Latent Space Physics: Towards Learning the Temporal Evolution of Fluid Flow

@article{Wiewel2019LatentSP,
  title={Latent Space Physics: Towards Learning the Temporal Evolution of Fluid Flow},
  author={Steffen Wiewel and M. Becher and N. Th{\"u}rey},
  journal={Computer Graphics Forum},
  year={2019},
  volume={38}
}
  • Steffen Wiewel, M. Becher, N. Thürey
  • Published 2019
  • Computer Science, Mathematics
  • Computer Graphics Forum
  • Our work explores methods for the data-driven inference of temporal evolutions of physical functions with deep learning techniques. [...] Key Method Key for arriving at a feasible algorithm is a technique for dimensionality reduction based on convolutional neural networks, as well as a special architecture for temporal prediction. We demonstrate that dense 3D+time functions of physics system can be predicted with neural networks, and we arrive at a neural-network based simulation algorithm with significant…Expand Abstract
    Neural Ordinary Differential Equations
    • 681
    • PDF
    Towards Physics-informed Deep Learning for Turbulent Flow Prediction
    • 12
    • PDF
    SPNets: Differentiable Fluid Dynamics for Deep Neural Networks
    • 41
    • PDF
    Learning to Simulate Complex Physics with Graph Networks
    • 28
    • PDF
    Lagrangian Fluid Simulation with Continuous Convolutions
    • 15
    • PDF

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 96 REFERENCES
    Data-driven fluid simulations using regression forests
    • 142
    • PDF
    Deep Learning the Physics of Transport Phenomena
    • 52
    • PDF
    PDE-Net: Learning PDEs from Data
    • 199
    • PDF
    Long Short-Term Memory
    • 30,975
    • PDF
    Stochastic Backpropagation and Approximate Inference in Deep Generative Models
    • 2,704
    • PDF
    Data-driven synthesis of smoke flows with CNN-based feature descriptors
    • 86
    • PDF
    ImageNet classification with deep convolutional neural networks
    • 52,359
    • PDF
    Learning representations by back-propagating errors
    • 16,006
    • PDF