4095

ScEventTree

EventHeader ScDATA

runno

eventno

trigger

evdate

evtime1

evtime2

evcurdtime

evabsdtime

ndet

detno

adchh

adchl

adclh

adcll

tdc

tdctype

Tree structure for scintillator data

Root[0] scevtree->Draw(“tdc>>h1(4095,0,4095)”, ”detno==1 && trigger==2&& evtime>010000 && evtime <=020000”)

Interactive Debugging

Self Trigger

Monitoring Tools● Automated Analysis of Scintillator Detector

Calibration data with Muons● Remote Monitoring of each Scintillator detector

in the array– Performance of detector and DAQ can be

monitored by any of the GRAPES-3 collaborator on daily basis

● Remote Monitoring of Muon Detector

Detectors calibrated by generating muon trigger using two paddles placed below scintillator

Several detectors calibrated in dayby moving paddles to different detectors

Algorithm developed to identify calibrated detectors in the data

Muon data analysed to get calibratedparameters and stored into thedatabase

Scintillator Calibration with Muons

Diagnostic Parameters for good and Bad Detector

Remote Monitoring

● The monitoring plots and logs uploaded to a common gmail account on daily basis.

The remote shifter check the plots, enters the detail of the problem and his feed backs

to an excel file and sends back. The feedbacks very useful to take necessary action

taken by the people at the experimental site.

● Remote Shifters at present

Supriya Das, Sumana Das (Bose Institute, Kolkata), Sonali Bhatnagar (Dalbag

Institute, Agra), S.R. Dugad, S.K.Gupta, P.K. Nayak, P.K. Mohanty, S.D. Morris

(Mumbai)

We are expecting more participation in this activity from the collaborating institutes.

Summary

● The Object Oriented design of GRAPES-3 data analysis software is robust and efficient

● GRAPES-3 ROOT framework is a team effort.

● ROOT framework implemented for scintillator and muon data

● Reconstruction program ready in ROOT framework

● Plans for online reconstruction, online alert for solar activity.

● Still many more things to be developed. Needs involvement of more people.

THANKS

GRAPES-3 Data Analysis Architecture

ROOT DATA

SC RAW DATA

Monitoring

Shower Reconstruction

Shower Reconstruction

GEANT4

CORSIKAShower Parameters in ROOT(Xc,Yc, Ne, S, , , E, N)

Ne – E Relation

Calibrations + Bad data summary

MU RAW DATA

ROOT DATA

Event Matching

Muon Reconstruction

-ray astronomy

Energy spectrum and composition

GRAPES-3 Scintillator Data Structure

RUN

Event Event Event Event Event

Trigger ADC TDCTime

Det1

Det2

Det1

Det2

~8000/RUN

~ 350/day

Shower Reconstruction● Arrival direction (,) reconstruction using plane fit and cone

fit

● Shower size Ne, Age S and Core location (Xc,Yc) by fitting NKG function using log likely hood method

● ROOT TMinuit class for minimization

● CORSIKA for shower simulation and GEANT4 for detector

simulation.

GRAPES-3 analysis code summary ● The code consists of

– 25 classes

– 30 000 lines

Scintillation Array Monitor● Detector Monitor Parameters

– Noise: Pedestal Mean/RMS– Uniformity of Analog Data: Signal Rate, Mean and RMS– Signal Timing (TDC): Rate, Mean, RMS

● Determine all 8 parameters for each run in a day– Obtain performance index (PI) of each parameter

Signal Rate

P.I

. (0-

100%

)