Improving Long-term Autoregressive Spatiotemporal Predictions: A Proof of Concept with Fluid Dynamics

Abstract

Data-driven methods are emerging as efficient alternatives to traditional numerical forecasting, offering fast inference and lower computational cost. Yet, for complex systems, long-term accuracy often deteriorates due to error accumulation, and autoregressive training (though effective) demands large GPU memory and may sacrifice short-term performance. We propose the Stochastic PushForward (SPF) framework, which retains one-step-ahead training while enabling multi-step learning. SPF builds a supplementary dataset from model predictions and combines it with ground truth via a stochastic acquisition strategy, balancing short- and long-term performance while reducing overfitting. Multi-step predictions are precomputed between epochs, keeping memory usage stable without storing full unrolled sequences. Experiments on the Burgers' equation and the Shallow Water benchmark show that SPF achieves higher long-term accuracy than autoregressive methods while lowering memory requirements, making it promising for resource-limited and complex simulations.

References in corpus (10)

• An Experimental Review on Deep Learning Architectures for Time Series Forecasting
• Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations
• Reduced-order modeling of advection-dominated systems with recurrent neural networks and convolutional autoencoders
• Machine learning based spatio-temporal super resolution reconstruction of turbulent flows
• Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge
• Machine-learning-based reduced order modeling for unsteady flows around bluff bodies of various shapes
• Data-driven prediction of a multi-scale Lorenz 96 chaotic system using deep learning methods: Reservoir computing, ANN, and RNN-LSTM
• $\text{FastSVD-ML-ROM}$: A Reduced-Order Modeling Framework based on Machine Learning for Real-Time Applications
• Multi-fidelity physics constrained neural networks for dynamical systems
• Dynamical system prediction from sparse observations using deep neural networks with Voronoi tessellation and physics constraint