论文标题
部分可观测时空混沌系统的无模型预测
100 prisoners and a lightbulb -- looking back
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。
100 prisoners and a light bulb is a long standing mathematical puzzle. The problem was studied mostly in 2002 [5], 2003 [1], and 2004 [3]. Solutions in published articles had average number of visits above 3850, but best solutions on forums had (declared) average number of visits around 3500. I spent some time in 2007-2009 to optimize the communication strategy and I pushed the average number of visits below 3390, seems no new ideas appear after it. Recently I have met several people familiar with published papers from 2002-2003 but not knowing newer results. Even after 2009 several papers on the topic were published where the new results were not mentioned [4]. Whole book was written about the problem [2]. This is why I am writing this summary.
