Complex Systems comp4006 / 7011 Project suggestions

Last updated 24/07/06

 

New tools allow us to look where others have looked, and see what they did not see. (paraphrased)

 

Requirements

 

The requirement for the project is to simulate and/or analyse an interesting complex system. This provides a lot of flexibility. Projects must involve simulations of complex systems using networks, cellular automata, agent-based models; or analyses of real world complex systems. For example, characteristics of real world networks, phase space analyses; [or something you can convince one of the lecturers is appropriate].

 

The projects are intended to be ~60 hours of work, including the planning and write-up, so the idea is to choose something concise but interesting. The most straight forward projects will involve using network analysis (eg Pajek) or cellular automata (eg NetLogo).

 

The following may give you ideas. Each one needs more work to flesh out into a full project.

 

1.      Calculate the collaboration graph for researchers in one of the schools or centres at UQ

a.       Draw the graph

b.      Calculate the network characteristics such as clustering, and diameter

c.       How do authorship patterns differ across disciplines?

d.      Can you predict how collaborations for one individual change over time?

2.      Protein-protein interaction networks and their subcellular localisation.
Proposed by Mikael Boden.
Overview: It is reasonable to believe that protein-protein interactions are partially constrained by their subcellular compartments. This project would create a network representing protein-protein interactions (e.g. using MIPS Mammalian Protein-Protein Interaction Database http://mips.gsf.de/proj/ppi/, DIP: Database of Interacting Proteins http://dip.doe-mbi.ucla.edu/ or HPID: Human Protein Interaction database http://www.hpid.org) and analyse it in terms of their known (or predicted subcellular location). Both processes (interactions and localisation) are dynamic so it is not known to which extent the aforementioned constraint is present in the static data.

3.      Complex Systems challenges in Health care
See Invited Lecture in Week 1 by Prof Michael Ward, QLD Health.

4.      Complex systems challenges in network security (cops and robbers in real life)
See Invited Lecture in Week 3 by Kathryn Kerr, Auscert.
linkages between web sites, Graham Ingram, Auscert.

5.      Cell-cell interactions (cadherin paint).
details tba, potential collaboration with IMB.

6.      Wound healing: cells as agents with signals and 2D membranes.
Proposed by Kevin Burrage, reference Rod Smallwood

7.      Plant modelling: mapping parameter spaces
with Jim Hanan and Peter Gresshoff, CILR

8.      Additional projects are possible in a variety of topic areas

a.       Visualising the structure of optimisation landscapes (big valley landscapes)

b.      Analysing the network structure of conceptual spaces

c.       Calculating the effective dimensionality of high dimensional data

d.      Developing metrics for tree edit operations

9.      Cognitive science: Systems that learn. Networks have been used as memory and other models in a variety of forms, including both neural (Hopfield 1982) and symbolic (semantic) forms. Choose a cognitive model from the literature and analyse its structural, dynamical and computational properties of a memory model.

10.  Complex systems biology and evolutionary computation. Systems that evolve

11.  Systems that grow (ontogeny, plant growth)

12.  Choose your own project: Choose a domain that interests you, and consider how you could map real world phenomena into complex systems models:

a.       What are the component parts?  (they will form the nodes in the network)

b.      Are they discrete stable objects, like cars, or are they shifting under the evolving system, like a child’s understanding of the concepts in a language? (Is the alphabet fixed, stable, growing, or changing?)

c.       What are the interactions? (links in the networks)

d.      What are their properties? Are they measurable, like links between web pages? How do they change?

e.       What is the environment? Is it static, predictable or complex in its own right?

 

Questions that you might ask include (ask the easy questions first)

a.       Structural: What are the nodes and links?

                                       i.      Plot the connectivity distributions as increasing numbers of patterns are stored

                                     ii.      Where do the phase transitions in connectivity lie?

b.      Dynamical: Determine the parameter ranges for stable, cyclic, chaotic and class 4 (cf Wolfram science paper)

a.       Where do the phase transitions in dynamical behaviour lie?

b.      Basins of attraction

c.       Lapunovs during the change (learning) process

d.      Computation: in which dynamic class does the functional behaviour lie?