Accuracy and Architecture Studies of Residual Neural Network solving Ordinary Differential Equations
This work provides an alternative neural network-based solver for ordinary differential equations, which could be beneficial for researchers and practitioners in scientific computing.
This paper explores using a Residual Neural Network (ResNet) to solve ordinary differential equations (ODEs). The ResNet, trained with stochastic gradient descent, is shown to achieve accuracy comparable to its finite difference method counterparts, such as forward Euler, Runge-Kutta2, and Runge-Kutta4.
In this paper we consider utilizing a residual neural network (ResNet) to solve ordinary differential equations. Stochastic gradient descent method is applied to obtain the optimal parameter set of weights and biases of the network. We apply forward Euler, Runge-Kutta2 and Runge-Kutta4 finite difference methods to generate three sets of targets training the ResNet and carry out the target study. The well trained ResNet behaves just as its counterpart of the corresponding one-step finite difference method. In particular, we carry out (1) the architecture study in terms of number of hidden layers and neurons per layer to find the optimal ResNet structure; (2) the target study to verify the ResNet solver behaves as accurate as its finite difference method counterpart; (3) solution trajectory simulation. Even the ResNet solver looks like and is implemented in a way similar to forward Euler scheme, its accuracy can be as high as any one step method. A sequence of numerical examples are presented to demonstrate the performance of the ResNet solver.