Abstract:SIAM Journal on Applied Dynamical Systems, Volume 23, Issue 1, Page 505-552, March 2024. Traditional sensitivity analysis methods fail for chaotic systems due to the unstable characteristics of the linearized equations. To overcome these issues two methods have been developed in the literature, one being the shadowing approach, which results in a minimization problem, and the other being numerical viscosity, where a damping term is added to the linearized equations to suppress the instability. The shadowing approach is computationally expensive but produces accurate sensitivities, while numerical viscosity can produce less accurate sensitivities but with significantly reduced computational cost. However, it is not fully clear how the solutions generated by these two approaches compare to each other. In this work we aim to bridge this gap by introducing a control term, found with optimal control theory techniques, to prevent the exponential growth of solution of the linearized equations. We will refer to this method as optimal control shadowing. We investigate the computational aspects and performance of this new method on the Lorenz and Kuramoto–Sivashinsky systems and compare its performance with simple numerical viscosity schemes. We show that the tangent solution generated by the proposed approach is similar to that generated by shadowing methods, suggesting that optimal control attempts to stabilize the unstable shadowing direction. Further, for the spatially extended system, we examine the energy budget of the tangent equation and show that the control term found via the solution of the optimal control problem acts only at length scales where production of tangent energy dominates dissipation, which is not necessarily the case for the numerical viscosity methods.

Backpropagation in hyperbolic chaos via adjoint shadowing

Backpropagation in hyperbolic chaos via adjoint shadowing

Backstepping Control of Mixed Hyperbolic-Parabolic PDE System with Multiple Coupling Terms

Backstepping Control of an Underactuated Hyperbolic-parabolic Coupled PDE System

Backstepping Synthesis for Feedback Control of First-Order Hyperbolic PDEs with Spatial-Temporal Actuation

On the Application of Optimal Control Techniques to the Shadowing Approach for Time Averaged Sensitivity Analysis of Chaotic Systems

Parameterized shadowing for nonautonomous dynamics

Probability density adjoint for sensitivity analysis of the Mean of Chaos

Shadowing, Hyers--Ulam stability and hyperbolicity for nonautonomous linear delay differential equations

Shadowing and hyperbolicity for linear delay difference equations

No-propagate algorithm for linear responses of random chaotic systems

On Shadowing and Chain Recurrence in Linear Dynamics

Shadowing the rotating annulus. Part II: Gradient descent in the perfect model scenario

Adjoint Sensitivities of Chaotic Flows without Adjoint Solvers: A Data-Driven Approach

Hyperbolicity, shadowing directions and sensitivity analysis of a turbulent three-dimensional flow

Kinetic Functions for Nonclassical Shocks, Entropy Stability, and Discrete Summation by Parts

Sensitivity computation of statistically stationary quantities in turbulent flows

DeepSPoC: A Deep Learning-Based PDE Solver Governed by Sequential Propagation of Chaos

The inverse problem of identifying complex hyperbolic equation source terms in electromagnetic propagation

Sharp Propagation of Chaos for the Ensemble Langevin Sampler

Least Squares Shadowing Sensitivity Analysis of a Modified Kuramoto-Sivashinsky Equation