Evolving a network

(updated  31st August 2007)

Keywords: simulating people, simulating crowds, simulating crowd dynamics

WelcomeIntoductionLadies LoosCar ParkingQueueing Facts

NetworksQueue Maths 1Queue Maths 2CongestionQueue LoS

During my PhD research I developed a wide range of network applications. The original work is shown below and detailed in my thesis Crowd Dynamics. This work has now taken a major step forward following some intensive developments over the two last years and based on many discussions and experiments during my time at "Starlab".  Myriad includes a range of network analysis tools and you can find numerous references to network analysis in a wide range of scientific text books.

The screen shot (above) shows the development of a simple connection network using 100 agents. It iterates between a routing (the four corners) and erosion algorithm. An optimal route (including optimal width) develops by using ONLY the interactions. This is a very simple model of the type of trails that develop, for example, on grass in a park. This experiment is documented in my PhD Thesis reference section.

The series of screen shots (101 - 112) show the trail as it develops over time. The interesting features are the symmetry breaking and route to optimal. The principles used have been extended to search for optimal routes in a network. These models serve to illustrate the process of using multiple agents to search for optimal solutions (note the path width and the intersection widths) which accommodate the agents requirements.

Figures 113 to 124 show the same experiment but with fewer agents. This develops the same optimal - but takes a different series of paths to get to the solution. The width of the paths also indicate the stability of the system as the experiment evolves with time. By using these agent based algorithms we have developed models which optimize traffic and routing networks.

Basically these models are multi-layered Cellular Automata. With them we have been modelling all manner of processes from efficient network design, optimal office layouts, efficient shopping areas, building layouts and traffic networks. Myriad now incorporates many more advanced algorithms for evolving spatial requirements for pedestrian planning and design.