Physarum polycephalum and Networks
As gas prices rise and traffic congestion get worse in the large cities around the world, more and more people turn to public transportation to get from place to place. It goes without saying that being able to optimize these public transit networks is in the best interest of everyone. There are a lot of computer algorithms on how best to optimize these networks. Then nearly two years ago a paper titled “Rules for Biologically Inspired Adaptive Network Design”, published in Science magazine, called for a new way to think about solutions to the problem.
Researchers from the Research Institute of Electronic Science at Hokkaido University wanted to test the efficiency of the networks created by a yellow slime mold Physarum polycephalum. They duplicated the map of the area around Tokyo and placed the mold on the position of Tokyo and food sources at the surrounding 36 towns. At first, the mold populated all the available space with a finely meshed network of thin tubes. But over time, the tubes that connected the food sources together inefficiently started to die out, leaving (and strengthening) the tubes that were connected efficiently behind. In time, only the most efficient paths between the food sources resulted. To the researchers’ surprise, the resulting network looked extremely similar to the actual Tokyo subway system.
The researchers eventually developed a Physarum polycephalum-inspired mathematical model that is adaptable to dynamic networks. Their model produced solutions to real-world situations that were comparable to or better than the actual real-world solution. An extension of their model allows for the adjustment of the cost-benefit ratio to create even more interesting results. Other researchers have applied this with networks in the United Kingdom and the Iberian Peninsula and have acquired similar results of efficiency.
It will be interesting to see what more will come about from this research as the demand for more efficient networks grows.
Source: http://www.sciencemag.org/content/327/5964/439.full
(http://www.sciencemag.org.proxy.library.cornell.edu/content/327/5964/439.full)