arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1906.04015physics.comp-phcs.LGstat.ML
收藏

Cormorant:协变分子神经网络

Cormorant: Covariant Molecular Neural Networks

The University of Chicago(芝加哥大学);Flatiron Institute(熨斗研究所);Atomwise(Atomwise公司)

Brandon Anderson, Truong-Son Hy, Risi Kondor

摘要:

我们提出了Cormorant,一个旋转协变神经网络架构,用于学习复杂的多体物理系统的行为和属性。我们将这些网络应用于分子系统,有两个目标:学习原子势能面用于分子动力学模拟,并学习由密度泛函理论计算的分子基态性质。我们的网络的一些关键特征是:(a)每个神经元明确对应于原子的子集;(b)每个神经元的激活与旋转是协变的,确保整个网络是完全旋转不变的。此外,我们网络中的非线性是基于张量积和Clebsch-Gordan分解,允许网络完全在傅立叶空间中运行。Cormorant在从MD-17数据集中的构象几何学学习分子势能面方面显著优于竞争算法,并且在学习GDB-9数据集上的分子的几何、能量、电子和热力学性质方面与其他方法具有竞争力。

英文摘要:

We propose Cormorant, a rotationally covariant neural network architecture for learning the behavior and properties of complex many-body physical systems. We apply these networks to molecular systems with two goals: learning atomic potential energy surfaces for use in Molecular Dynamics simulations, and learning ground state properties of molecules calculated by Density Functional Theory. Some of the key features of our network are that (a) each neuron explicitly corresponds to a subset of atoms; (b) the activation of each neuron is covariant to rotations, ensuring that overall the network is fully rotationally invariant. Furthermore, the non-linearity in our network is based upon tensor products and the Clebsch-Gordan decomposition, allowing the network to operate entirely in Fourier space. Cormorant significantly outperforms competing algorithms in learning molecular Potential Energy Surfaces from conformational geometries in the MD-17 dataset, and is competitive with other methods at learning geometric, energetic, electronic, and thermodynamic properties of molecules on the GDB-9 dataset.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑