Zobrazeno 1 - 4
of 4
pro vyhledávání: '"Chakraborty, Dibyajyoti"'
Predicting the long-term behavior of chaotic systems remains a formidable challenge due to their extreme sensitivity to initial conditions and the inherent limitations of traditional data-driven modeling approaches. This paper introduces a novel fram
Externí odkaz:
http://arxiv.org/abs/2410.05572
Forecasting high-dimensional dynamical systems is a fundamental challenge in various fields, such as geosciences and engineering. Neural Ordinary Differential Equations (NODEs), which combine the power of neural networks and numerical solvers, have e
Externí odkaz:
http://arxiv.org/abs/2407.00568
In this work, we provide a mathematical formulation for error propagation in flow trajectory prediction using data-driven turbulence closure modeling. Under the assumption that the predicted state of a large eddy simulation prediction must be close t
Externí odkaz:
http://arxiv.org/abs/2405.17612
Deep learning is increasingly becoming a promising pathway to improving the accuracy of sub-grid scale (SGS) turbulence closure models for large eddy simulations (LES). We leverage the concept of differentiable turbulence, whereby an end-to-end diffe
Externí odkaz:
http://arxiv.org/abs/2307.03683