From Academic Conferences to the Classroom

Portrait of Qian Qin
Photo by Samiraa Amin, Backpack student

“Teaching and research are actually very similar,” proclaims Assistant Professor Qian Qin of the School of Statistics. “At conferences, we get to introduce our own research, but in the classroom we are just introducing others’ research, some from hundreds of years ago.” Qin realized he enjoyed teaching after serving as the instructor for a couple courses at the University of Florida, where he received his PhD in statistics in 2019.

Although it was Qin's research that motivated him to seek a career in higher education, teaching has been an added benefit. "When you're in the classroom, you have to think about how to best convey your ideas to the students. You need to have a strategy." While many agree that this is a challenging undertaking, it is more than that to Qin. "Yes, it takes some time, but it's not a boring task. It can be fun. It can be exciting." 

The Intersection Between Physics and Statistics

As a physics major during his undergraduate career, Qin didn’t start college knowing he wanted to pursue a PhD in statistics. Qin spent a year working in a physics research lab at Peking University, after which he knew he wanted to continue his education but wasn’t convinced that physics was the right fit. At the time, statistics was an intriguing program for Qin because it was a growing field and the job prospects were good.

It also helped that he had prior exposure to the subject. “My parents are both mathematics professors, but they work mainly in statistics. I learned a lot from them, and they were obviously very supportive of [my] decision [to study statistics],” Qin laughs. Although Qin chose to study a sub-field of statistics that’s quite different from his parents’ area of research, they have served as great role models to him, especially early on in his education. 

Collaboration in Research

Qin joined the School of Statistics faculty in fall 2019 after completing his PhD in statistics at the University of Florida. When asked why he decided to come to the University of Minnesota, he explains, “The University of Minnesota has a great reputation. The statistics department in particular is pretty prestigious.” But that’s not why he took a teaching position here. During his interview, faculty members in the School of Statistics were straightforward and honest with Qin, traits he greatly admires.  He adds, “The atmosphere in the department is great. Everyone’s very friendly, and I enjoy a lot of academic freedom.” Qin feels supported by the department doing research in areas he likes, even if they are not “trendy.” 

Qin’s research in theoretical statistics is another reason he chose to come to Minnesota. While working toward his PhD, Qin’s main focus was Markov Chain Monte Carlo (MCMC), and after graduating he decided to continue this research. The School of Statistics is home to a few other MCMC researchers who Qin hopes to collaborate with, including Department Chair Galin Jones. Jones is actually Qin’s academic brother, meaning that they both studied at the University of Florida and shared the same PhD advisor, Jim Hobert. While MCMC is a broad field, it is unique for a medium-sized statistics department, like the one here at the U, to have several faculty members studying the subject, let alone academic siblings.

Markov Chain Monte Carlo is a tool that’s commonly used in Bayesian Statistics, which studies the distributions of unknown parameters. Starting from one point, MCMC simulates complex systems by generating random numbers based on the previous number. “The hope is that if you iterate the algorithm long enough, eventually the elements will approximately follow the complicated distribution that you are interested in,” Qin elaborates. “It’s a little like computer science theory.” 

Qin’s particular research focuses on the theory behind MCMC algorithms. As he explains, “The hope is we can analyze practical MCMC algorithms using mathematics and calculate the time one needs to run them [in order] for them to converge. Without this type of theoretical analysis, people can only run MCMC algorithms for as long as they can, and hope that these algorithms have converged by the time they stop.” Without the theoretical analysis, there is no way to know for sure that the algorithms have converged. 

What’s Next? 

As far as the future goes, “there are lots of interesting problems that are not solved.” A major problem in Qin’s research area, MCMC, is making the theoretical analysis practical and useful. Qin explains that while MCMC algorithms are very useful in practice, the theoretical analysis needs developing. “MCMC algorithms are very complicated. As of right now, analyzing the properties of these algorithms from a theoretical standpoint remains challenging,” Qin states. 

Personally, Qin looks forward to teaching more students at UMN.  "As professors, we like to share our ideas and opinions on subjects that we are familiar with," explains Qin. "This is definitely something I see myself continuing in the future. Maybe eventually I’ll create my own course,” he says, “and contribute to the school [in that way].”

This story was written by an undergraduate student in Backpack. Meet the team.

Share on: