3 ms·

I don't get it... It's this just a variant on the shortest path problem. Even a bad solution is going to look similar to a real rail network. There are plenty

by zemaj 17y ago

I don't get it... It's this just a variant on the shortest path problem. Even a bad solution is going to look similar to a real rail network. There are plenty of good, easy algorithms to solve this, much faster than waiting a day for slime to grow. Confused as to how this got in Science & Wired. What am I missing (I didn't read the Science report, so maybe there's something there?)

jvdh 17y ago
It's actually a lot more like the travelling salesman problem, for which there are no algorithms (well, some heuristical algorithms).
- Sukotto 17y ago
  It's actually more of a traveling salaryman problem :-)
m_eiman 17y ago
Well, they did construct a computerized version of the method the slime uses to construct the network. Also they say that it's suitable for use in a dynamic network where traffic flows change and nodes are added and removed.
tel 17y ago
Because prior to seeing it happen, slime mold configuration wasn't expected to be equivalent to railroad optimization. And then it was. Saying things are obvious after they've happened is 100% useless. Demonstrating you have the mental models to say these things predictively is pretty exciting, though. If you can do that, go become a scientist. Use those models. Publish papers. You'll get into Science, too.