论文标题

部分可观测时空混沌系统的无模型预测

Geometric Signatures of Switching Behavior in Mechanobiology

论文作者

Barkan, Casey O., Bruinsma, Robijn F.

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。

The proteins involved in cells' mechanobiological processes have evolved specialized and surprising responses to applied forces. Biochemical transformations that show catch-to-slip switching and force-induced pathway switching serve important functions in cell adhesion, mechano-sensing and signaling, and protein folding. We show that these switching behaviors are generated by singularities in the flow field that describes force-induced deformation of bound and transition states. These singularities allow for a complete characterization of switching mechanisms in 2-dimensional (2D) free energy landscapes, and provide a path toward elucidating novel forms of switching in higher dimensional models. Remarkably, the singularity that generates a catch-slip switch occurs in almost every 2D free energy landscape, implying that almost any bond admitting a 2D model will exhibit catch-slip behavior under appropriate force. We apply our analysis to models of P-selectin and antigen extraction to illustrate how these singularities provide an intuitive framework for explaining known behaviors and predicting new behaviors.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源