https://youtu.be/cZVxTeUeez8
Most days, math graduate student Veronica Ciocanel spends her time modeling how frog eggs go from jelly-like blobs to tiny tadpoles having a well-defined front and back, top and bottom. But for a week this summer, she used some of the same mathematical tools from her Ph.D. research at Brown to help a manufacturing company brainstorm better ways to filter nasty-smelling pollutants from industrial exhaust fumes.
Math professor Ryan Pellico of Trinity College took a similar leap. Most of his research aims to model suspension bridges that twist and bounce to the point of collapse. But he spent a week trying to help a defense and energy startup devise better ways to detect landmines using ground-penetrating radar.
Ciocanel and Pellico are among more than 85 people from across the U.S., Canada and the U.K. who met at Duke University June 13-17 for a five-day problem-solving workshop for mathematicians, scientists and engineers from industry and academia.
The concept got its start at Oxford University in 1968 and has convened 32 times. Now the Mathematical Problems in Industry workshop (MPI) takes place every summer at a different university around the U.S. This is the first time Duke has hosted the event.
The participants’ first task was to make sense of the problems presented by the companies and identify areas where math, modeling or computer simulation might help.
One healthcare services startup, for example, was developing a smartphone app to help asthma sufferers and their doctors monitor symptoms and decide when patients should come in for care. But the company needed additional modeling and machine learning expertise to perfect their product.
Another company wanted to improve the marketing software they use to schedule TV ads. Using a technique called integer programming, their goal was to ensure that advertisers reach their target audiences and stay within budget, while also maximizing revenue for the networks selling the ad time.
“Once we understood what the company really cared about, we had to translate that into a math problem,” said University of South Carolina graduate student Erik Palmer. “The first day was really about listening and letting the industry partner lead.”
Mathematicians Chris Breward of the University of Oxford and Sean Bohun of the University of Ontario Institute of Technology were among more than 80 people who met at Duke in June for a week-long problem solving workshop for scientists and engineers from industry and academia.
For the rest of the week, the participants broke up into teams and fanned out into classrooms scattered throughout the math and physics building, one classroom for each problem. There they worked for the next several days, armed with little more than caffeine and WiFi.
In one room, a dozen or so faculty and students sat in a circle of desks in deep concentration, intently poring over their laptops and programming in silence.
Another team paced amidst a jumble of power cords and coffee cups, peppering their industry partner with questions and furiously scribbling ideas on a whiteboard.
“Invariably we write down things that turn out later in the week to be completely wrong, because that’s the way mathematical modeling works,” said University of Oxford math professor Chris Breward, who has participated in the workshop for more than two decades. “During the rest of the week we refine the models, build on them, correct them.”
Working side by side for five days, often late into the night, was intense.
“It’s about learning to work with people in a group on math and coding, which are usually things you do by yourself,” Ciocanel said.
“By the end of the week you’re drained,” said math graduate student Ann Marie Weideman of the University of Maryland, Baltimore County.
For Weideman, one of the draws of the workshop was the fresh input of new ideas. “Everyone comes from different universities, so you get outside of your bubble,” she said.
“Here people have tons of different approaches to problems, even for things like dealing with missing data, that I never would have thought of,” Weideman added. “If I don’t know something I just turn to the person next to me and say, ‘hey, do you know how to do this?’ We’ve been able to work through problems that I never could have solved on my own in a week’s worth of time.”
Supported by funding from the National Science Foundation and the industry partners, the workshop attracts a wide range of people from math, statistics, biostatistics, data science, computer science and engineering.
More than 50 graduate students participated in this year’s event. For them, one of the most powerful parts of the workshop was discovering that the specialized training they received in graduate school could be applied to other areas, ranging from finance and forensics to computer animation and nanotechnology.
“It’s really cool to find out that you have some skills that are valuable to people who are not mathematicians,” Pellico said. “We have some results that will hopefully be of value to the company.”
On the last day of the workshop, someone from each group presented their results to their company partner and discussed possible future directions.
The participants rarely produce tidy solutions or solve all the problems in a week. But they often uncover new avenues that might be worth exploring, and point to new approaches to try and questions to ask.
“We got lots of new ideas,” said industry representative Marco Montes de Oca, whose company participated in the MPI workshop for the second time this year. “This allows us to look at our problems with new eyes.”
Next year’s MPI workshop will be held at the New Jersey Institute of Technology in Newark.
Post by Robin A. Smith