From ‘New Guy’ to Successful Start-Up Founder, UM Alum Rick Chartrand Shows How It’s Done
For a man who spends his days analyzing satellite imagery, Rick Chartrand [BSc/93] is refreshingly down to earth. When asked to describe his decision to study mathematics in university, Chartrand gets straight to the point:
“When I was younger, I wanted to be a lawyer. I liked to argue. But that notion didn’t make it past high school, when I didn’t like my courses that weren’t science or math. At times it felt like I ended up doing math by a process of elimination. I started at the U of M intending to get a degree in theoretical physics. But I didn’t like the experimental component of physics courses. It was frustrating to understand the theory perfectly, but fail to demonstrate it experimentally, partly due to issues like friction and air resistance, but mostly due to not being a good experimentalist.”
So it was that the young Chartrand graduated from the UofM in 1993 with a degree in Mathematics, winning a University Gold Medal and a Governor General’s Silver medal in the process. His next stop was the University of California (Berkeley) where he earned a Master’s and then a PhD under the supervision of Dr. Donald Sarason. More awards followed, including the UC Berkeley Mathematics Department Research Fellowship and the Earl C. Anthony Fellowship. Chartrand worked as a teaching assistant and then as an instructor, winning an Outstanding Graduate Student Instructor Award and a Teaching Effectiveness Award for his implementation of an innovative homework workshop.
Two post-doc positions followed, one at Middlebury College in Vermont, another at the University of Illinois at Chicago. Both were temporary, and Chartrand remembers that period of being “the new guy” as particularly stressful.
“The worst part of being new is what it takes to get there. Not knowing what your next job is going to be is profoundly draining, and the effort put into trying to get that next job can take a lot out of your current job. I just visited Middlebury, Vermont, the site of my first faculty position 18 years ago, a one-year position but with the department doing a search for a permanent position during that year. It brought back a flood of memories of anxiety, being in what amounted to a months-long job interview, while also applying for faculty positions elsewhere for the likely case of not getting the permanent position.”
Chartrand and his wife decided to make a change, moving to the Four Corners region of the U.S. Having spent a great deal of time traveling through the area, they decided it was just as good a place as any to settle down. It was at this point that Chartrand changed careers, working at Los Alamos National Laboratory (LANL) as an applied mathematician.
Chartrand excelled at Los Alamos, working on such topics as numerical differentiation, algorithms for the Monge-Kantorovich problem of optimal transport and mathematical image processing. Underneath it all was his affinity for math. When asked to explain his love for the subject, Chartrand elaborates:
“Its logical precision. That fits with how I like to think. Some of that is training: much of the struggle of the first theoretical math courses is in learning how to think logically, breaking statements down to their logical components. But I’ve always been a sequential thinker. The same thought process has also been useful in writing code. I’ve also always been fascinated by numbers, but that turns out to have very little to do with mathematics.”
In addition to his love of math, Chartrand is a natural teacher. Asked to explain the area of his greatest research impact (the recently-emerged field of compressive sensing), Chartrand easily breaks down the topic, making it accessible for individuals far less familiar with the subject:
“The field isn’t well named. To me it’s the ability to reconstruct an image (or other type of signal) from what would seem to be too few measurements. The best examples come from medical imaging, like CT or MRI scanning. Imagine we want to get an image of a cross-section of a patient, and to see enough detail, we want that image to be a square that’s 1,000 pixels on a side. That’s 1,000,000 pixels in total, so 1,000,000 numbers we need to find to represent the gray levels of all the pixels. One would expect that we would need our scan to give us 1,000,000 measurements from which to reconstruct our image, and with traditional methods, that’s essentially true. With compressive sensing, however, we might need only 100,000 measurements, 1/10th as much data. That’s 1/10th as much X-ray radiation in the case of CT, or 1/10th the scanning time (and hospital expense) in the case of MRI. What lets us do that is that much of the pixel-level information in the image is redundant; if we’re clever, we might be able to describe the information content of the image using only 30,000 parameters. Then solving for those parameters from 100,000 measurements seems possible, though it takes advanced mathematics to do it. The name ‘compressive sensing’ in this context refers to the fact that if we can recover an image from 1/10th as much data, then the data can be thought of as a compression of the image. A compression that we measured directly with our scan; we’ve sensed compressively.”
One of the applications of compressive sensing allows video to be separated into moving and stationary components. How this happens mathematically is that the matrix that contains the frames of the video as columns is decomposed into the sum of a sparse matrix and a low-rank matrix. Chartrand’s work in this area became part of a $5M LANL-funded research project, for which he was one of the principal investigators. The project’s success allowed Chartrand to co-found Descartes Labs along with a group of other investigators.
Descartes gathers and processes satellite imagery in order to provide analysis services to global entities such as agricultural giant Cargill. It’s taken only three years for Descartes to expand from 5 employees to over 50 and to attract some $38M in funding. Chartrand is understandably proud of what his company has accomplished to date. His decision to leave Winnipeg in search of new horizons made him “the new guy”, but it also had its upside:
“The best part has been being in so many places, both places to live and workplaces. Since leaving Winnipeg, I’ve lived in all four U.S. time zones. I’ve worked at four universities, a nuclear weapons laboratory, and a tech startup. It’s been interesting to see how many different ways there are to do similar things, whether it’s running a research organization, or curbside recycling, or what have you. Being in one place, you make assumptions about the way things are that keep you from realizing that they can be done differently. I can’t imagine who I’d be if I hadn’t flown the coop, and students should never do graduate study at their undergraduate institution.
“It also struck me how little my career has resembled what I thought back then it was going to be. Leaving academia, starting over in a new field (applied rather than pure math), making a name for myself, and then starting over again in starting a tech startup. I didn’t think there would be so many starts!”
Dr. Chartrand is one of seven outstanding Alumni being honoured at the upcoming 2018 Faculty of Science Careers in Science – Pathways to Achievement Honoured Alumni Awards event.
What: 2018 Pathways to Achievement Honoured Alumni Awards
Topic: Careers in Science, Alumni and Student Mixer
When: Thursday, February 1, 2018, 3:00 p.m. (Doors open), 3:30 p.m. (Panel starts)
Where: Marshall McLuhan Hall, 2nd Floor, University Centre, University of Manitoba, Fort Garry campus
Reception to follow, all are welcome to attend.
Individual department events also planned. More details to follow.