力学信息学简介

数据在力学的发展中始终属于最基础和最重要的角色。在古典力学时代，通过对海量数据的总结归纳，科学大师们得出了以牛顿运动三大定律为代表的自然世界运行的客观规律。在当今时代，快速发展的力学实验自动化技术和高通量技术，使力学数据呈爆炸式增长，如何基于迅猛增长的数据来快速发现、发展和革新力学理论，成为一个迫切需要解决的问题。力学工作者可以借助当下快速发展的人工智能算法，直接智能地优化实验和生产工艺，或者利用诸如符号回归、稀疏回归和流形学习等机器学习方法对数据进行挖掘处理，发现并给出数据所遵循的公式形式，将数据上升为知识。这一人工智能和力学相结合的交叉学科便是“力学信息学”。基于力学信息学方法，古老的力学学科也必将迎来新的春天。

A brief introduction of mechanoinformations

Data play an essential, important and fundamental role in the development of mechanics. The classical mechanics, represented by the three famous Newton’s laws, is developed by summarizing and analyzing vast data from the observations of natural behaviors, especially, from the observations of star movement. It is nowadays the information era and data grow explosively owing to the developments of high-throughput computation, high-throughput experiment and experimental automatization. Scientists in mechanics are facing great challenge: how to utilize the explosive growth of data to further develop mechanics, or how to gainmechanics knowledge from big data. In this article, we propose to develop “mechanoinformatics” by employing techniques, tools, and theories drawn from the emerging fields of data science, internet, computer science and engineering, digital technologies, machine learning, and artificial intelligence (AI) to the field of mechanics to accelerate the development of mechanics. In “mechanoinformatics”, we emphasize the construction of mechanics databases and the combination of “human being learning” and “machine learning”. As examples, we sketchily elucidate symbolic regression, sparse regression and manifold learning methods, through which equations, theory and knowledge are achieved from data. Developing “mechanoinformatics” will definitely bring further prosperities to mechanics.