He obtained his doctorate in 2 years and published 11 papers in top conferences! Zhejiang University alumnus Li Minchen won the SIGGRAPH Best Doctoral Dissertation Award

Jin Lei from Aofei Temple
Quantum Bit Report | Public Account QbitAI

For four consecutive years, the SIGGRAPH Best Doctoral Dissertation Award has been won by Chinese scholars.

This year, Zhejiang University alumnus and UCLA mathematics postdoctoral fellow Li Minchen won the award.

However, what emerged was not only the glory, but also his amazing academic career.

Come, feel a set of data:

In just 2 years , he obtained his doctorate degree.

During his doctoral studies, he published 11 SIGGRAPH papers.

This experience can't help but cause netizens to exclaim:

Have you ever felt what dimensionality reduction attack means?

A very short doctoral study experience

In September 2018, he entered the University of Pennsylvania (UPenn) to pursue a doctorate and graduated in December 2020.

A little over two years is indeed a bit short.

Li Minchen himself also said this in a previous interview with "Li Ding Chat Room".

But speed does not mean lowering standards.

There are generally two requirements for graduating with a doctorate in North America: one hard and one soft .

More rigid requirements include taking a certain number of courses, doing a certain number of defenses, publishing a certain number of papers, etc.

For example, Penn requires three papers published in top journals , and they must be the first author .

The slightly softer requirements come from the instructor (you know) .

To put it bluntly, it depends on whether your supervisor thinks you have met the graduation standards.

And even if you meet the standards, whether he is willing to let you graduate depends on the situation.

So facing these hurdles, how did Li Minchen "unlock" them step by step?

In fact, this is inseparable from his early experience of studying for a master's degree at UBC (University of British Columbia) .

The courses that Li Minchen took during his two and a half years of master's studies are recognized in the Penn system.

Based on this, he saved the time of taking two courses and one thesis defense during his doctoral period.

But more importantly, Li Minchen has published two first-author research papers during his master's degree (UBC master's graduation requires one first-author paper) .

At that time, I already had some knowledge, ideas and experience in the field of computer graphics.

Such master's experience laid a good foundation for his doctoral studies.

When it comes to the "soft" aspect, Li Minchen believes that he is very lucky to have met a very good mentor - Professor Jiang Chenfanfu .

(Yes, he is the mentor of Hu Yuanming and others whom we all know .)

Under the leadership of Professor Jiang, Li Minchen's main research direction is the application of optimization problems and numerical methods in geometric processing and physical simulation.

After he saw that Li Minchen had reached the level required of a PhD, he took the initiative to approach him and said:

Why don’t you think about whether you can prepare for graduation?

"Very shocked" was Li Minchen's first reaction when he heard his mentor's words.

But this also indirectly reflects his outstanding academic achievements.

After all, if Bole wants to select a horse, the “horse” must also have the quality of “Di Lu”.

But compared to academic achievements, through his study experience at UBC and UPenn, a seed had already taken root in Li Minchen's heart - to become a professor .

The person who had the most profound influence on him was Professor Jiang Chenfanfu:

He not only made me see the importance of being a professor in terms of teaching and educating people, exploring the boundaries of human knowledge, and inheriting the spirit, but also made me see the freedom and infinite possibilities of a professor's career.

At the same time, I also learned a lot from him about scientific research and interpersonal relationships, which made me want to become an outstanding professor like him.

So, after graduating with a Ph.D., he made his decision resolutely - to join the UCLA Department of Mathematics.

Li Minchen's title in the department is Assistant Adjunct Professor, which can be translated as temporary assistant professor.

It is worth noting that this is different from what we usually call assistant professor .

The former is more like a postdoctoral fellow in scientific research, but also has teaching work; while the latter is the familiar "through train" to tenured positions.

It is understood that Li Minchen is currently teaching a summer course (multivariable calculus) at UCLA , and will also teach numerical calculation methods in the fall semester.

It seems that Li Minchen has realized his dream in scientific research.

But how did he get into this scientific research path?

Where did your interest in computer graphics come from?

This has to start with Li Minchen’s earlier experience.

The undergraduate experience was influential

Like most boys when he was a child, Li Minchen loved playing games.

Apart from normal studies, he had not thought about what he wanted to do or which major to choose before going to college.

But just after entering Zhejiang University, two opportunities strengthened Li Minchen's determination to choose computer graphics.

When he was doing exercises online in a programming class during his freshman year, he wrote a program to solve a problem.

Then we find that the output results can meet the requirements of this problem.

Great sense of accomplishment .

This is Li Minchen’s most intuitive feeling about programming.

Before he came into contact with programming, when he saw something like a calculator, he simply thought it was a precision instrument.

But he feels very happy when he understands the principles and can implement them by doing it himself.

"Programming is an interesting and meaningful thing," said Li Minchen.

This can be regarded as a "good start" for his scientific research path.

Another opportunity is the story of Li Minchen choosing his major at Zhejiang University.

Before choosing a major, students will be arranged to visit various laboratories.

Li Minchen visited about five laboratories at that time, but after "touring" the first four laboratories, he felt confused - "I don't quite understand what they do . "

But the last one we visited was the computer graphics laboratory, the CAD&CG laboratory of Professor Zhou Kun of Zhejiang University .

After seeing the results of their work, Li Minchen suddenly felt:

Huh? Doesn't this look like a series of screenshots of me playing games? Interesting.

This aroused Li Minchen's interest, so he began to read the laboratory introduction carefully.

Later he discovered that the knowledge and achievements of computer graphics could not only be used in games or movies, but also had a certain relevance to the industrial world.

For example, CAD drawing, parts modeling, and even civil engineering simulation.

So, when it comes to choosing a major, Li Minchen said:

It is promising and interesting, so it is the right choice.

These two opportunities can be regarded as Li Minchen's "guides" on the road of scientific research.

But as he continued to delve deeper into computer graphics, Li Minchen discovered that there are actually more sub-fields in this field.

"It is a very comprehensive and inclusive discipline that involves research in all directions," Li Minchen explained.

When he just graduated from college, he took the initiative to read the SIGGRAPH papers.

Li Minchen found that the contents of these papers are very different. For example, some are about modeling, some are about rendering, some are about simulation, and some are about image manipulation, etc.

It's dazzling, and there are also big differences in style.

Some articles will focus on more fundamental issues and delve into numerical calculations or numerical optimization.

Some articles will focus on larger but more practical issues. Although they will not delve into the details, the pipeline of the entire system is very complete.

It was precisely because of this in-depth understanding that Li Minchen's original intention towards computer graphics changed slightly.

It is no longer just a simple thought that it is cool as it was at the beginning, but I have developed a strong interest in those more "basic and fundamental" but difficult problems in computer graphics.

This "basic" does not mean simple, but it means that the problem is not complicated and can be clearly stated in one or two sentences, but it is difficult to solve.

Therefore, Li Minchen became more and more fond of using mathematical modeling or numerical optimization methods to solve them:

I think it’s a very fulfilling and pleasurable thing.

Of course, since it is more theoretical, there will definitely be some boring elements in it.

In this regard, Li Minchen explained:

Things that are interesting, things that are popular, and things that are influential may not be so lucky to overlap.

But the achievements made by Li Minchen also prove one thing: as long as you stick to what you love, you will eventually reap rewards.

Best Doctoral Dissertation Award

Perhaps it was because of this persistence that Li Minchen won the Outstanding Doctoral Dissertation Award at this year's SIGGRAPH .

A related research is Robust and Accurate Simulation of Elastodynamics and Contact .

SIGGRAPH's official comment on this is:

Important breakthroughs have been made in the problem of "robust friction contact simulation in nonlinear solid dynamics with non-intersecting surfaces".

In previous solid motion simulation and force analysis, it is very difficult to accurately model the solids themselves or the contacts between them, especially for solids with complex shapes, fine structures, and large deformations.

The reason for this problem is that the elasticity of the solid is nonlinear and non-convex in mathematical description, and the contact is not only nonlinear and non-convex, but also not smooth.

These are properties that are extremely difficult to handle when applying numerical methods for physical simulation.

Traditional methods either have a certain degree of stability and accuracy but are very time-consuming, or are fast but visibly unrealistic (because of large errors) , and are even prone to numerical explosion, leading to simulation failure.

The IPC (Incremental Potential Contact) method proposed by Li Minchen in this study solves this problem well.

Specifically, within the framework of optimization theory, through rigorous mathematical definitions of all collision scenarios, dimensionality reduction acceleration of the obstacle function method, controllable modeling of static-dynamic friction conversion, and clever discretization of the friction operator.

This provides a new approach to solving this problem, thereby achieving unprecedentedly accurate, stable, and non-penetrating simulation effects at a faster speed.

But why did Li Minchen do this research?

After all, the official said that this is a "long-standing problem that is difficult to break through."

This is inseparable from the first SIGGRAPH article published by Li Minchen:

My first SIGGRAPH article, FoldSketch, was a research project when I was a master’s student at UBC.

In order to ensure the feasibility of the design, we applied fabric simulation to obtain the actual wearing effect of the designed clothing. At that time, we used a very popular open source fabric simulation system that could obtain simulation effects in real time.

But once the clothes come into contact with more clothes, the algorithm becomes very slow and easily fails, and it is impossible to guarantee no penetration. This problem caused me a lot of trouble in my research at the time, and I still remember it clearly, and I am determined to solve this problem.

It just so happened that through continuous accumulation, learning, and research on optimization theories and methods, I discovered the compatibility of the interior point method with this problem and the challenges that existed at the same time, so I tried to explore it without hesitation.

Of course, this is also inseparable from the support and guidance of my mentor Professor Jiang Chenfanfu, as well as the challenges and encouragement of Adobe collaborator Dr. Danny Kaufman.

Moreover, Li Minchen is not only deeply involved in academia, but also has made achievements in industry.

For example, it has been working closely with CG expert Zhang Xinxin's Zesen Technology, and can be considered as half a founder.

It is understood that they also plan to jointly open a company in the United States to industrialize IPC while promoting talent exchanges at home and abroad, so that higher quality physical simulation systems can truly improve the production methods of multiple industries in a simple and easy-to-use form.

Finally, we also chatted with Li Minchen about some light topics - Does he have a girlfriend?

In response, Li Minchen said:

Of course you can have both academics and love, the important thing is to find the right person with all your heart.

(Manually clicked the like button!)

Reference Links:

[1] Li Ding Chat Room "Sharing about completing a PhD in Computer Science in two years":
https://www.zhihu.com/zvideo/1348264350231289856
https://www.lidingzeyu.com/minchen-on-phd-reflection/
[2] Best Doctoral Dissertation:
https://www.proquest.com/openview/178e7289d88473c8df74c493cc3b501c/1
[3] SIGGRAPH official website:
https://www.siggraph.org/acm-siggraph-2021-awards/