News for hackers. Stuff that matters.
Re:Sussman anomaly by jpfr on Thursday January 18, 2018 @03:05PM (#16180819)
Convex optimization and its mixed-integer counterparts have developed amazing results. But not all planning problems fit this framework.

There is an entire community dealing with "AI planning". They try to find action-sequences leading to a desired outcome. Every action is defined by preconditions and effects. Most of the benchmark cannot even be sensibly formulated in terms of a mixed-integer optimization problem.

This seminal paper is a good entry-point to the field: http://www.jair.org/media/1705/live-1705-2731-jair.pdf

A bit of background research would do good before calling the problem at the center of a very active community "childish".

There are no comments on this comment yet.