论文标题

部分可观测时空混沌系统的无模型预测

$py$GWBSE: A high throughput workflow package for GW-BSE calculations

论文作者

Biswas, Tathagata, Singh, Arunima K.

论文摘要

储层计算是预测湍流的有力工具,其简单的架构具有处理大型系统的计算效率。然而,其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间,然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用,该系统模拟了湍流的若干特征。我们表明,使用径向基函数作为非线性投影器,即使只有部分观测并且不知道控制方程,也能稳健地捕捉复杂的系统非线性。最后,我们表明,当测量稀疏、不完整且带有噪声,甚至控制方程变得不准确时,我们的网络仍然可以产生相当准确的预测,从而为实际湍流系统的无模型预测铺平了道路。

We develop an open-source python workflow package, $py$GWBSE to perform automated first-principles calculations within the GW-BSE (Bethe-Salpeter) framework. GW-BSE is a many body perturbation theory based approach to explore the quasiparticle (QP) and excitonic properties of materials. The GW approximation has proven to be effective in accurately predicting bandgaps of a wide range of materials by overcoming the bandgap underestimation issues of the more widely used density functional theory (DFT). The BSE formalism, in spite of being computationally expensive, produces absorption spectra directly comparable with experimental observations. The $py$GWBSE package achieves complete automation of the entire multi-step GW-BSE computation, including the convergence tests of several parameters that are crucial for the accuracy of these calculations. $py$GWBSE is integrated with $Wannier90$, a program for calculating maximally-localized wannier functions, allowing the generation of QP bandstructures. $py$GWBSE also enables automated creation of databases of metadata and data, including QP and excitonic properties, which can be extremely useful for future material discovery studies in the field of ultra-wide bandgap semiconductors, electronics, photovoltaics, and photocatalysis.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源