Software

Software Status
M. Ellis - MICE CM18 - 16th June
2007
Malcolm Ellis – MICE
CM18 – 16th June
2007
1
Question from Alain’s Talk:
“Goals of meeting
-- status of software and analysis: are we ready to receive data?
My pet questions:
-- can we get “online” results ? How reliable can they be? “
 Executive Summary of Answer:

No
 “Yes Minister” Answer:

Yes and No
M. Ellis - MICE CM18 - 16th June 2007
2
Outline

Recent Progress and Current Status








Urgent/Important work needed





GRID
PID Detector Code
Online Monitoring
Data Base
Short and Medium term plans





Decoding (SciFi – Aron)
Simulation (CKOV – Yordan & Hideyuki)
Event Display (Michael)
Reconstruction (SciFi – Aron and Tamas)
Optics/Analysis (Chris)
G4MICE G4MANX
Tests
Complete reconstruction release
Focus on online/data taking needs
Software Workshops
Data Taking & Online Analysis – a preview
Conclusions
M. Ellis - MICE CM18 - 16th June 2007
3
Recent Progress
 Similar report to past CMs:


Continue to make progress in all areas of interest.
Continue to be limited by very small number of effective FTEs.
 Good news:

Several new people are making an increasing contribution:




Tamas Sashalmi (IC)
Yordan Karadzhov (Sofia)
Michael Wojcik (IIT)
Ben Freemire (IIT)
 Bad news:


Still very few people in MICE who can use the software to
perform routine simple tasks (e.g. produce a histogram in order
to do a Monte Carlo study or calibrate a detector element).
While simple computing tasks are being done by developers,
development of code is held up (and vice versa).
M. Ellis - MICE CM18 - 16th June 2007
4
SciFi Decoding
 Ongoing saga that has been recently resolved for the current
iteration of readout electronics and waveguides.
 Aron has produced a number of decoding files for the various
tests made of Station 5 and with assistance from Michael
Wojcik has verified that it is correct.
 N.B.

We also have decoding code and files for the KEK test beam for
the TOF system used there but have no input from the other
detector groups as to how their detectors will be read out,
cabled, etc and hence no work has been done on the “decoding”
of the other detectors!
M. Ellis - MICE CM18 - 16th June 2007
5
Simulation
 CKOV and TOF Simulations have been improved by Yordan
(continuing work started by Hideyuki).
 The CKOV detector geometry looks reasonable and emphasis is
now moving towards implementing a reasonable model of the
transport and collection of the Cherenkov light by the PMTs.
 Aim to simulate the test beam data that was used for
selection of the aerogel in order to validate this aspect of the
simulation.
 TOF simulation already existed
but has been improved by Yordan.
 He has also checked and fixed
the digitisation and reconstruction.
 So once we can unpack the TOF
data fully, it can be reconstructed.
M. Ellis - MICE CM18 - 16th June 2007
6
Event Display
 HepRep based event display described in past CM talks.
 Slowly adding information to the output, now focusing on
reconstructed information (hits, clusters, slabs, points,
tracks, etc).
 Michael recently added visualisation of SciFiDoubletClusters
(i.e. a fibre or pair of fibres in one plane of a SciFi tracking
station).
 Next step will be to visualise reconstructed tracks and fix
known issue with visualisation of simulated particles.
M. Ellis - MICE CM18 - 16th June 2007
7
Station 5 Cosmic Ray Events
M. Ellis - MICE CM18 - 16th June 2007
8
SciFi Reconstruction
 Migrated to new version of RecPack Kalman fitting package
(used to use old version used in HARP, now changed to
redesigned package that is undergoing development and which
is supported).
 Pattern Recognition written by Aron (described at last CM)
provides a seed to the Kalman Fit.
 One known problem (incorrect determination of the process
noise Q matrix for the correct treatment of MCS) otherwise,
new version works well.
 Release of new track reconstruction was made at the end of
the software workshop that finished on Tuesday. This is being
used by Tamas to verify the baseline station spacing now.
 Once MCS problem is fixed, will add field map use and will
then be able to do global matching and return to work on PID.
M. Ellis - MICE CM18 - 16th June 2007
9
Track Fit
M. Ellis - MICE CM18 - 16th June 2007
10
Optics/Analysis/General Black Magic
 Tests for BeamTools

Solenoid, Dipole, RFMap, PillBox, SolenoidMap
 Simulation integration tests


Field placement, engineering model, phasing
Not yet automated
 Partial implementation of sector dipoles for MANX
 Beam generation algorithms

Amplitude momentum correlation, momentum dependent beta, fully
automated linear matching
 Optics upgrades


Second order transport
Third order transport (ongoing)
M. Ellis - MICE CM18 - 16th June 2007
11
G4MICE -> G4MANX
 As described at past meetings, it is possible to describe quite
complex geometries in G4MICE now without writing a single
line of C++ code.
 This has allowed some colleagues from the MANX experiment
to develop a “configuration” which contains two standard MICE
trackers but places the helical cooling channel and matching
sections in the middle.
 This is currently being used by them to understand and fix
some problems in the matching section and prepare results for
a proposal that is due soon.
M. Ellis - MICE CM18 - 16th June 2007
12
G4MANX
M. Ellis - MICE CM18 - 16th June 2007
13
Tests
 MICETESTS area continues to grow and become a part of the
MICE software team’s culture.
 The SciFi Recon tests (space point, PR and Kalman fit) were a
crucial part of the preparation of the latest release and will
continue to serve to monitor future releases to ensure that
the performance is not degraded.
 Recent additions include a more detailed check of the material
properties used throughout G4MICE (so allow us to see if
there are any changes in properties when we change G4, for
example).
 Currently in the process of adding a geometry checker that
validates all MICE models and ensures that there are no
problems (e.g. a tracker station sticking outside of the
solenoid volume).
M. Ellis - MICE CM18 - 16th June 2007
14
Urgent/Important Work
 Efforts on the simulation and reconstruction are beginning to
ramp down (at least for now).
 With only a few important tasks left (and apparently little
serious desire by the collaboration to use the tools...) the
focus will now shift to the needs for data taking and analysis.
M. Ellis - MICE CM18 - 16th June 2007
15
GRID

Very little progress since last CM basically due to other calls on my
time.

Lack of computing support at RAL for MICE continues to concern me
(and I’m not just referring to the lack of network access during
CMs!).

CSF at RAL still does not recognise the MICE VO (despite a request
placed in February), nor do any of the RAL GRID machines support
the MICE VO.

The list of institutes that support the MICE VO has not changed
since CM17 (Sofia, QMUL, RHUL, Brunel and ScotGRID).

I’ve requested that Sheffield and Imperial provide resources for the
MICE VO, but have not succeeded yet.
M. Ellis - MICE CM18 - 16th June 2007
16
PID Detector & Monitoring Code
 Simulations of the PID detectors are making progress,
however there has been no effort on the code required for
decoding and analysing real data.
 We also have no scheme for accessing or analysing the data
that will presumably be recorded as part of the controls and
monitoring system (see slides on DataBase later)
 An opportunity to start this work was not taken up when the
Frascati test beam data was analysed.
 Assuming the code to unpack the data format produced by
DATE exists (and it will), we still need code and data files for
things such as:



Cabling (i.e. what electronic channel is connected to what physical
part of a detector).
Calibration (e.g. pedestals, gains, t0s, etc).
Clusterisation, etc if required (e.g. flash ADC vs ADC + TDC).
M. Ellis - MICE CM18 - 16th June 2007
17
Online Monitoring






Recently had the opportunity to get help on the online monitoring
from Akram Khan (Brunel).
He requested a ROOT file containing samples of the kind of
histograms we might use for monitoring and offered to produce a
GUI interface.
I was unable to do this for him (too busy with tracker hardware and
other software tasks) and until recently was unable to find anyone
else (other software people already overloaded and no-one else can
do something like this).
Paul Kyberd has since taken up the challenge and during the
collaboration meeting has produced a simulation and written a
G4MICE application to produce a set of histograms for Akram.
I use this example to illustrate the problem that I fear we will soon
face to a much greater extent.
Paul will soon circulate a questionnaire to the collaboration to try to
work out in detail the parameters and plots that are needed.
M. Ellis - MICE CM18 - 16th June 2007
18
Data Base
 Requested requirements from the experiment during/after
CM17.
 Received responses from:




Tracker group
CKOV (Lucien)
Target (Paul)
Analysis (John)
 Decision has been taken to use PostgreSQL.
 This decision will be documented as time allows.
 Does this need to be checked/agreed on by CB, TB, EB, etc?
M. Ellis - MICE CM18 - 16th June 2007
19
Short and Medium Term Plans
 Complete remaining simulation and reconstruction code
development and testing (mostly tweaking a few details,
ensuring adequate test coverage and double checking results
of tests).
 Write unpacking code to read the files produced by DATE.
 Generate a reasonable set of monitoring histograms.
 Start implementation of online DB.
 Support detector commissioning.
 Continue attempts to train new users of G4MICE.
M. Ellis - MICE CM18 - 16th June 2007
20
Software Workshops
 Next workshop:
15th – 17th August at Fermilab.
 Followed by:
3rd – 5th October at RAL.
 Current plans are not finalised for the next few, but looking
at the following:




November
December
January
February
(Brunel or RAL)
(Fermilab)
(RAL)
(with CM20)
 Increase in frequency is for two reasons:


Higher average productivity of the small software group when we
can work together in a focused manner.
Continue to use workshops to train new users in the use of
G4MICE. Hopefully as data appears more people will develop an
interest in producing plots and doing an analysis of some
description.
M. Ellis - MICE CM18 - 16th June 2007
21
Tracker in G4MICE
 G4MICE has been used to analyse over 10 million events of
tracker related data (test beam, cosmic ray, radioactive
source and LED pulser).
 Applications have been generated that perform tasks such as:




Histogramming physical quantities.
Reconstructing and visualising points and tracks in the prototype
tracking stations
Performing fits for calibration.
Determining light yield, looking for dead channels, etc.
 (Note this does not include the various Monte Carlo studies
that have been performed or are ongoing).
 While tasks such as emittance calculation have so far only
been performed on simulated data, a reconstructed track
produced by G4MICE can be used for emittance calculation in
the same manner whether it was derived from simulated or
real hits.
M. Ellis - MICE CM18 - 16th June 2007
22
Data Taking & Online Analysis
 In order to (finally) answer Alain’s question, the next few
slides will describe how the present state of the software
could be used to analyse the MICE data.
 Almost all of the tools required for an analysis of the MICE
tracker(s) already exists in G4MICE. The rest of the
experiment is not in such a healthy state.
 Data is produced by the DAQ (DATE).
Does not exist now, but will do so soon (VLSB readout code -> DATE)
 Code in the RealData area is used to generate VlpcHits.
Does not exist now, but will be written in parallel with DAQ debugging
 Code in the Reconstruction area makes SciFiDigits from the
VlpcHits.
Code already exists, just need new cabling file for final tracker √
M. Ellis - MICE CM18 - 16th June 2007
23
Data Taking & Analysis Example
 Code in the Reconstruction area makes SciFiDoubletClusters
Code already exists, just need calibration files √
 Code in the Reconstruction area makes SciFiSpacePoints
Code already exists, nominal geometry already known √
 Code in the Reconstruction area makes SciFiKalTracks
Code already exists, nothing further required √
 User code extrapolates reconstructed tracks to required
surfaces (e.g. TOF planes)
Code already exists, can be used as an example for whatever is required
 User code generates histograms of desired values
Code already exists, can be used as an example for whatever is required
√
√
 Calculate emittance of beam, SPE distributions, etc
Code already exists, can be used as an example for whatever is required
M. Ellis - MICE CM18 - 16th June 2007
√
24
Examples
M. Ellis - MICE CM18 - 16th June 2007
25
Examples 2
M. Ellis - MICE CM18 - 16th June 2007
26
Examples 3
M. Ellis - MICE CM18 - 16th June 2007
27
Examples 4
M. Ellis - MICE CM18 - 16th June 2007
28
Conclusions





We continue to see a slow but steady increase in the number of
people AND FTE available to G4MICE development.
We still lag behind where I would have liked to have been in terms
of general MICE users of the software.
Focus is now turning towards the online needs for detector and beam
line commissioning and analysis in the coming year.
We will be able to reconstruct the tracker when it is commissioned.
I’m hopeful that the TOF and CKOV will also be available, but that is
not obvious at the moment.

I continue to encourage everyone to attend a workshop to learn how
to use G4MICE (you don’t need to know C++, GEANT4, etc in order
to make histograms).

I am also happy to travel (within other work constraints) to hold
workshops, software tutorials, etc if invited, so let me know if you
would like to organise something.
M. Ellis - MICE CM18 - 16th June 2007
29