海归学者发起的公益学术平台
分享信息,整合资源

交流学术,偶尔风月
密度泛函理论(DFT)是研究分子和材料电子结构的强大工具,它能够揭示许多物质性质的内在机制。然而,由于DFT在大型系统上计算时所需的高昂计算成本和运行时间,使得在此类系统中成功实施DFT计算仍然受到诸多限制。近年来深度学习的出现,使具有DFT精度的高效原子模拟成为可能,比较典型的例子是机器学习原子间势(MLIP)的广泛使用。机器学习原子间势模型在若干种构型构建的训练集上训练之后,可以对训练集之外的构型进行准确预测。如果能够开发一种具有与MLIP相似的高可转移性的模型,实现从不同构型的原子位置直接映射到电子哈密顿量,将极大地推动机器学习在实际电子结构计算问题中的应用。
Fig. 1 HamGNN architecture and the illustration of its subnetworks.
复旦大学的向红军教授等人设计了HamGNN图神经网络模型,该网络显式地考虑了哈密顿矩阵在三维实空间中的旋转等变性和宇称对称性,并在训练时以倒空间中随机采样的k点处的能带误差作为正则化项,使得该模型对训练集之外的分子和固体的电子结构具有很高的拟合能力和可迁移性。
Fig. 2 The prediction of HamGNN on several periodic solids that are not present in the training sets.
在碳同素异形体、硅同素异形体和SiO2异构体的哈密顿矩阵上分别进行训练后的HamGNN模型对训练集之外的同类结构预测的能带与DFT计算得到的能带高度一致。在硅同素异形体结构上训练后的HamGNN模型对包含4,284个原子的硅位错缺陷的能带和缺陷波函数进行了预测,揭示了硅位错引起的缺陷能级的高度局域性。在无转角的双层MoS2结构上训练之后的HamGNN模型准确预测出含1626个原子的Moiré双层MoS2中的Dirac锥色散和价带顶波函数的空间分布。HamGNN还在测试中准确拟合了不同化学计量比的BixSey族材料的含自旋轨道耦合效应(SOC)的电子哈密顿矩阵。
这些实际测试证明该研究提出的机器学习模型对电子哈密顿量的预测具有很高的精度和可迁移性,可以替代DFT用于高效计算大型系统的电子结构。该文近期发表于npj Computational Materials 9:182 (2023)英文标题与摘要如下,点击左下角“阅读原文”可以自由获取论文PDF。
Fig. 3 The electronic structure prediction on the BixSey quantum materials.
Transferable equivariant graph neural networks for the Hamiltonians of molecules and solids
Yang Zhong, Hongyu Yu, Mao Su, Xingao Gong & Hongjun Xiang
 This work presents an E(3) equivariant graph neural network called HamGNN, which can fit the electronic Hamiltonian matrix of molecules and solids by a complete data-driven method. Unlike invariant models that achieve equivariance approximately through data augmentation, HamGNN employs E(3) equivariant convolutions to construct the Hamiltonian matrix, ensuring strict adherence to all equivariant constraints inherent in the physical system. In contrast to previous models with limited transferability, HamGNN demonstrates exceptional accuracy on various datasets, including QM9 molecular datasets, carbon allotropes, silicon allotropes, SiO2 isomers, and BixSey compounds. The trained HamGNN models exhibit accurate predictions of electronic structures for large crystals beyond the training set, including the Moiré twisted bilayer MoS2 and silicon supercells with dislocation defects, showcasing remarkable transferability and generalization capabilities. The HamGNN model, trained on small systems, can serve as an efficient alternative to density functional theory (DFT) for accurately computing the electronic structures of large systems.
本文系网易新闻·网易号“各有态度”特色内容
媒体转载联系授权请看下方
继续阅读
阅读原文