Computational Complexity of Air Travel Planning [pdf] (2003) | Hacker News Reader

Computational Complexity of Air Travel Planning [pdf] (2003)

demarcken.org

76 pointsrochoa10 comments

The PDF was produced by ITA, which famously used Common Lisp:

* https://en.wikipedia.org/wiki/ITA_Software

From 2001, a message from the same author as the linked paper:

> (Here's an email Carl de Marcken of ITA Software sent to a friend, describing their experiences using Lisp in one of the software industry's most demanding applications.)

* https://www.paulgraham.com/carl.html

This is well over 20 years old and is based on pre 9/11 flight data. I would suspect that a lot has changed since then. So proceed with no caution at all.

Since these sorts of things usually only get more and more complex over time, I would guess that it's all still true, but much more so.

This is a very popular article that get submitted every now and then (nearly every year) [1].

I think this kind of problem would be a very nice for logic, optimization and constraint programming that probably can be solved with modern tools like Google OR-Tool or Monash University MiniZinc [1],[2],[3].

[1] Past:

https://hn.algolia.com/?query=Computational%20Complexity%20o...

[2] Logic, Optimization, and Constraint Programming: A Fruitful Collaboration - John Hooker - CMU (2023) [video]:

https://www.youtube.com/live/TknN8fCQvRk

[3] Google OR-Tools:

https://developers.google.com/optimization

[4] MiniZinc:

https://www.minizinc.org/

A very interesting dive into, as the title says, the computational complexity of air travel planning. Graph algorithms with lots of complexity added due to the wide variety of fare conditions that airlines have dreamt up over the years.

The article may be from 2003 but I would call it an evergreen. While I imagine some of the details have changed since then, I suspect that the complexity has only grown since then.

It makes me wonder: Would an airline that drastically simplified its fares be more likely to appear in flight search results?

Simplifying the fares would make it less computationally expensive and, in theory, could take fewer steps to answer a flight planning query.

Imagine a flight search planner that, say, fanned out N airline-specific workers when handling a planning query and then displayed to the user whatever results it got back within some time limit. If FooAir had simple fares, the FooAir searcher would likely run faster than searchers for other airlines. Thus it would be more likely to return results for more queries, assuming the deadline is fairly tight because of usability metrics. (People don't tend to stick around waiting for slow results.)

At least a few years ago (~2014), the fare search was actually nearly instant, but all major airfare search sites added a delay because customers had the impression they were getting a better deal when they had to wait. It seems like the delay has been dialed back lately.

Are there any public, open, comprehensive datasets on flights?

> Are there any public, open, comprehensive datasets on flights?

Airlines and commercial aviation operators schedule their own flights. That is a dynamic schedulle. So, perhaps there is no "comprehensive data set".

However, FlightAware makes publicly available scheduled and completed flight data over many routes in the USA. You can search by route and get a list of flights.

Flight information includes filed departure time, route of flight, and speed. For completed flights actual time, altitude, and route is shown. For example, a search on the route Dallas/Fort Worth to Austin lists 45 flights.

I hope that helps.

(2003)

Reply on news.ycombinator.com