Help is at hand for commuters frustrated that their city's transportation network never seems to go where they need it, or anyone whose street has suffered an electricity blackout. The solution comes not from the vast quantity of computer power tackling such issues, but from watching how ants deal with similar problems.
Dr Tanya Latty of the University of Sydney specializes in “swarm intelligence” – the way collectives of animals, or even single celled organisms, can arrive at decisions even though no individual has the brain power or knowledge that would appear to be needed.
In Interface Latty explores the way Australian meat ants Iridomyrmex purpureus connect widespread nests. “What is amazing about these ants is that they don't rely on engineering to plan their networks,” Latty said in a statement. “However it turns out that they are able to find a specific balance between cheapness, efficiency and robustness.”
Argentine ants have previously been found to build extremely efficient networks of paths. “They can solve the shortest path problem,” Latty told IFLScience. However, I purpureus carve paths between sites and appear to value robustness as well as efficiency, incorporating redundancy so the network survives when individual connections are cut.
While the Argentine ants lay down trails in a matter of a few hours, Latty says meat ants can take a century to build their network, adding “I would be at end of my career before I had results,” if she relied on observing the building process in real life.
Latty observed that new nests are connected to the closest existing nest, and to a tree in which the ants can feed, provided one was sufficiently handy. She shared this information with mathematicians from Uppsala University, “Once we have found what nature does, we have tried to apply the same simple rules to predict what would happen to man-made systems, electric grids for example, if they were built by these ants,” said Arianna Bottinelli, a PhD student at Uppsala, in the statement.
Latty told IFLScience that she and Bottinelli created a mathematical model of the ants' environment and entered plausible sets of rules for path building, seeking the set that most closely resembled the observed nests.
“It worked well if you have 20 nodes like meat ants,” Latty told IFLScience, “But humans need many more. We tried the algorithm we had produced with 300 nodes, but it became a shortest path solution,” not a robust one like the meat ants produce.
However, with some algorithm tweaking the authors made something that produces networks that balance efficiency and resilience. Latty said the result has not been compared with the best work generations of computer models have managed but, “It is something different. It is almost like we are asking the experts on how to make decisions when no one is in charge.”
Latty has previously explored the way collections of the slime mold Physarum polycephalum form a search strategy, showing that even brainless organisms can balance competing forces as they produce efficient networks between food supplies.
