Microsimulation is a discrete modelling technique that allows the study of populations based on the behaviour of each indivdual in the model rather than based on aggregated data. An individual's properties are sampled from data distributions for the property or from picking a probability. For example, in infection studies, transmission of disease from one person to another is a simple probability while the period of incubation is chosen from a lognormal or similar distribution.
Each individual behaviour can be altered by internal and external influences. This is not something that can be achieved with aggregated data on behaviours and is a more realistic aproach for studying threat in society.
We have developed a microsimulation framework, simulacron, and analysis tools for display:
Simulacron is our microsimulation framework; it abstracts away the common functionality used in our simulations to allow for more efficient construction of new models. Rather than a fixed set of capabilities, simulation functionality is implemented in modules. A number of related tools also fall under the umbrella of this project.
Associated toools include Refinery, Monotony and Spiderweb, which are infection analysis engines that allow matching to historical data and display of infection chains.
A number of modules have been constructed to increase the functionality of Simulacron:
Jazz is our visualisation package. Presently at the prototype stage, it will be a flexible, extensible framework for building data visualisation and processing tools.