Comparison of Complex Systems Approaches and Applications

From COSNet

back to Complex Dynamics of Urban Systems

Structural complexity models

Approach	Common Applications	Particular focus	Constraints and advantages	Software
Cellular Automata	Simple land use change, urban developments, fire spread,	Spatial interactions, diffusion processes, pattern formation	Grid/discrete based. Mostly used for spatial phenomena. Can be computationally expensive.	[1]
Network analysis	Social networks, electricity transmission networks, transport and industrial networks	Interactions between nodes	Can be used to analyse structural features of interactions. Useful for mapping social relationships. Often relatively data hungry. Often Mathematically demanding.	[2]
Systems dynamics	Water supply, carbon cycles, urban metabolism, resilience assessment	Dynamic systems with multiple interacting components involving feedback	Requires mathematical formulation. Limited capacity for handling uncertainty. Can model feedback in systems. Good for exploring effects of non-linearity. Can show importance of initial and boundary conditions as well as finding stable and unstable regions, and attractors.	[3]
Evolutionary algorithms	Numerical optimization, Learning, Artificial Life, Robotics	Learning processes, Evolutionary processes, Optimization	Allows configurations to emerge without a designer. Can mimic real evolutionary processes. Allows for optimization over huge, undefined, complex and variable fitness landscapes.	[4]

Social complexity approaches

Approach	Common Applications	Particular focus	Constraints and advantages
Game theory (incl. evolutionary)	Economics, Behavioural modelling, Political science, Biology	Decision making, behaviour and cooperation	Requires mathematical formulation. Usually assumes rational actors. Deals with issues of repeated interactions. Deals with issues of perfect and imperfect information. Provides formulation of social dilemmas.
Heuristic decision making models	Resource use, Psychology, Computer science	Human decision making	Does not rely on assumptions of perfect information or rational actors. Computationally inexpensive. Allows for contradictions. Can embed cultural and institutional aspects. Can feed into evolutionary algorithms. Allows for modelling adaptive behaviour.
Experimental economics	Economics, Markets, Auctions	Collecting information about decision making and preferences	Difficult to generalize findings. Requires considerable skill to set up experiments. Can uncover surprising and important insights.

Hybrid approaches

Approach	Common Applications	Particular focus	Constraints and advantages	Software
Agent based models	Markets, epidemics, social interactions, autonomous segregation of populations, traffic congestion modelling	Dynamic interactions between large systems of diverse agents. Integration framework for complexity models	Success of ABM relies on the success of its components. Not straightforward to validate. Requires programming effort. Can have powerful visualization features. Good for scenario analysis. Can be linked to role-playing games. Based on Ontology meaning that narratives can be used as input. Very useful when the diversity of agents, and agent outcomes, is important.	[5]
Companion modelling	Natural Resource Problems	Social dilemmas. Rapid assessments. Mixing qualitative and quantitative information. Post normal dialectic.	Supports social learning and collective action. Relies on assumptions made by participants. Acknowledges multiple possible realities (post-normal). Allows for eliciting knowledge about complex human interactions. Relies on social validation.
Bayesian Belief Networks (Graphical Models)	Bioinformatics, Decision support systems	Decision making under uncertainty and imperfect information	Interactions under uncertainty encapsulating beliefs. Excellent for risk assessments. Can incorporate stakeholder perceptions and evaluations. "doesn't really solve the problem of establishing causal relations". Computationally infeasible for large graphs.	[6]

back to Complex Dynamics of Urban Systems