Massively parallel implementation of gradients within the Random Phase Approximation: Application to the polymorphs of benzene

The Random-Phase approximation (RPA) provides an appealing framework for semi-local density functional theory. In its current formulation, it is cost-effective and has a better scaling behaviour compared to other wavefunction based correlation methods. To broaden the application field for RPA, it is necessary to have first order properties available. RPA nuclear gradients allow for structure optimizations and data sampling for machine learning applications. We report on an efficient implementation of RPA nuclear gradients for massively parallel computers. We apply the implementation to two polymorphs of the benzene crystal obtaining very good cohesive and relative energies. Different correction and extrapolation schemes are investigated for further improvement of the results and in order to estimate error bars.

随机相位近似（RPA）为半局部密度泛函理论提供了一个颇具吸引力的框架。在其目前的表述中，RPA具有较高的成本效益，相较于其他基于波函数的相关方法，其标度行为更为优越。为了拓宽RPA的应用领域，有必要提供一阶性质。RPA的核梯度允许进行结构优化和数据采样，以适用于机器学习应用。本文报道了针对大规模并行计算机的高效RPA核梯度实现。我们将该实现应用于苯晶体的两种同质异形体，获得了非常优异的凝聚能和相对能。为了进一步优化结果并估计误差范围，本文探讨了不同的校正和外推方案。