I was on the El to the airport the other day (for my Michigan visit, actually), and I was wondering how the train system could be improved. Someone once suggested that the network should be built like a spider web centered around down town Chicago. There would be trains which take you down town, but also other lines which circle the city. A line like that would certainly make the trip from Evanston to O'Hare a lot shorter, but it's a costly way to make the train system efficient.
Instead, as a nerd computer scientist, I thought about how trains are kind of like linked lists. You start at some node, but to get to another node you have to go through everything that's in between. This is, of course, as opposed to arrays, where you can just jump to the node you need.
And then I realized, people have solved this problem in computer science before. The solution: skip lists. These are not data structures normally taught in courses (although I heard that has changed since I took it... sigh), but the principle it functions on is simple. Going with the train analogy (and only going one way for the moment), instead of having one train stop at every station, you have several trains/tracks. One would stop at every station, another would stop at every other station, then every fourth, eighth, and so on. Getting to your destination would then involve transferring to trains that skip more and more stations, stay on that train until its next stop is past yours, then transfer to slower and slower trains.
Of course, this idea is not new to transportation companies either. Usually it's given the name "express". The only difference is that the express tends only to run during rush hours, and not otherwise. That is understandable - if there aren't that many passengers, profit will be low or even non-existent if there are several trains going around rather than just one.
And yet, the tracks are already there. Between Howard and Fullerton the Red and Purple Lines run on separate tracks. They don't stop at the same stations (except Belmont), but the tracks are there for this to happen.
So here's what I suggest. Instead of doing the whole skip-2, skip-4, skip-8 system, which would require a separate track for each train, just use two tracks. Have the express do a skip-4 or something otherwise in the middle. The express will then travel roughly 4 times faster than the normal train, greatly speeding up travel. At the end of each track, just do a merge to change sides, and the trains are ready to go back.
Unfortunately, in the current economic climate no one will do this. It's probably prohibitively expensive, and will never get enough passengers. And the space to get 4 tracks (2 there and 2 back) is also pricey. But it looks good on paper!
And that is how computer science (tries to) impact real life.
Subscribe to:
Post Comments
(
Atom
)
NYC's train system works a lot like that. There are express lines and local lines for the same routes. But it also gets complicated, because at the ends of the routes where there are less popular stops, the express and local trains split and all become local. In that way, they can serve more locations but still be efficient for the popular stops.
ReplyDelete