Transcript Document

Atlantis tutorial
Hans Drevermann (CERN),
Janice Drohan (UCL),
Charles Timmermans (KUN)
4 March 2004
ATLAS Software Week/CERN
Outline
•
•
•
•
•
•
•
•
•
Introduction
Data visualization – Hans
Basic concepts (presentation) – Janice
Hands on demonstration of basic functionality
User session
---- Coffee ---Access to events – Janice
Advanced features (presentation) - Charles
Hands on demonstration of advanced features
Questions & answers
Atlantis goals
Primary
• visual investigation and physical understanding of
complete Atlas events.
Secondary
• help develop reconstruction and analysis
algorithms
• debugging during commissioning
• pictures and animations for publications,
presentations and exhibitions
• event display for simple test-beams
• online event display
Data
The following data may currently be visualized by the program
•
•
•
•
•
•
•
•
•
3D silicon points, silicon strip clusters and TRT straws
Si Geant Hits, Trigger space points
Simulated tracks, neutral particles and vertices
Reconstructed tracks iPatRec, xKalman, IDScan
Reconstructed secondary vertices
LAr, TILE, HEC and FCAL calorimeter cells and clusters.
MDT, RPC, TGC, CSC hits and CSC clusters
Simulated and reconstructed muon tracks (MOORE)
Associations (hit to track, cell to cluster etc)
Detector/Data oriented projections
3D Cartesian coordinates x,y,z are not always optimal for
colliding beam experiments
More natural and useful are the non-linear combinations which
reflect the design of ATLAS
r = (x2+y2),
f = atan2(y, x),
h = log( z/r + ((z/r)2+1) ,
where x, y, z need to be slightly modified to take into account
the primary vertex of the underlying event (xvtx yvtx zvtx )
x' = x-xvtx , y' = y-yvtx , z' = z-zvtx
Y/X projection – TILE, LAr barrel, RPC (intuitive)
RPC
f/r projection – like Y/X, but prompt tracks are straight lines
r/Z projection – calorimeters, muon hits (rsector) (intuitive)
X'/Z projection – muon hits and their association to Moore tracks
f/Z projection – TRT and LAr endcaps, HEC, TGC phi strips
TGC
TRT
f/h projection and the V-Plot
f
r = rMax
V-Plot
Draw each space point
twice at
F,h+k*(r-rMax)
and
F,h-k*(r-rMax)
r = 2 cm
h
low p, -ve
Distorted V’s
track not from IP
high p, +ve
 3D information
For tracks can judge
•f
•h
• pt (slope of V arms)
• charge ( L -ve
V +ve)
f/h projection – track to calorimeter associations (30 GeV electron)
LAr Presampler
Pt=29.3 GeV
E =31.2 GeV
Cell geometry
LAr Layer 1
LAr Layer 2
LAr Layer 3
Track
(enters LAr here)
Area a E
Island (guides eye)
User interface
Menu bar
Window Control
(zoom,copy, DnD)
(IO,preferences,help)
Interaction Control
Parameter groups
Commands
Parameters
Output window
Online help – available for every component
Right click on component for online help (hyper-linked HTML)
Hover for
tool-tip help
Interactions
•
•
•
•
•
•
•
ZMR - zoom, move and rotate w.r.t defined center
Rubberband - selection and zooming
Pick - pick and move to (selection and query)
Fisheye - relative expansion of central region
Clock - relative expansion of angular region
Synchro-cursors - correlation between different projections
Scale - copy scales between windows
Mostly mouse driven with sometimes a
modifier key pressed on the keyboard
Input Data
Atlantis is a JAVA application
It communicates with Athena via dedicated XML files
produced by JiveXML ( see talk later)
These files are best grouped and compressed inside zip files
Single design luminosity event is approx 20 MB (XML)
4 MB (zip)
Detector Geometry
• Used to convey quickly to the user the context in
which hits are to be viewed.
• Idealized geometry is adequate and desirable. (e.g.
LAr pre-sampler is only 1 cm thick and would be
invisible if drawn as such
• Stored in two separate XML files.
• muon geometry derived from parameter book.
Printing
File => print => EPS, PNG, GIF
EPS – high quality vector graphics
good for posters, publications
(file size 200KB - 2MB)
PNG – compressed bitmap
good for ppt presentations + web
(file size 20-50 KB)
AtlFast
User defined geometry (e.g. MDT - cosmic test stand)
Web page
• www.cern.ch/atlantis
• How to download, install and run Atlantis
• Picture database (example event displays)
• Presentations
Contributors
Many people contributed to the development of Atlantis. In particular:
Gary Taylor (UC Santa Cruz)
Hans Drevermann (CERN/EP)
Dumitru Petrusca (Siegen/CERN)
Jon Couchman (UCL)
Frans Crijns (Nijmegen)
Peter Klok (Nijmegen)
Current Developers:
Janice Drohan (UCL)
Charles Timmermans (Nijmegen)
Principal developer
Original ideas, FORTRAN version
Initial work on GUI, calorimeters
Athena algorithm (JiveXML)
Muon geometry
Picture database
Atlantis tutorial-2
Hans Drevermann (CERN),
Janice Drohan (UCL),
Charles Timmermans (KUN)
4 March 2004
ATLAS Software Week/CERN
Analysis Techniques
Data to be viewed may be
Cut - e.g. by pT, energy, association…
Colored - by associations , layer, sub-detector
more powerful when used in combination
e.g. selected only hits belonging to kine tracks
and color them by their associated reconstructed track
( inconsistencies indicate problems)
Superimposed – iPatRec tracks over true tracks
Check track reconstruction in difficult design luminosity event
• Selected event has two high pt (>560GeV) jets ( DC1dataset 2045)
• Luminosity
• Silicon space points
• TRT hits
• Reconstructed tracks
• Reconstructed in
1034
27,000
240,000
120
20 minutes
2D projections of Inner Detector data not very useful at design luminosity
(TRT -ve barrel only)
V-Plot silicon space points
Fh calorimeters
Filtering of space points available inside Atlantis
Filter space points
with a histogram
based technique
which selects hits
consistent with
tracks originating
from the primary
vertex.
Time = 1 sec/event
ATLAS note in
preparation
Filtered hits
iPatRec tracks
36 tracks
440 GeV
25 tracks
222 GeV
True tracks
34 tracks
410 GeV
27 tracks
270 GeV
Tracks lost in core of central jet
STr,iPat,S3D(STr,iPat)
iPat, xKal
iPat,S3D(Filter,iPat)
Lists
Up till now we have seen how to investigate
data and association present on the input file.
Lists allow user to dynamically create and
manage their own associations
grouping of object
perform context dependent operations
e.g. vertex a set of reconstructed tracks
Identifying secondary vertices
Look for a group of nearby kinked V’s in the VPlot
True tracks
Reconstructed tracks
D
B
Y/X projection – region around the primary vertex
• Region around primary vertex
B
D
3s error
ellipse
Reconstructed tracks
Primary
True tracks
Secondary vertex region best displayed in abstract 3D Box
Plane
containing
primary
vertex
Plane
containing
secondary
vertex
secondary
vertex
primary
vertex
Ellipses
represent
track
error (1s)
Comparison of muon and inner detector track fits
f
p = 28 GeV
Dp = 5 GeV
Df = 1.5 deg
Dh = 0.02
muon
tracks
Inner detector
tracks
p = 25 GeV
Dp = 4 GeV
Df = 1.0 deg
Dh = 0.01
V-Plot allows comparison
of f, h and pt
h
Cell clustering and Jet reconstruction – AtlFast (DC1- QCD event)
Cells coloured by cluster (Area a E)
Jet (Area a E)
E= 13 GeV
Bug ?
SE= 347 GeV
Details of cell
clustering
?