论文标题
部分可观测时空混沌系统的无模型预测
Integral formulas for a foliated sub-Riemannian manifold
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。
In this article, we deduce a series of integral formulas for a foliated sub-Riemannian manifold, which is a new geometric concept denoting a Riemannian manifold equipped with a distribution ${\mathcal D}$ and a foliation ${\mathcal F}$, whose tangent bundle is a subbundle of ${\mathcal D}$. Our integral formulas generalize some results for foliated Riemannian manifolds and involve the shape operators of ${\mathcal F}$ with respect to normals in ${\mathcal D}$ and the curvature tensor of induced connection on ${\mathcal D}$. The formulas also include arbitrary functions $f_j\ (0\le j<\dim{\mathcal F})$ depending on scalar invariants of the shape operators, and for a special choice of $f_j$ reduce to integral formulas with the Newton transformations of the shape operators. We apply our formulas to foliated sub-Riemannian manifolds with restrictions on the curvature and extrinsic geometry of ${\mathcal F}$ and to codimension-one foliations.