Won the champion of international EDA competition and the post-90s team of computer school of Huazhong University of science and technology -- Exploring cutting-edge algorithms to solve application problems
2021-11-23
Professor Lv Zhipeng (middle) and his team members. Not long ago, at the International Conference on computer aided design, Professor Lv Zhipeng's team from the school of computer science of Huazhong University of science and technology won the global champion of the electronic design automation layout and routing algorithm competition. This young team focuses on applied research, hoping to get through the "last mile" of algorithm research from the laboratory to the enterprise and write the scientific research achievements on the land of the motherland. Luo Canhui, a graduate student in the Institute of artificial intelligence and optimization, School of computer science, Huazhong University of science and technology, sits nervously in front of the computer and clicks the mouse and refreshes the web page every few minutes. Suddenly, the web page was updated. In the list of the first place in the International Conference on computer aided design (ICCAD) electronic design automation (EDA) layout and routing algorithm competition, the names of their team appeared impressively: Luo Canhui, Liang Jinghu, Xie Zhenxuan, Su Zhouxing and LV Zhipeng. "I was so excited that I shared the good news with the team for the first time," Luo Canhui said. The average age of Luo Canhui's team members is only 24 years old. Except for the instructor Professor Lv Zhipeng, they are all post-90s. This year is their first time to participate in the ICCAD competition. Such a young team can stand out from 137 teams in 12 countries and regions, but LV Zhipeng is very calm: "our goal is far more than this. Engaging in EDA research is not to issue papers or win awards, but to serve specific industrial applications and write scientific research achievements on the land of the motherland." Winning the championship is inseparable from deep accumulation EDA is a design method that uses computer-aided design software to complete the functional design, synthesis, verification and other processes of the chip. EDA layout and wiring design is an important part. Just like decorating a house, we need an optimal "construction drawing" to make all kinds of furniture, household appliances and wire network layout in the most appropriate place, so as to be beautiful, save space and perfect interconnection. Chips and other precision devices are only the size of fingernails, but they have to load 10 billion units. The interconnected circuits are more complex, and the optimal "construction drawings" can only be designed by algorithms. EDA is the cornerstone industry of electronic design. In the field of precision manufacturing, EDA is the basis of precision device production, processing and testing. "It can be said that if we master the best EDA, we will have the leading power in the field of high-end industry," said LV Zhipeng. The basis of EDA is algorithm. The winning of this young team is inseparable from the accumulation of more than 40 years in the algorithm field of the Institute. In the 1980s, the Institute of artificial intelligence and optimization was established in the school of computer science of Huazhong University of science and technology. Huang Wenqi, the first director of the Institute, became one of the earliest experts in China who paid attention to algorithm research. "In the early stage, the institute focused more on pure theoretical research and focused on solving NP hard problems," Lv Zhipeng said. NP hard problem is one of the seven mathematical problems in the world. In the field of algorithms, generally speaking, NP hard problems refer to those uncertain problems with high computational complexity and no "standard answer". How to solve problems without standard answers and standard paths? Huang Wenqi proposed the algorithm of personification and object imitation, which has been inherited to this day. "Imitating people and things is to solve mathematical algorithm problems with some design and layout wisdom of nature and human society," said LV Zhipeng. This algorithm is inspired by a crowded bus experience. At that time, Huang Wenqi received a research topic, which required machining a large circle, drilling some circular holes of different sizes, and the center of gravity of the large circle after drilling should be close to the original large circle. How? Huang Wenqi thought hard, but the seemingly full bus crowded into the scene of new passengers again and again, which gave him a flash of inspiration. He equivalent the crowded bus scene to a physical evolution model. After many algorithm exercises, he finally found the optimal solution to the big circle punching problem. Solve the practical application problems of enterprises There are many nurses in every hospital, but the level and specialty of nurses are different. Hospital nurses can not be cut off, but also ensure the rest time of nurses... How can each hospital quickly find the best scheduling scheme through algorithm? During his study abroad, LV Zhipeng participated in such an algorithm competition and won the third place. To his surprise, the results of the competition were immediately applied to major hospitals. "In foreign countries, algorithms are used in everything from nurse scheduling to express logistics and urban planning and design. Many enterprises will closely combine algorithms with industry. With China's rapid economic development and in-depth digital transformation, we should closely combine scientific research with industrial development," said LV Zhipeng. After becoming the director of the Institute in 2011, he devoted himself to the applied transformation of scientific research. The opportunity came unexpectedly. In the graduation season of 2011, an enterprise came to Huake for recruitment. When the interviewer talked with a graduate student applying for a job, he talked about a topic that the graduate student was participating in, which happened to be a difficult problem to be solved by the enterprise. As soon as the interview was over, the interviewer immediately asked the graduate student to take him to meet LV Zhipeng, the project leader, and put forward the intention of cooperative scientific research on behalf of the enterprise. "This is the first application-oriented project our team has done. Although it does not have much economic benefits, we think it is worth it, which indicates that our team has officially applied algorithm research to industry," Lv Zhipeng said. A year and a half later, they came up with an optimization scheme, which greatly reduced the production cost of the enterprise. After taking the first step, the applied research of the Institute is becoming more and more smooth. From laboratories to enterprises, their algorithms have been applied to express logistics, electronic devices, smart city construction and many other fields. Adhere to the "breakthrough from 0 to 1" In many international competitions in the field of algorithms, the "problem setter" is an important enterprise in the industry, and the problem is a difficult technical problem for the enterprise itself. Therefore, promoting practice through competition is a tradition of the Institute of artificial intelligence and optimization. The ICCAD that LV Zhipeng's team participated in this time is one of the top academic conferences with the longest history in the EDA field. The topic of this competition is a problem in the current cutting-edge chip design. There is a thick one when the topic is printed, with a large amount of data and many restrictive conditions. The first contact with such a difficult topic, several post-90s were a little scared at first. Dr. Su Zexing decomposed the tasks according to everyone's strengths and advantages. Everyone quickly buried themselves in their specific problems and explored different solutions. "The NP hard problem is like this. There is no definite answer. We can only continue to explore, fail and think of new ways," said graduate student Liang Jinghu. "Perseverance and never give up" is the deepest impression of Su Zhouxing on the team members. The realization of "breakthrough from 0 to 1" is the most exciting moment for the team members. After four months of unremitting efforts, after countless trials and errors, they finally found a correct problem-solving route. Although the algorithm is widely used, Luo Canhui and other team members have strengthened their determination to engage in EDA research in the future. They are well aware that when many key technologies cannot be bought or learned, China needs researchers to settle down and break the "neck sticking" problem. "The results of this competition once again prove that young people can carry the big beam." when it comes to the future, LV Zhipeng is full of confidence. "We also hope to share experience with other researchers, jointly solve the" last mile "problem of algorithm research from laboratory to enterprise, and cultivate more talents for the country." (outlook new era)
Edit：Yuanqi Tang Responsible editor：Xiao Yu
Source：
Special statement: if the pictures and texts reproduced or quoted on this site infringe your legitimate rights and interests, please contact this site, and this site will correct and delete them in time. For copyright issues and website cooperation, please contact through outlook new era email：lwxsd@liaowanghn.com
Recommended Reading Change it 
Home
