Ab Initio Simulations and Materials Chemistry in the Age of Big Data
Ab Initio Simulations and Materials Chemistry in the Age of Big Data
复制标题
大数据时代的从头算模拟和材料化学
DOI：
10.1021/acs.jcim.9b00781
复制
发表时间：
2020
影响因子：
5.6
通讯作者：
A. Fazzio
中科院分区：
文献类型：
--
作者：
G. R. Schleder;A. C. Padilha;A. Rocha;G. Dalpian;A. Fazzio
在这篇透视文章中，我们讨论了过去几十年来在算法和技术方面的计算进展，这些进展使分子和材料系统的模拟方法得以发展，广泛使用和成熟。这些进步导致了大量数据的产生，这需要创建几个计算数据库。在这种情况下，随着数据访问的民主化，该领域现在遇到了几个机会，以数据驱动的方法来解决化学和材料问题。特别是，用于预测新材料或性能的机器学习方法正越来越多地获得巨大成功。然而，黑盒使用在许多情况下失败;一些技术细节需要专业知识才能使预测有用，例如描述符和算法选择。这些方法代表了进一步发展的方向，特别是允许发达国家和新兴国家的进步与适度的计算基础设施。
In this perspective article we discuss computational advances in the last decades, both in algorithms as well as in technologies, that enabled the development, widespread use, and maturity of simulation methods for molecular and materials systems. Such advances led to the generation of large amounts of data, which required the creation of several computational databases. Within this scenario, with the democratization of data access, the field now encounters several opportunities for data-driven approaches towards chemical and materials problems. Especially, machine learning methods for predictions of novel materials or properties are being increasingly used with great success. However, black-box usage fails in many instances; several technical details require expert knowledge in order to the predictions to be useful, such as with descriptors and algorithm selection. These approaches represent a direction for further developments, notably allowing advances for both developed and emerging countries with modest computational infrastructures.
影响因子：
5.6
作者：
Legrain, Fleur;van Roekeghem, Ambroise;Mingo, Natalio
通讯作者：
Mingo, Natalio