Planning a subway system? Forget the computers, break out some slime mold!

[Ayatollah So]

Physics Today writes:

Designers of transportation networks have to weigh the cost of serving customers against the need for an efficient, robust system. Natural organisms, too, confront tasks in which they need to balance competing desiderata. As it forages for food, for example, a slime mold must balance cost (that is, the amount of protoplasm it extrudes), efficiency, and the ability to withstand injury. Remarkably, as recently reported by Atsushi Tero and colleagues from Japan and the UK, the molds do as well as transportation engineers in balancing their analogous competing needs. Panel a of the figure re-creates a 17-cm-wide map of the principal cities served by the Tokyo railway system with a slime mold (yellow) at the location of Tokyo and food flakes (white) representing other cities. In about a day’s time, the slime mold finds where the nourishment is and generates a protoplasm network with the food flakes as nodes. Standard metrics for analyzing transportation networks reveal that the mold’s foraging network and the Tokyo railway system perform similarly.

Hmm, which is cheaper, an engineering team, or a slime mold?

Actually, I lied. The researchers built a simple computer model of slime mold growth, and found that by tweaking some parameters, they could do better than the actual slime mold (or what human engineers have done to date).

So I guess you'll need that computer after all.

Still, pretty damn cool. It just goes to show you -
Evolutionary algorithms rock!

 

[Brighteye]

An acquaintance of mine has a paper coming out in Nature soon about the mathematics behind foraging shapes and behaviour. It's really interesting stuff.

 

[Ayatollah So]

I forgot the picture:

there ya go

 

[GoodGame]

So in a way a slime mold is like a neural net solving a problem?

I suspose that makes sense since it's really a collection of single-celled organisms. So is chemotaxis/intercellular signalling involved to help it solve the problem?

 

[ParadigmShifter]

My! That was a yummy slime mold!

EDIT: Nethack, simply great You fall into a pit! You die...

 

[Brighteye]

My acquaintance has worked out the mathematical principles governing how it is most efficient to search for new resources; the shapes that the mould forms when expanding to find the food flakes.

 

[Mise]

There was a thread on this before. Biological optimisation algorithms are really cool... It seems like a kind of "step up" from similar methods based on physical systems, such as simulated annealing, but I'm not sure whether they actually are... They all seem to depend on a feedback mechanism, which reinforces combinations that more closely fit some objective function.

What's even cooler than that is that evolution is itself an optimisation method. That is to say, one optimisation method (evolution) was employed to produce the "optimal optimisation method" for, separately, slime molds, bees, ants, etc, using different feedbacks, different objective functions, and even different evolutionary paths...

 

[Earthling]

My! That was a yummy slime mold!

This x100. For the longest time I didn't realize it was even a customizable name, and I decided not to try to edit the file to permanently give it a new name (say pamplemousse) Slime mold juice forever.

Actually already read about this from another science magazine/source I think, not the original journal article though. It is quite cool; the OP sums up my thoughts pretty well too.