论文标题
部分可观测时空混沌系统的无模型预测
Point singularity array with metasurfaces
论文作者
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。
Phase singularities are loci of darkness surrounded by monochromatic light in a scalar field, with applications in optical trapping, super-resolution imaging, and structured light-matter interactions. Although 1D singular structures, such as optical vortices, are the most common due to their robust topological properties, uncommon 0D (point) and 2D (sheet) singular structures can be generated by wavefront-shaping devices such as metasurfaces. Here, using the design flexibility of metasurfaces, we deterministically position ten identical point singularities in a cylindrically symmetric field generated by a single illumination source. The phasefront is inverse-designed using phase gradient maximization with an automatically-differentiable propagator. This process produces tight longitudinal intensity confinement. The singularity array is experimentally realized with a 1 mm diameter TiO2 metasurface. One possible application is blue-detuned neutral atom trap arrays, for which this light field would enforce 3D confinement and a potential depth around 0.22 mK per watt of incident trapping laser power. Metasurface-enabled point singularity engineering may significantly simplify and miniaturize the optical architecture required to produce super-resolution microscopes and dark traps.