论文标题
部分可观测时空混沌系统的无模型预测
Dependence Logics in Temporal Settings
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。
Many forms of dependence manifest themselves over time, with behavior of variables in dynamical systems as a paradigmatic example. This paper studies temporal dependence in dynamical systems from a logical perspective, by enriching a minimal modal base logic of static functional dependencies. We first introduce a logic for dynamical systems featuring temporalized variables, provide a complete axiomatic proof calculus, and show that its satisfiability problem is decidable. Then, to capture explicit reasoning about dynamic transition functions, we enhance the framework with function symbols and term identity. Next we combine temporalized variables with a modality for next-time truth from standard temporal logic, where modal correspondence analysis reveals the principles needed for a complete and decidable logic of timed dynamical systems supporting reductions between the two ways of referring to time. Our final result is an axiomatization of a general decidable logic of dependencies in arbitrary dynamical systems. We conclude with a brief outlook on how the systems introduced here mesh with richer temporal logics of system behavior, and with dynamic topological logic.