Portrait of Associate Professor Yiqiu Dong, DTU Compute

Written by Marianne Vang Ryde.


Coming from mainland China working her way through Hong Kong, Austria and Germany, Yiqiu Dong has finally settled down in Denmark. She never thought she should live so far north, but the love of mathematics and a group of kindred spirits in the DTU based group for computational uncertainty quantification, CUQI, led her here.

Yiqiu Dong loves to travel – to meet people and experience their culture and way of living and to find out how this is related to the landscape and the physical surroundings.

“It is quite funny to see these patterns – for me it’s a kind of logic not unlike mathematics,” she smiles.

Mathematics is Yiqiu’s great passion. Her father is a well-known painter in the traditional Chinese style, and as a child she thought she was going to follow in his footsteps, but quickly that became impossible.

“I liked math more,” she explains.

Long journey on the wings of mathematics

So, she left her city to study mathematics at Peking University. Though her supervisor worked with purely theoretical mathematic problems he pointed out to her that Prof. Raymond Chan at the Chinese University of Hong Kong was working with an interesting problem concerning image processing. And he suggested she tried to do a PhD in that field.

This turned out to be a very good idea, and Yiqiu never went back to pure mathematics.

“I like the math in itself but I don’t want to only work with pen and paper. I would like to improve things. That’s why I started with this image processing,” she says.

After the PhD, she also thought of choosing to work in the industry rather than academia, partly because she didn’t trust her own ability to do research. She even got a job offer from a big company in Shanghai. But her supervisor believed in her and pushed her towards academia. And just before she accepted the industrial job, an opening showed up in Austria at the University of Graz.

“I give it a year, and then I return to China,” she said to herself. But Austria turned out to be nice. A beautiful country, an interesting project, and an extremely good group of scientists. Also, she met her husband there, so in the end she stayed both in Europe and in academia.

Real problems with mathematical solutions

In her PhD, Yiqiu was looking at noise and errors in images. But she wanted to do more.

“We can’t remove errors; they will always exist. For instance, when you take a photo the camera itself will have small errors in the mechanical parts, and maybe you are shaking it a little. The mathematical formulas are very complicated, so you need to simplify, and this also introduces errors. If you take all those uncertainties into account, your mathematical formula will be too complicated. Furthermore, when you want your computer to solve it you introduce even more errors. Computers are not perfect either.”

“So, what we want to do is to make our mathematical methods more robust to error. Even if you have an error here or there, your result will still be acceptable. Our goal is to make the solution reliable,” she tells.

This ambition of hers is exactly the reason why she fits so well into the CUQI (Computational Uncertainty Quantification for Inverse Problems) project at DTU which aims to characterize and study the sensitivity of a solution taking into account errors and inaccuracies in the data, models, algorithms, etc.

The CUQI scientists want to develop a mathematical and computational framework for applying uncertainty quantifications to inverse problems such as image deblurring and fault inspection. They want to create a computational platform also suited for non-experts in many areas, e.g. those who use X-ray imaging to inspect objects for defects, thus aiding to analyse whether a crack is actually present.

“When you use CT scanning, either medical or industrial, the measurement geometry is not always stable, for instance, when scanning under water it will be affected by the waves,” Yiqiu explains.

“We try to find out how to handle that mathematically. And we also try to expand the solutions so they can be used for image restoration where a wrong restoration model will lead to a degraded image.”

Modesty and engagement

When asked to point out a certain result which makes her especially proud, Yiqiu hesitates for a long time, then says:

“I don’t know – I am never satisfied. I always feel I can do better. I wouldn’t use the word proud – I am interested and really engaged. I like everything related to errors and after so many years of work in this field, I think I have a quite good picture of different kinds of error and how to model and handle them. How to make the solutions robust, stable, reliable, and quick. I like to understand what is happening inside.”

“But I am not only satisfied with good results. No matter if it’s a good result or a bad one, I want to understand why. Maybe that’s why I like academia more than industry, I never have to hand in a final solution to a specific problem. A boss in the industry might be very angry with me.”

After eight years in Denmark and 14 years in Europe, Yiqiu is feeling very much at home in Europe and of course has most of her scientific network here.

But it hasn’t been easy for her to adapt to Western culture. Language is one thing, especially learning Danish is difficult when you are afraid to make mistakes and therefore shy to start speaking as Yiqiu is. But it has also been difficult to learn the cultural codes.

“In China we don’t like to say how we feel. We give different other signs to show that we disagree. But here you need to say it loud, otherwise nobody understands. That was a very important lesson for me to learn,” she says.

Still, she is convinced that she will stay in Denmark or at least in Europe. Of course, China will always be her home country, but professionally she cannot imagine living anywhere else. Now, Denmark is definitely home for her and her family. 

“I like the group, I like being at DTU. And the work covers all my professional needs,” she concludes.