Modelling and Analysis of Large Complex Systems

Download Report

Transcript Modelling and Analysis of Large Complex Systems

Introducing the
Prof. Marta Kwiatkowska
Launched 7th May, 2003
www.MeSC.ac.uk
Overview
 The Midlands e-Science Centre
– Area of Excellence Modelling and Analysis of Large Complex
Systems
– Applications focus, rather than Grid middleware
– Hope to work with Grid middleware developers…
 Partner institutions
–
–
–
–
University of Birmingham
University of Warwick, Centre for Scientific Computation
University of Coventry
University of Wolverhampton
 Infrastructure and resources
 Projects
 Next steps
Complex systems
New field of science - study how parts of a system give rise to the
collective behaviours, and how it interacts with its environment.
Social science, medicine, weather, engineering, economy, management...
Meeting the complexity challenge
 Why study and analyse?
– knowledge, discovery, prediction
 Sources of complexity
–
–
–
–
millions of components
huge data sets
interaction, motion in space
unpredictability
–
–
–
–
mathematical modelling
computational modelling, simulation
high-performance visualisation
collaboration
 Solutions
 Delivery via e-Science
– harness the power of global computer
– answers in real-time
Model

Simulate

Predict

Control

Avoid disaster
The Midlands e-Science Centre
 Virtual Centre
– open, possible still to join
 University of Birmingham
–
–
–
–
–
–
–
–
–
home Computer Science
Physics and Astronomy
Chemical Sciences
Biosciences
Engineering
Geography, Earth and Env. Sci.
Mathematics and Statistics
Medical School
Information Services
 University of Warwick
– Centre for Scientific Computing
 University of Coventry
 University of Wolverhampton
MeSC objectives
 Connect the Midlands
– provide accessibility and connectivity for the Grid for the
Midlands region
 Excellence in Complex Systems
– focus on modelling of very large complex systems
– act as source of relevant expertise for industry
 Enable long-term research
– numerical algorithms
– simulation techniques for the Grid
 Foster collaboration
– different disciplines in science and engineering
– academics and industry
Research at MeSC
 Research themes
–
–
–
–
Simulation of evolving systems of interacting components
Large-scale Grid-enabled distributed simulation
Mathematical solutions of large complex systems
Data mining and large-scale visualisation
 Hope to stimulate crossover of techniques
–
–
–
–
–
from
from
from
from
etc
evolutionary techniques to organisation management
physics motion models to understanding mobile processes
concurrency formalisms to modelling particulate processes
algorithms research to bioinformatics
People at MeSC
 Management Board
–
–
–
–
–
–
–
–
–
Marta Kwiatkowska, CS, Director
Peter Watkins, Phys
Peter Knowles, Chem
Georgios Theodoropoulos, CS
Andrew Chan, Eng
John Owen, IS
Peter Taylor, CSC, Warwick
Keith Burnham, Eng, Coventry
Richard Hall, Eng, Wolverhampton
 Technical/User Support
– Paul Hatton, IS
– Steve Jarvis, CS, Warwick
– PDRA (offer made)
 Many more existing/potential collaborators
Infrastructure
 Networking
– High-speed campus network, multi-million pound investment
(SRIF and University)
– midMAN
 Computing facilities
– SRIF-2 funding, £200K, currently considering future strategy
– About to purchase dedicated cluster for e-Science Centre
– HPC facility at Birmingham, and various clusters
 Access Grid Node
– at Birmingham (2x), Warwick and Wolverhampton
– for virtual meetings and and collaboration
 VISTA
– State-of-the-art visualisation centre
Visual and Spatial Technology Centre
 Set up in partnership with HP
 £4M investment
 Association with several
industrial partners (AVS, CFX,
Fakespace, etc)
 Scientific visualisation
– geodata, medical imaging
 Information visualisation
– knowledge discovery
 Data representation
– understanding complex data
 Immersive environments
www.vista.bham.ac.uk/index.htm
Part of the
internal
structure
of a
hydrogen
atom.
Image
fusion
of a
series
of
MRI
scans.
Complexity in… Hardware Design
Microprocessor
Size 7.5x3.5mm
Millions of transistors on chip
Errors found after manufacture
(cf Intel)
 Research in Modelling and Analysis of Systems Group
–
–
–
–
distributed simulation to assess performance
automatic verification to ensure no design errors
also can find errors in software (security protocols, etc)
funding from EPSRC, DTI, QinetiQ, BT, EU
 The Grid technology enables
– larger models, faster analysis, improved reliability
– reduced costs & time to manufacture
www.cs.bham.ac.uk/research/systems/, www.cs.bham.ac.uk/~gkt/Research/par-lard/
Complexity in… Social Science
 Managing complex social scenarios
– develop new ways of thinking about
social processes, modelling and complex
organisations (e.g. hospitals)
– uses agent technology and evolutionary
computation
– real-time disaster management
response with the Grid
 Research in Natural Computation
Group
– also includes neural networks, evolvable
hardware, self-organising systems, ...
– funding from EPSRC, EU, Advantage
West Midlands, Marconi, Honda
www.irit.fr/COSI/, www.cs.bham.ac.uk/research/NC/
Real situation

Model

Agent-based simulation
Complexity in… the Human Genome
 Modelling of biology of
immune response
– large-scale genomics
– data mining,
computationally intensive
– modelling physiology of
the immune response
– understanding molecular
basis
 Research in
ImmunoGenomics Group
– gene expression profiling,
infection modelling
Cancer Research
Components of a probabilistic model
describing a lymphocyte in a chronic
inflammatory disease
– childhood cancer
www.irit.fr/COSI/, www.cs.bham.ac.uk/research/NC/
Complexity in… Urban Pollution Control
 Difficult to model
– air movement in street
– effect of road dust
 The Grid technology
– better accuracy
– feasibility of response
on regional/national
scale
Concentration of pollutants in street lanes
 Research in Climate and Atmospheric Research and Wind
Engineering Groups
– various project concerning the effect of wind, turbulence,
dispersion of particles, etc
– large eddy simulation
– funding from NERC, EPSRC, industry
www.ges.bham.ac.uk/research/physical/Atmospheric/atmospheric.htm, www.eng.bham.ac.uk/civil/
Complexity in… Fluids and Flows
 Modelling bubble formation
– relevant for laser surgery,
bubble contrast agents in
ultrasound imaging, underwater
explosions, water waves, ship
bow waves, etc
– computationally demanding, would
benefit from the Grid
 Research in Applied Mathematics
Group
– also detonation and flame
processes (Fuel Cells, to be
displayed at Royal Society)
– cancer modelling
– funding from EPSRC, Kodak,
Unilever, Nestle, Pilkingtons, etc
www.mat.bham.ac.uk/research/applied/applied1.htm
Laser-generated bubble near
boundary
Complexity in… Granular Substances
Pharmaceuticals,
foods,
powders,
aerosols,
soils, ...
 Modelling and Simulation (DEM) of Particulate Processes
–
–
–
–
discontinuous, composed of many millions of particles
particles interact in various ways
aim to calculate properties of substance: elasticity, texture, feel
Grid technology needed because of sheer scale of models
 Research in Chemical and Civil Engineering
– funding from EPSRC, Cadbury, Unilever, BNFL
www.eng.bham.ac.uk/chemical/
Complexity in… the Universe
Einstein’s Theory of
General Relativity
Mass-energy produces
space-time warpage
Black hole collisions,
Supernovae,
The Big Bang, ...
Gravitational waves are
time dependent
gravitational fields
produced by the
acceleration of masses.
Colliding black holes (courtesy NCSA)
Gravitational Waves and e-Science
 Measure the stretch
and squeeze of space
with light beams,
approx. 10-16 cm
 Signals drastically
dominated by noise
 Extract signals from
the noise while keeping
up with the data flow
(approx. a few Mb/sec)
LIGO - Livingston
4km
 Research in Gravitational Waves Group
– partners in LIGO and LISA international scientific collaborations
– funding from PPARC
 Grid technology the only solution
www.sr.bham.ac.uk/research/gravity/, www.ligo.caltech.edu/, http://lisa.jpl.nasa.gov/
Research examples: Warwick
 New methods for
quantum-chemical
calculations
(Chemistry/Maths)
 Monte Carlo simulation
of condensed matter
(Physics/Statistics)
 Analysis of turbulence
simulations:
distributed data
visualisation via the
Grid (Eng/Maths/Computer Science)
Studying
molecular
properties
of
aromatic
systems
with
DALTON.
Simulation of
molecular
structures
and
interactions.
http://qcwizards.warwick.ac.uk/~taylor/research.htm, www.phys.warwick.ac.uk/molecularsim/home.html
Research examples: Coventry
 Control,
optimisation
 Industrial
collaborators
– Corus,
Jaguar,
Rolls-Royce,
TRW,
Walsgrave
Hospitals
NHS Trust,
etc
 Funding from
– EPSRC, DTI
and HEFCE
Control methods for improving
annealing furnace
Research examples: Wolverhampton
Simulation
of a new hip
and joint
replacement.
VR
simulation of
a prototype
gear
assembly.
Projects
 At Birmingham
–
–
–
–
–
–
GridPP
LIGO & LISA (GW) and STAR (Nuclear Physics)
Grid-enabled distributed simulation and numerical solutions
COSI (Complexity in Social Sciences, EU)
BioSimGrid
Integrative Biology (cancer modelling, fluid dynamics)
–
–
–
–
e-TUMOUR (EU FP6 IP)
Bioinformatics (Bioinformatics Regional Institute)
Randomised trials (Primary Care, national network)
Pollution modelling and control (Geography and Env. Science)
Projects continued…
 At Warwick
– PACE, Performance Analysis and Characterisation Environment
– Molecular modelling
– Turbulence
 At Coventry
– Biomedical engineering
– Industrial control, optimisation
 At Wolverhampton
– VR
– Simulation for manufacturing, SMEs
Next steps
 Infrastructure improvements
– AGN rooms, dedicated cluster, etc
 Application areas
– medical applications
– bioinformatics
– pervasive e-Science? (sensor networks, mobile wearable
computing)
– industrial solutions
– etc
 Collaborate and build on collaborations
– with other e-Science centres
– collaborate with e-Science ontology, workflow and visualisation
tool developers