Dynamical Analysis and Modeling of Experimental Data from Nonlinear Circuit Elements ("DYNAMODE")

本项目的一个重要目标是通过应用电路和系统理论的方法，为项目描述中提出的一系列物理系统推导和分析新的非线性忆阻器模型。尽管同时也有大量关于忆阻器的文献，但仅在少数情况下提出并分析了具有适当定义忆阻器的方程的定量紧凑模型，即依赖于状态的欧姆定律以及定义状态方程。值得注意的是，忆阻器形式的应用极大地加速了非易失性记忆和各种形式的神经形态计算的研究。为了未来发展新的创新忆阻电路，取代部分传统的CMOS技术，有必要应用物理和化学方法，并进行精确和详细的测量，以获得电路理论器件模型，从而在标准设计软件中精确和稳定地模拟忆阻元件和电路。我们的研究将集中于推导具有适当调谐参数的新动力学模型，使模型能够被其他研究人员推广到其他系统。这项计划工作的重点是为惠普实验室和劳伦斯伯克利国家实验室创建的NbO2忆阻器建立一个新的非常全面的实验数据库，Stanley Williams博士已经向德累斯顿的工作组提供了这个数据库，在那里将与Namlab gGmbH合作进行额外的测量。与我们正在进行的工作相比，我们需要基于这些数据进行全面的调查，以详细的形式监测设备动态，目前还没有。我们还计划确定更广泛的memi -element的动态非线性模型，包括各种类型的电感器和电容器，为现有的电路元件提供新的功能，可以刺激在传感，信号分析和局部计算方面取得更大的进展，即在具有局部存储器的非冯·诺伊曼架构上进行计算。特别是，我们将应用衍生模型来探索非线性动态电路元件的各种应用，如信号转导和分析，放大，数据传输和局部计算。

One important aim of this project is to derive and analyze new nonlinear memristor models by applying methods of circuit and system theory for a range of physical systems as presented in the project description. Although there is meanwhile a considerable bulk of literature on memristors, quantitative compact models with equations that properly define memristors, i.e. a state-dependent Ohm’s Law along with the defining state equations, have been proposed and analyzed in a few cases only. Here it is important to note that the application of the memristor formalism has dramatically accelerated research in nonvolatile memories and various forms of neuromorphic computation. For the future development of new innovative memristor circuits replacing partly conventional CMOS technology, it is necessary, to apply methods from physics and chemistry and to perform accurate and detailed measurements in order to obtain circuit theoretic device models allowing an accurate and stable simulation of memristor elements and circuits in standard designer software. Our investigations will be focused on the derivation of new dynamical models with appropriate tuning parameters to enable the models to be generalized to other systems by other researchers. This planned work is focused on a new very comprehensive experimental data base for the NbO2 memristor created at HP Labs and at the Lawrence Berkeley National Laboratory, it is already provided by Dr. Stanley Williams to the working group in Dresden where additional measurements will be performed in cooperation with the Namlab gGmbH. As compared to our ongoing work, here we need comprehensive investigations based on these data monitoring the device dynamics in a detailed form not being available so far. We are also planning to determine dynamical nonlinear models of a wider sets of mem-elements including various types of inductors and capacitors, giving new capabilities to existing circuit elements that could spur even more progress in sensing, signal analysis and local computation, i.e. computing on non-von Neumann architectures with local memory. Especially, we will apply derived models to explore a variety of applications for nonlinear dynamical circuit elements, such as signal transduction and analysis, amplification, data transmission and local computation.