Friday, November 18, 2011

TSPs and shit

think about this for a second.
you place a request into the stingerette system, and it determines where to insert that job such that you can pick it up en route, taking into account not only the source destination but also the sink. if you're going from IC to crecine, and you've got 4 other queued jobs/students, where they place a new job could differ than if you were to have entered 'IC to north ave', or 'Perry to crecine'. the variability increases exponentially with the number of queued jobs. but also with the fluctuation (e.g at one point you may have 4 jobs, at another you may have one).

additionally, consider vehicle constraints. the system must be programmed such that the van doesn't exceed capacity, and this is on a purely stochastic level when jobs are constantly entering and leaving. which is to say, REALLY FUCKING DIFFICULT.

i really wish this dude in front of me would wear shirts that covered his midriff. why does that even. lookofdisapproval.jpg

anyway. what you essentially have is a traveling salesman problem, except all the nodes are CHANGING ALL OVER THE PLACE LIKE MOTHERFUCKERS, and really, i never find my isye more enthralling when i consider the sheer complexity and usefulness of its applications.

can you even imagine the depth of that model? jesus christ.

a few phd people/professors are working on optimizing the trolley system, but (and yes, that is nigh impossible) it doesnt even seem as impressive as the stingerettes. well, i suppose it's difficult in its own way, with stochastic interarrival times x 3827363819147. idk, whatever. tbdiscussed at a later time.

No comments:

Post a Comment