论文标题
部分可观测时空混沌系统的无模型预测
Superconducting bimodal ionic photo-memristor
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。
Memristive circuit elements constitute a cornerstone for novel electronic applications, such as neuromorphic computing, called to revolutionize information technologies. By definition, memristors are sensitive to the history of electrical stimuli, to which they respond by varying their electrical resistance across a continuum of nonvolatile states. Recently, much effort has been devoted to developing devices that present an analogous response to optical excitation. Here we realize a new class of device, a tunnelling photo-memristor, whose behaviour is bimodal: both electrical and optical stimuli can trigger the switching across resistance states in a way determined by the dual optical-electrical history. This unique behaviour is obtained in a device of ultimate simplicity: an interface between a high-temperature superconductor and a transparent semiconductor. The microscopic mechanism at play is a reversible nanoscale redox reaction between both materials, whose oxygen content determines the electron tunnelling rate across their interface. Oxygen exchange is controlled here via illumination by exploiting a competition between electrochemistry, photovoltaic effects and photo-assisted ion migration. In addition to their fundamental interest, the unveiled electro-optic memory effects have considerable technological potential. Especially in combination with high-temperature superconductivity which, beyond facilitating the high connectivity required in neuromorphic circuits, brings photo-memristive effects to the realm of superconducting electronics.