Generative modeling of turbulence

نویسندگان

چکیده

We present a mathematically well founded approach for the synthetic modeling of turbulent flows using generative adversarial networks (GAN). Based on analysis chaotic, deterministic systems in terms ergodicity, we outline mathematical proof that GAN can actually learn to sample state snapshots form invariant measure chaotic system. this analysis, study hierarchy starting with Lorenz attractor and then carry GAN. As training data, use fields velocity fluctuations obtained from large eddy simulations (LES). Two architectures are investigated detail: deep, convolutional (DCGAN) synthesise flow around cylinder. furthermore simulate low pressure turbine stator pix2pixHD architecture conditional DCGAN being conditioned position rotating wake front stator. The settings effects specific explained. thereby show efficient simulating turbulence technically challenging problems basis moderate amount data. inference times significantly fall short when compared classical numerical methods, particular LES, while still providing high resolution. analyse statistical properties synthesized LES fields, which agree excellently. also ability generalize over changes geometry by generating positions not included

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

2DV Nonlinear k-ε Turbulence Modeling of Stratified Flows

The commonly used linear k-ε turbulence model is shown to be incapable of accurate prediction of turbulent flows, where non-isotropy is dominant. Two examples of non-isotropic flows, which have a wide range of applications in marine waters, are saline water flow and the stratified flows due to temperature gradients. These relate to stratification and consequently, variation of density through...

متن کامل

Generative Modeling by PRISM

PRISM is a probabilistic extension of Prolog. It is a high level language for probabilistic modeling capable of learning statistical parameters from observed data. After reviewing it from various viewpoints, we examine some technical details related to logic programming, including semantics, search and program synthesis.

متن کامل

Generative mesh modeling

We propose a novel model representation method. Its main feature is that 3D shapes are represented in terms of functions instead of geometric primitives. Given a set of – typically only a few – specific parameters, evaluating such a function results in a model that is one instance of a general shape. The shape description language is a full programming language, but it has an extremely simple s...

متن کامل

Lifelong Generative Modeling

Lifelong learning is the problem of learning multiple consecutive tasks in a sequential manner where knowledge gained from previous tasks is retained and used for future learning. It is essential towards the development of intelligent machines that can adapt to their surroundings. In this work we focus on a lifelong learning approach to generative modeling where we continuously incorporate newl...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Physics of Fluids

سال: 2022

ISSN: ['1527-2435', '1089-7666', '1070-6631']

DOI: https://doi.org/10.1063/5.0082562