Deep Learning the Physics of Transport Phenomena

By Amir Barati Farimani, Joseph Gomes, Vijay S. Pande. 2017

117 citations recorded by the source corpus.

We have developed a new data-driven paradigm for the rapid inference, modeling and simulation of the physics of transport phenomena by deep learning. Using conditional generative adversarial networks (cGAN), we train models for the direct generation of solutions to steady state heat conduction and incompressible fluid flow purely on observation without knowledge of the underlying governing equations. Rather than using iterative numerical methods to approximate the solution of the constitutive equations, cGANs learn to directly generate the solutions to these phenomena, given arbitrary boundary conditions and domain, with high test accuracy (MAE$<$1\%) and state-of-the-art computational performance. The cGAN framework can be used to learn causal models directly from experimental observations where the underlying physical model is complex or unknown.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Statistical and Nonlinear Physics

Citations in the corpus

Integrating Physics-Based Modeling with Machine Learning: A Survey (2020, background): “For example, Farimani et al. [78] have shown that Conditional Generative Adversarial Networks (cGAN) can be trained to simulate heat conduction and fluid flow purely based on observations without using knowledge of the underlying governing equations.”
Deep Generative Design: Integration of Topology Optimization and Generative Models (2019, background): “[12] also propose the integration of cGAN to solve the steady-state problem for heat conduction and incompressible fluid flow.”
Generating Liquid Simulations with Deformation-aware Neural Networks (2017, background): “Recently, first works have also addressed replacing numerical solvers with trained models for more generic PDEs [13, 27].”
Data-driven synthesis of smoke flows with CNN-based feature descriptors (2017, unknown): “E.g., a conditional GAN was used to compute solutions for smaller, two-dimensional advection-diffusion problems [Farimani et al. 2017; Long et al. 2017].”
Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systems (2020, background): “[91] have shown that Conditional Generative Adversarial Networks (cGAN) can be trained to simulate heat conduction and fluid flow purely based on observations without using knowledge of the underlying governing equations.”
Deep learning for determining a near-optimal topological design without any iteration (2018, unknown): “Farimani et al .”