pETNNs: Partial Evolutionary Tensor Neural Networks for Solving Time-dependent Partial Differential Equations
Authors:
Tunan Kao,
** Zhao,
Lei Zhang
Abstract:
We present partial evolutionary tensor neural networks (pETNNs), a novel framework for solving time-dependent partial differential equations with both of high accuracy and remarkable extrapolation. Our proposed architecture leverages the inherent accuracy of tensor neural networks, while incorporating evolutionary parameters that enable remarkable extrapolation capabilities. By adopting innovative…
▽ More
We present partial evolutionary tensor neural networks (pETNNs), a novel framework for solving time-dependent partial differential equations with both of high accuracy and remarkable extrapolation. Our proposed architecture leverages the inherent accuracy of tensor neural networks, while incorporating evolutionary parameters that enable remarkable extrapolation capabilities. By adopting innovative parameter update strategies, the pETNNs achieve a significant reduction in computational cost while maintaining precision and robustness. Notably, the pETNNs enhance the accuracy of conventional evolutional deep neural networks and empowers computational abilities to address high-dimensional problems. Numerical experiments demonstrate the superior performance of the pETNNs in solving time-dependent complex equations, including the Navier-Stokes equations, high-dimensional heat equation, high-dimensional transport equation and Korteweg-de Vries type equation.
△ Less
Submitted 9 March, 2024;
originally announced March 2024.
Automatic differentiation of Sylvester, Lyapunov, and algebraic Riccati equations
Authors:
Ta-Chu Kao,
Guillaume Hennequin
Abstract:
Sylvester, Lyapunov, and algebraic Riccati equations are the bread and butter of control theorists. They are used to compute infinite-horizon Gramians, solve optimal control problems in continuous or discrete time, and design observers. While popular numerical computing frameworks (e.g., scipy) provide efficient solvers for these equations, these solvers are still largely missing from most automat…
▽ More
Sylvester, Lyapunov, and algebraic Riccati equations are the bread and butter of control theorists. They are used to compute infinite-horizon Gramians, solve optimal control problems in continuous or discrete time, and design observers. While popular numerical computing frameworks (e.g., scipy) provide efficient solvers for these equations, these solvers are still largely missing from most automatic differentiation libraries. Here, we derive the forward and reverse-mode derivatives of the solutions to all three types of equations, and showcase their application on an inverse control problem.
△ Less
Submitted 24 November, 2020; v1 submitted 23 November, 2020;
originally announced November 2020.
