Accelerating high order discontinuous Galerkin solvers using neural networks: 1D Burgers' equation

Article Mechanics

Convolutional-network models to predict wall-bounded turbulence from wall quantities

Luca Guastoni et al.

Summary: Two convolutional neural network models were trained to predict velocity-fluctuation fields in turbulent open-channel flow, with one directly predicting fluctuations and the other reconstructing flow fields using a linear combination of orthonormal basis functions. Both models outperformed traditional methods in predicting nonlinear interactions in the flow.

JOURNAL OF FLUID MECHANICS (2021)

Add to Collection

Article Multidisciplinary Sciences

Machine learning-accelerated computational fluid dynamics

Dmitrii Kochkov et al.

Summary: Numerical simulation of fluids is crucial in modeling physical phenomena like weather and aerodynamics, but solving the Navier-Stokes equations at scale remains challenging due to computational costs. This study uses end-to-end deep learning to improve approximations in computational fluid dynamics for modeling two-dimensional turbulent flows, achieving computational speedups and stability while generalizing well to different scenarios. Leveraging machine learning and hardware accelerators, this approach showcases the potential to enhance simulations without sacrificing accuracy or generalization.

PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA (2021)

Add to Collection

Review Computer Science, Interdisciplinary Applications

A review on deep reinforcement learning for fluid mechanics

Paul Garnier et al.

Summary: Deep reinforcement learning has been widely adopted in solving decision-making problems in physics and engineering, particularly in the field of fluid dynamics. This review provides insights into existing DRL applications in fluid mechanics, highlighting their potential and advantages in addressing new challenges. Recent results and comparison with classical methods are presented to showcase the progress made in this field.

COMPUTERS & FLUIDS (2021)

Add to Collection

Article Mechanics

From coarse wall measurements to turbulent velocity fields through deep learning

A. Guemes et al.

Summary: This study evaluates the applicability of super-resolution generative adversarial networks (SRGANs) for reconstructing turbulent flow quantities, demonstrating that SRGAN can enhance the resolution of coarse wall measurements and provide better instant reconstructions. Despite challenges with lower resolution inputs, SRGAN yields very good reconstruction results, capturing large-scale flow structures effectively.

PHYSICS OF FLUIDS (2021)

Add to Collection

Article Physics, Multidisciplinary

Integration of neural networks with numerical solution of PDEs for closure models development

Arsen S. Iskhakov et al.

Summary: Modeling often requires closures to consider the multiscale/multiphysics nature of certain phenomena. Interest has grown in applying machine learning for their development, with a focus on integrating knowledgebase to enhance flexibility and predictive capability. This paper presents a PDE-integrated ML framework that uses convolutional operators to solve PDEs and integrate neural networks directly on observed field variables, demonstrating its viability by integrating NNs with heat conduction, Navier-Stokes, and RANS equations.

PHYSICS LETTERS A (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications