Operator learning for predicting multiscale bubble growth dynamics

نویسندگان

چکیده

Simulating and predicting multiscale problems that couple multiple physics dynamics across many orders of spatiotemporal scales is a great challenge has not been investigated systematically by deep neural networks (DNNs). Herein, we develop framework based on operator regression, the so-called network (DeepONet), with long-term objective to simplify modeling avoiding fragile time-consuming “hand-shaking” interface algorithms for stitching together heterogeneous descriptions phenomena. To this end, as first step, investigate if DeepONet can learn different scale regimes, one at deterministic macroscale other stochastic microscale regime inherent thermal fluctuations. Specifically, test effectiveness accuracy in multirate bubble growth dynamics, which described Rayleigh–Plesset (R–P) equation modeled nucleation cavitation process dissipative particle (DPD). First, generate data using R–P caused randomly time-varying liquid pressures drawn from Gaussian random fields (GRFs). Our results show properly trained DeepONets accurately predict outperform long short-term memory networks. We also demonstrate extrapolate outside input distribution only very few new measurements. Subsequently, train DPD corresponding dynamics. Although are noisy collect sparse points trajectories, model able mean GRF pressures. Taken together, our findings be employed unify models problem, hence providing insight into role regression via DNNs tackling realistic simplifying descriptions.

برای دانلود رایگان 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

Online Machine Learning Techniques for Predicting Operator Performance

Electrical Engineering and Computer Science Institute of Software Engineering and Theoretical Computer Science

متن کامل

Dynamics of Macro–Nano Mechanical Systems; Fixed Interfacial Multiscale Method

The continuum based approaches don’t provide the correct physics in atomic scales. On the other hand, the molecular based approaches are limited by the length and simulated process time. As an attractive alternative, this paper proposes the Fixed Interfacial Multiscale Method (FIMM) for computationally and mathematically efficient modeling of solid structures. The approach is applicable to mult...

متن کامل

Operator Based Multiscale Method for Compressible Flow

Accurate simulation of subsurface flow with detailed geologic description is of great academic and industrial interest. Fully fine-scale simulation is usually too expensive. The multiscale method is developed to capture fine-scale information without solving fine-scale equations. It is more efficient than fine-scale simulation methods and more accurate than traditional upscaling techniques. Pre...

متن کامل

Bubble dynamics in DNA

The formation of local denaturation zones (bubbles) in double-stranded DNA is an important example for conformational changes of biological macromolecules. We study the dynamics of bubble formation in terms of a Fokker-Planck equation for the probability density to find a bubble of size n base pairs at time t, on the basis of the free energy in the Poland-Scheraga model. Characteristic bubble c...

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: Journal of Chemical Physics

سال: 2021

ISSN: ['1520-9032', '1089-7690', '0021-9606']

DOI: https://doi.org/10.1063/5.0041203