Simulating the human heart is an interesting and challenging problem.
For clinical applications we would like to simulate a whole heart in
detail, but this would take days to compute even on high performance
computing resources. Reliability is also a concern - simulation
software should be easily testable (against empirical data) and
maintainable, which is often not the case with extensively
hand-optimised codes. There is thus a conflict between developing
conceptually clear models, and efficient implementations of models
for simulation purposes. This talk will describe our initial attempts
at using automatic optimisation to resolve this conflict, outline how
we will formally prove that our optimisations are not only valid but
also improve efficiency, and discuss our plans for using such
techniques to aid simulation steering.

2) Dr. Yudong Sun , ComLab, University of Oxford

A Grid-enabled Multiscale Biomolecular Simulation Model and BioSimML
Markup Language

Biomolecular simulations are useful to relate static protein
structures to dynamic functional properties. The simulations can be
conducted at atomistic or coarse-grained scale. An atomistic
simulation can model the details of a biological process but are
highly time-expensive. A coarse-grained simulation is time-efficient
but cannot reveal the atomic details of a biological process. To
combine the advantages of such simulations, we have developed a
multiscale simulation model that integrates the simulations at both
atomistic and coarse-grained scales. The model supports dynamic
integration and parallel execution of the simulations at two scales.
The model enables automatic workflow management, resource-aware task
allocation, and distributed simulation transformation on a grid-based
system. A BioSimML markup language has been developed to capture the
data representation and formulate the data exchange between
multiscale simulations. An experimental simulation on a viral
membrane fusion peptide demonstrates the considerable performance
gains of our simulation model.

This page last modified 4 December, 2006 by [ ]