The Apsara Conference 2022, an annual technology event organized by Alibaba Cloud, was held from November 3rd to November 5th in Hangzhou. Gong Xingao, Academician of the Chinese Academy of Sciences (CAS), and Chancellor of Guangdong Technion – Israel Institute of Technology (GTIIT), delivered a speech titled “From Atoms to Computer, and Back”.

Today’s conference is themed on “computing drives future”, and I’d like to discuss the relationship between physics and computing. It’s widely accepted that the last century is about physics. It was a period when many outstanding achievements in physics were accomplished. Quantum Mechanics is one example. It provided the theoretical framework for understanding the basic law of the microworld and injected huge impetus into the social development of the last century. This field has produced numerous Nobel laureates, including this year’s winner in physics.

Professor Paul Dirac, who made fundamental contributions to the early development of quantum mechanics, has commented, “The fundamental laws necessary for the mathematical treatment of a large part of physics and the whole chemistry are thus completely known, and the difficulty lies only in the fact that application of these laws leads to equations that are too complex to be solved.” In the four decades that followed, a lot of distinguished physicists have dug into those difficult mathematical solutions. It was not until the mid-1960s that some breakthroughs occurred.

One particular theory was developed, which enabled people to understand the true essence of semiconductors and boost their development. That was the foundation of today's computing technology – the “Energy Band Theory”.

The first time humans used computers to solve physical problems was back in the 1950s when the esteemed physicist Professor Enrico Fermi visited the US mega computer and thought of using it to solve physical problems. Two young lads made it happen, and their research has created the major impetus for future physics development.

Starting from the early 1950s, a large number of computers have been used to advance materials science. After decades of time, computational physical science has emerged as a new research paradigm in physical science. Nowadays, computational physicists outnumber theoretical physicists and even researchers. The number of talents working in this brand-new physics subfield proves the success and importance of computing.

Why is computing so crucial to physics? In addition to being an indispensable tool for understanding experiment results, it also has the benefit of addressing three issues: 1) physics that is challenging to measure directly, such as topology; 2) experimental conditions that are hard to achieve, such as the physical behavior under extreme conditions; 3) complicated systems that are difficult to disentangle influencing factors.

Computing capabilities are essential to boost the development of computational physical sciences. In the last 60 years, the speed of computing power has multiplied a trillion or even a hundred trillion times. Over the last three decades, with the expansion of computational physics, a vast majority of core software are developed overseas. At that stage, China’s computing power was quite constrained. That’s why I’d like to call upon all friends and scholars in attendance to collaborate to develop this software on our own.

We have made a lot of expansion in this field, but challenges still abound.

Machine-learning has tremendous potential and application in computational physical sciences. We are leveraging computing power at different levels, ranging from the electronic level and atomic level to the structural level and material level. Take the arrangement of atoms as an example. How do they arrange, and in what kind of format? That’s a complicated mathematics problem. Despite our best efforts, we can only extract it at the scale of 30 to 40 atoms. With AI, we are able to anticipate the mechanism for more atoms.

In addition, how can we develop actual material in terms of computing? Scientists have demonstrated the capability of AI in accelerating development in this field.

Computational physical science can feed back into the development of computing hardware. We shouldn't neglect the quantum effect when developing electronic chips at the nanoscale, or even smaller. Scientists are attempting to optimize device performance, and I believe the so-called computational electronics can offer good support for the upcoming device generation.

To conclude, the combination of machine-illuminated learning and computational material science will provide new opportunities for reconstructing computational material science. Computers were born because of physics, and physics progressed because of computers. Thank you!

Text: The Beijing News, GTIIT News & Public Affairs

Photos: Apsara Conference 2022