Ant Colonies

It's a beautiful morning here--bright blue skies and 45 degrees. There's a new bird at my feeder--I am pretty sure it is a house finch (longer and less purple), but it could possibly be a purple finch. They are brown with some dark streaks, sort of like a sparrow with a reddish purple head and a sturdy bill for cracking seeds. We have lots of them in New Orleans. They are devoted to their mates--one watches while the other eats--and they sing a lovely song, a lot.

Yesterday I got my econ mid-term back and I did fail, but so did many people in the class so he curved the grades and I ended up with a C. I made 90 on the latest homework. That tells me that I do understand the material, the test is just too much for me.

I also got back homework in Adaptive Optimization--98. I'm working on the Ant Colony System(ACS) computer program this morning. It is the latest optimization technique we're studying that is based on something biological. We watched a YouTube video that someone created showing how ants search for food. That got us in the mood to model the ants in constructing a trail. We're solving the Traveling Salesman Problem (TSP) with this algorithm. In the problem a salesman wants to minimize the distance he travels in visiting "n" cities. To solve the problem with ACS, I've generated "n" ants and they start randomly to build their route. When all of the ants have built their route, I calculate the distance. Each ant deposits pheromone on their route (trail) and I make sure the ant with the best route so far deposits extra pheromone to encourage the other ants to follow his trail. Each ant continues building trails and depositing pheromone until all of them are on the same shortest (I hope) trail. This algorithm is quite complicated and only works on problems that are combinatorial (not continuous).

I find it interesting that the TSP problem (think about routing for airlines, for trucking companies, for network traffic, for telephone lines) is not easily solvable since it is so common and important. People everywhere for years have studied this problem and continue trying to find better (less computer time) solutions. The link I gave you for the TSP is from Scientific American and our teacher said it was a good basic article about the algorithm that people unfamiliar with optimization programs could understand.

I think I'll look at ants in a new light from now on!