Visualization for VOTech:

1
Visualization for VOTech Visualization_at_Leeds Mult
ivariate Data Visualization
Ken Brodlie School of Computing University of
Leeds
2
Background

• Involved in a number of UK e-Science projects
• Developing visualization middleware to provide a
  visual front-end to distributed and Grid
  computing
• Range of application areas from environmental
  science to computational biology
• gViz project has studied middleware to link
  simulation and visualization processes
• Simulation runs remotely
• Pollution dispersion as demonstrator application
• Plus collaborative visualization

3
Dataflow Visualization Systems

• Visualization represented as pipeline
• Read in data
• Construct a visualization in terms of geometry
• Render geometry as image
• Realised as modular visualization environment
• IRIS Explorer is one example
• Visual programming paradigm
• Extensible add your own modules
• Others include IBM Open Visualization Data
  Explorer

4
Imagine this .

• An explosion!
• A dangerous chemical escapes!
• Where is the fugitive pollutant headed?
• Who needs to be evacuated?

5
Understanding what will happen

• Model the dispersion by solving system of PDEs
• Understand solution by visualization

• What if scenarios need to be able to steer the
  simulation
• For example, what if the wind changes direction?

6
Linking Simulation and Visualization - Steering

• Computational steering
• By including a control module in the pipeline, we
  can direct the simulation in response to the
  visualization

PRO not only can we track, we can alter the
actual course of the simulation
Human-in-the-loop
Question for VOTech Is this a potential paradigm
for data mining?
7
Tracking the Pollution
8
Bring on the Grid!

• Real time computing is not fast enough for this
  application
• we need to predict the possible pollutant paths
  before they reach critical areas..
• So can we run the simulation module on a
  powerful remote compute node, keeping
  visualization on the desktop?
• Solution Grid-enabled IRIS Explorer

9
Harnessing Remote Compute Resources
Explorer on multiple hosts
Explorer on single host

• Automatic authentication using
• Globus certificate
• SSH Key pair

10
Simulation Runs Remotely
Here the simulation runs on Grid machine
but note it is often useful to run visualization
modules remotely too
Again in VOTech, we might mine on the Grid, vis
on the desktop
11
Gathering the expertise

• Environmental disaster!!!!
• We need to gather together group of experts..
• .. To understand the science
• .. and get the message to the politicians
• Again do it fast.. No time to physically collocate

12
Sharing the Visualization

• Extend the dataflow model to interlink pipelines
  across the Internet
• Each person has their own pipeline but they can
  share data
• Collaborative server provides the link
• So one user for example - can send geometry to
  another person for viewing

visualize
data
render
share
collaborative server
share
13
Programming the Collaboration

• It is useful to be able to program the
  collaboration
• To adapt to how people want to collaborate
• To adapt to network bandwidths
• Here raw data is exchanged so a different
  visualization can be created

visualize
data
render
share
collaborative server
share
14
Bring in the Meteorologist Remotely
Is there an analogy for astrophysical
data analysis?
15
Background

• In Integrative Biology we are applying the gViz
  middleware to help biologists study models of
  electrical activity of the heart

• Multiple simulations initiated and monitored from
  the desktop
• Here IRIS Explorer as front-end

16
Detaching the Simulation the gViz Library

• gViz library allows simulation writer to expose
  steering parameters and return results
• Simulation has life of its own, independently
  of visualization system
• Scientist can tap-in to monitor long running
  simulation

Grid resources
Grid Information
gViz-lib
gViz-lib
Researcher Desktop
This work is quite general gViz links back-end
computation with front-end visualization
no dependence on IRIS Explorer
17
Background

• Other front-ends can be attached for example,
  Matlab

• Or a secure Web service

18
Web Visualization Services

• Web technology offers us ways of delivering
  visualization services to the wider community
• Early demonstrator air quality data
  visualization
• HTML form as front-end, CGI script drives IRIS
  Explorer on server, VRML returned
• New era of Web services brings new opportunities

19
Visualization Web Service - WebSerViz
Haoxiang Wang
visualization.leeds.ac.uk8080/jsp/webserviz/form.
html
20
WebSerViz - typical output
Combination of isosurface and slices
21
WebSerViz Architecture

• Apache Tomcat
• JavaBean
• JSP

22
Grid Services

• Grid services add authentication to Web services
• Heart Modelling Grid Service uses
• Web interface where user specifies user name and
  passphrase, and location of gViz directory
  service
• gViz library to connect with simulations
• ImageService to build image from simulation data
• Returned as a Web page

23
Anatomy of the Heart Modelling Grid Service
24
Multivariate Visualization Hypercell

• Hypercell is an approach to visualization of
  multivariate datasets
• Developed by Selan dos Santos
• Basic concept
• Map each observation to a position in
  N-dimensional space
• Define an N-d region of interest, and a focus
  point within it
• Navigate through this space by an organised
  sequence of projections

• Applied to range of applications
• Astrophysics
• E-Learning
• Nonlinear optimization
• Concept implemented in IRIS Explorer
• Complement to existing techniques available in eg
  Xmdvtool
• Parallel coordinates
• 2D scatter plots
• Glyphs

25
Define the N-d region
26
Select the Projection
The user can select 1D, 2D, 3D or 4D
projections from the graph tool
27
Astrophysical Application

• Joint study with Bob Mann
• SuperCOSMOS Science Archive
• Only looked at subset of 57 attributes and 1000
  observations
• Analytical task
• Calibration of SSA data
• Look for expected and unexpected correlations
• and made us rethink some ideas!

28
Location of Source in Galactic Coordinates
Subspace (l, b, ebmv) with colouring by meanclass
attribute An outlier is evident
29
Location of Source in Galactic Coordinates
Removing the green class reveals the outlier
30
Location of Source in Galactic Coordinates
Same cell of data but coloured according to
prfstatb attribute. Most candidates to
be classified as stars are at bottom,
segmented In red
31
Magnitude Values of Sources
Subspace defined by (classmag(b-r1),
classmag(r1-i), classmagb))
Coloured by meanclass
Colour and size by prfstatri
32
Magnitude Values of Sources
Subspace defined by (classmag(b-r1),
gcormag(b-r1), scormag(b-r1))
Colour mapped to meanclass
Colour and size to prfstatr1
33
Relating Colour to Shape Attributes
Subspace (prfstatb, prfstatr2, prfstati)
Subspace (ellipb, ellipr1, ellipi)
Colour mapped to meanclass
34
Following On

• Need to record history of explorations in Nd space

• Could provide as a Web service

35
Xmdvtool

• Here are some student attempts at the same data
  using Xmdvtool

36
Ellipticity of sources
Parallel Coordinates
37
Ellipticity of sources
2D Scatterplot
38
Profile stat of sources
Parallel Coordinates
39
Profile stat of sources
2D Scatterplot
40
DS6 Developments

• Visualization
• Understand the data to be visualized
• Determine the appropriate technique
• Parallel coordinates
• Scatter plots
• Glyphs
• Visualization and Data Mining
• Understand the relationship
• Can we borrow ideas from computational steering?

• Visualization software
• Many existing systems
• IRIS Explorer
• IBM Open Visualization Data Explorer
• Vtk
• Integration with other Astrogrid/VO tools
• Delivery
• Web service
• Grid service
• Collaboration in project
• How do we exploit the different skills and
  experiences in the project, to maximum effect?