School Seminars and Colloquia

Decoupled equations for studying Complex Systems


by Dr Rob Moss


Institution: The University of Melbourne
Date: Thu 7th May 2009
Time: 4:15 PM
Location: Room 225 James Hardie Theatre, 2nd Floor Architecture Bldg, The University of Melbourne

Abstract: A complex system comprises many interconnected units, whose behaviours are both
variable and dependent on the behaviour of the other units. Complex systems
exhibit emergent dynamics that may be deterministic, but which have properties
that can only be examined at a higher level than the individual parts. These
dynamics (e.g., the dynamics of the system as a whole) cannot be understood by
studying the units in isolation -- the system must be studied holistically.


In order to study emergent dynamics in the kidney, Dr Moss developed a
modelling approach ("hierarchical dynamical networks") that explicitly
represents both the inherent system structure and the interactions between
units in the system. With this approach the unit dynamics are captured by
difference equations rather than differential equations, and couplings between
units are represented as an explicit network rather than a set of coupled
equations. This decoupled approach results in analytical and computational
benefits when studying emergent dynamics.


By using this approach, Dr Moss was able to simulate emergent dynamics in the
kidney for far larger numbers of units (called "nephrons") than was possible
with existing models, and performance measurements indicate that this is the
first such model for which the simulation of whole-kidney function is
tractable. The model is also the first to have been used to study the
functional stability of the kidney in response to impaired function (such as
disease or injury).


In this talk Dr Moss will introduce this modelling approach in the context of
the kidney, contrasting his kidney model with existing models, which are
formulated as systems of differential equations. Some key results obtained from
the kidney model will also be presented. The talk will conclude by discussing
the suitability of this approach to other complex systems.


For More Information: contact Kerry Landman: email kerryl@unimelb.edu.au