Postdoc: Quantized First-Principles Quantum Dynamical Simulations of Condensed Matter Systems

采用一种计算策略，首次对凝聚态系统进行了完全量子化的第一原理量子动力学模拟，其中有效地包括了电子量子化和原子核量子化。该计算策略将质心分子动力学(CMD)方法和Car-Parrinello(CP)方法结合在一个可伸缩的算法中。量子动力学信息将通过在有效的、依赖于温度的从头算势上运行量子粒子的经典形式的CMD轨迹来获得。通过对MD方程进行时间积分的同时进行平衡Car-Parrinello量子路径积分平均来计算从头算有效势。这种方法将在并行CMD算法中实现，该算法应该能够实现近线性的性能扩展。模拟方法中的一个关键要素将是实施累积量平均策略，以显著减少计算工作量。该项目的成功将依赖于在高度并行的计算平台上实施的复杂可伸缩计算方法的开发。

A computational strategy will be used to carry out the first completely quantized first-principles quantum dynamical simulations of condensed matter systems, in which both the electronic and nuclear quantization are effectively included. The computational strategy combines the Centroid Molecular Dynamics (CMD) approach and the Car-Parrinello (CP) method within a scalable algorithm. The quantum dynamical information will be obtained by running classical-style CMD trajectories for the quantum particles on an effective, temperature-dependent ab initio potential. The ab initio effective potential will be calculated by performing equilibrium Car-Parrinello quantum path-integral averaging simultaneously with the time-integration of the MD equations. This methodology will be implemented in a parallel CMD algorithm, which should be capable of near linear performance scaling. A key element in the simulation methodology will be the implementation of a cumulant averaging strategy to significantly reduce the computational workload. The success of the project will rely on the development of intrisically scalable computational methods implemented on highly parallel computing platforms.