Skip to main content Skip to secondary navigation

The COVID-19 Health Alerts Website Provides the Stanford community with important updates related to the COVID-19 pandemic.

Visit Stanford COVID-19 Health Alerts Website

John Gunnar Carlsson

  

John Gunnar Carlsson, PhD '09

John Gunnar Carlsson is the Kellner Family Associate Professor of Industrial and Systems Engineering at the University of Southern California.

He received his PhD from the Institute for Computational and Mathematical Engineering (ICME) at Stanford University in 2009 under Yinyu Ye and his A.B. in mathematics and music from Harvard College in 2005. Carlsson works on algorithms for solving problems in continuous location theory, and more generally, optimization problems that have some kind of geographic element. His research is supported by DARPA, the Office of Naval Research, the Air Force Office of Scientific Research, the National Science Foundation, and the US Department of Transportation, and he was previously supported by an NSF GOALI grant, the Minnesota Department of Transportation, and the Boeing Company. A YouTube video of one of his paper and his website

What have you been up to since your ICME days?  / What are you working on now? After I graduated in 2009, I joined the Department of Industrial and Systems Engineering at the University of Minnesota; I stayed there until 2015, when I moved to USC.  I work on problems in transportation and logistics, like finding efficient ways to deliver packages or routing robots to pick up pallets in a warehouse.  At the moment I'm working on something that we call a "horsefly" problem, which is a vehicle routing problem where you have a truck and an aerial drone, and you're trying to find the best way to use the two vehicles in tandem to drop off some packages or gather information.  People often solve these kinds of problem using "discrete" or "network-based" optimization approaches, but a lot of my training in ICME was in computational geometry, so I like to think this brings something new to the table.

What is your fondest memory from your time as a student in ICME? There are so many; ICME is such a fun group of people to be with!  If I have to choose, it would be the time that I spent as one fourth of ICME's house band, "The Riot Squad," along with Paul Constantine, Mike Lesnick, and Arik Motskin.  The band got put together as a fun thing to do at the ICME Christmas party, but we got along so well that we stayed together for three more years until we graduated.  We ended up collaborating with a lot of musicians who are now famous, like Jidenna -- who I just learned was nominated for a Grammy -- and K.Flay.  There's even a stroller company, Orbit Baby, that uses one of our songs in one of their commercials.  I attached a tongue-in-cheek photo of us from 2007 (photo below).

Do you have any words of wisdom for current ICMErs?  One thing you wish you had studied or done while you were a student? On the really practical side, three pieces of software that I love and recommend are LyX, a graphical frontend for the LaTeX markup language, POV-Ray, which renders beautiful 3-dimensional pictures easily, and Inkscape, which I use for drawing all my technical diagrams.  I got my first research assistantship by making some nice looking drawings of networks for facility location problems.

How do you stay on top on the latest developments in your field? I read technology blogs like Slashdot, Ars Technica, and TechRadar religiously.  One of our latest research papers, which talks about the "Generalized Travelling Salesman Problem," was inspired by an article on TechCrunch about the rise in grocery delivery services.

Any favorite words-to-live by or a favorite quote to share? I'll go with Box and Draper's "All models are wrong, but some are useful."  It sounds cynical, but it's really helpful for putting things in perspective:  if I'm studying a real-world phenomenon, I could spend days and days trying to account for every single detail I could think of and build a really complex model to represent them, but there would still be things missing.  On the other hand, if I simplify the model to the point that I'm only capturing those elements that are absolutely critical to it, I'm more likely to end up with something that I can actually wrap my head around.