论文标题
部分可观测时空混沌系统的无模型预测
A Schelling Extended Model in Networks -- Characterization of Ghettos in Washington D.C
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。
Segregation affects millions of urban dwellers. The main expression of this reality is the creation of ghettos which are city parts characterized by a combination of features: low income, poor cultural level... Segregation models have been usually defined over regular lattices. However, in recent years, the focus has shifted from these unrealistic frameworks to other environments defined via geographic information systems (GIS) or networks. Nevertheless, each one of them has its drawbacks: GIS demands high-resolution data, that are not always available, and networks tend to have limited real-world applications. Our work tries to fill the gap between them. First, we use some basic GIS information to define the network, and then, run an extended Schelling model on it. As a result, we obtain the location of ghettos. After that, we analyze which parts of the city are segregated, via spatial analysis and machine learning and compare our results. For the case study of Washington D.C., we obtain an $80\%$ accuracy.
