We formulate a data-driven method for constructing finite volume discretizations of a dynamical system's underlying Continuity / Fokker-Planck equation. A method is employed that allows for flexibility in partitioning state space, generalizes to function spaces, applies to arbitrarily long sequences of time-series data, is robust to noise, and quantifies uncertainty with respect to finite sample effects. After applying the method, one is left with Markov states (cell centers) and a random matrix approximation to the generator. When used in tandem, they emulate the statistics of the underlying system. We apply the method to the Lorenz equations (a three-dimensional ordinary differential equation) and a modified Held-Suarez atmospheric simulation (a Flux-Differencing Discontinuous Galerkin discretization of the compressible Euler equations with gravity and rotation on a thin spherical shell). We show that a coarse discretization captures many essential statistical properties of the system, such as steady state moments, time autocorrelations, and residency times for subsets of state space.
翻译:我们提出了一种数据驱动的方法,用于构建动力系统底层的连续性/福克-普朗克方程的有限体积离散化。我们采用一种方法,允许在状态空间中进行灵活的分区,泛化到函数空间,适用于任意长的时间序列数据,对噪声具有鲁棒性,并测量对有限样本效应的不确定性。在应用方法后,我们得到了马尔可夫状态(单元格中心)和生成器的随机矩阵逼近。当它们一起使用时,它们模拟底层系统的统计特性。我们将该方法应用于 Lorenz 方程(三维常微分方程)和已修改的 Held-Suarez 大气模拟(基于薄球壳的可压欧拉方程,带有重力和旋转的通量差分非连续 Galerkin 离散化)。我们展示了一个粗糙的离散化可以捕捉到系统的许多重要的统计属性,例如稳态矩、时间自相关性以及状态空间子集的居留时间。