Closed mdlincoln closed 5 years ago
I'm drafting up this email... but I've encountered what looks like an early-stopping problem in the pathfinding package that evaded the tests I wrote but seems to break on the real Pittsburgh data 😱. Going to try to resolve it so I can send them one of our proper pathways to use as a solution...
It looks like the penalty function for dramatically increasing an edge's weight after crossing was doing it a bit too dramatically, and started returning Inf
weights.
Ruben Martins has assigned an undergraduate to work on this project, I'll be sending them some more documentation on our work so far along with #96. They hope to have some python code for pathfinding "in a couple of weeks". @scottbot and I will try to schedule some office hours time to meet with the student before they start working away, just to make sure they understand the data and the problem.