•  
  •  
 

Bulletin of Chinese Academy of Sciences (Chinese Version)

Keywords

scientific research driven by AI, scientific computing, Android paradigm

Document Type

Vigorously Promote Scientific Research Paradigm Transform

Abstract

The main purpose of scientific research is to discover fundamental principles and solve practical problems. Although tremendous progress has been made on both fronts, the lack of effective tools and efficient organizational structure still stands as the main bottleneck for scientific progress. The rapid development of artificial intelligence (AI) offers a new possibility. In recent years, deep learning has had an impressive performance, both in helping to solve fundamental scientific problems and in improving the effectiveness of scientific research tools. A new set of infrastructure is emerging, leading us to a new paradigm, the“Android paradigm”, for doing scientific research.

First page

10

Last Page

16

Language

Chinese

Publisher

Bulletin of Chinese Academy of Sciences

References

1 Dirac P A M. Quantum mechanics of many-electron systems. Proceedings of the Royal Society A, 1929, 123(792): 714-733.

2 E W N, Ma C, Wojtowytsch S, et al. Towards a mathematical understanding of neural network-based machine Learning: What we know and what we don’t. 2020, doi: arXiv:2009.10713v3.

3 Jumper J, Evans R, Pritzel A, et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021, 596: 583-589.

4 Zhang L F, Han J Q, Wang H, et al. End-to-end symmetry preserving inter-atomic potential energy model for finite and extended systems// Proceedings of the 32nd International Conference on Neural Information Processing Systems. New York: Curran And Associates, Inc., 2018: 4441-4451.

5 Zhang L F, Wang H, E W N. Reinforced dynamics for the enhanced sampling in large atomic and molecular systems. The Journal of Chemical Physics, 2018, 148: 124113.

6 E W N, Han J Q, Zhang L F. Machine learning-assisted modeling. Physics Today, 2021, 74(7): 36-41.

7 Guo Z Q, Lu D H, Yan Y J, et al. Extending the limit of molecular dynamics with ab initio accuracy to 10 billion atoms// Proceedings of the 27th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. Seoul: ACM, 2022: 205-218.

8 Zeng J Z, Zhang D, Lu D H, et al. DeePMD-kit v2: A software package for deep potential models. Journal of Chemical Physics, 2023, 159(5): 054801.

Download
Request a Copy

Share

COinS