We consider finite element discretizations of Maxwell's equations coupled with a non-local hydrodynamic Drude model that accurately accounts for electron motions in metallic nanostructures. Specifically, we focus on a posteriori error estimation and mesh adaptivity, which is of particular interest since the electromagnetic field usually exhibits strongly localized features near the interface between metals and their surrounding media. We propose a novel residual-based error estimator that is shown to be reliable and efficient. We also present a set of numerical examples where the estimator drives a mesh adaptive process. These examples highlight the quality of the proposed estimator, and the potential computational savings offered by mesh adaptivity.
翻译:我们考虑的是麦克斯韦尔方程式的有限元素分解,同时考虑的是非本地流体动力德鲁德模型,该模型准确计算了金属纳米结构的电子运动。具体地说,我们侧重于后置误差估计和网状适配性,这特别有意义,因为电磁场通常在金属及其周围介质的界面附近显示出强烈的局部性特征。我们建议了一个新的残余误差估计仪,该算法被证明是可靠和高效的。我们还展示了一组数字例子,让估计器驱动一个网状适应过程。这些例子凸显了拟议测算器的质量,以及网状适适配性所带来的潜在计算节余。