We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Physics > Fluid Dynamics

Title: Dimension Reduced Turbulent Flow Data From Deep Vector Quantizers

Abstract: Analyzing large-scale data from simulations of turbulent flows is memory intensive, requiring significant resources. This major challenge highlights the need for data compression techniques. In this study, we apply a physics-informed Deep Learning technique based on vector quantization to generate a discrete, low-dimensional representation of data from simulations of three-dimensional turbulent flows. The deep learning framework is composed of convolutional layers and incorporates physical constraints on the flow, such as preserving incompressibility and global statistical characteristics of the velocity gradients. The accuracy of the model is assessed using statistical, comparison-based similarity and physics-based metrics. The training data set is produced from Direct Numerical Simulation of an incompressible, statistically stationary, isotropic turbulent flow.
The performance of this lossy data compression scheme is evaluated not only with unseen data from the stationary, isotropic turbulent flow, but also with data from decaying isotropic turbulence, a Taylor-Green vortex flow, and a turbulent channel flow. Defining the compression ratio (CR) as the ratio of original data size to the compressed one, the results show that our model based on vector quantization can offer CR$=85$ with a mean square error (MSE) of $O(10^{-3})$, and predictions that faithfully reproduce the statistics of the flow, except at the very smallest scales where there is some loss. Compared to the recent study of Glaws. et. al. (Physical Review Fluids, 5(11):114602, 2020), which was based on a conventional autoencoder (where compression is performed in a continuous space), our model improves the CR by more than $30$ percent...
Subjects: Fluid Dynamics (physics.flu-dyn)
Journal reference: Journal of Turbulence 2022
DOI: 10.1080/14685248.2022.2060508
Cite as: arXiv:2103.01074 [physics.flu-dyn]
  (or arXiv:2103.01074v2 [physics.flu-dyn] for this version)

Submission history

From: Andrew Bragg [view email]
[v1] Mon, 1 Mar 2021 15:31:37 GMT (9865kb,D)
[v2] Mon, 24 May 2021 12:39:52 GMT (9866kb,D)

Link back to: arXiv, form interface, contact.