scholarly journals Delayed Curse of Dimension for Gaussian Integration

2000 ◽  
Vol 16 (2) ◽  
pp. 474-506 ◽  
Author(s):  
Francisco Curbera
Keyword(s):  
Gaussian Integration ◽  
Curse Of Dimension

scholarly journals Efficient approximation of solutions of parametric linear transport equations by ReLU DNNs

2021 ◽  
Vol 47 (1) ◽  
Author(s):  
Fabian Laakmann ◽  
Philipp Petersen
Keyword(s):  
Neural Networks ◽  
Deep Neural Networks ◽  
Initial Conditions ◽  
Activation Function ◽  
Transport Equations ◽  
High Dimensional ◽  
Linear Transport ◽  
Approximation Rates ◽  
Curse Of Dimension ◽  
Efficient Approximation

AbstractWe demonstrate that deep neural networks with the ReLU activation function can efficiently approximate the solutions of various types of parametric linear transport equations. For non-smooth initial conditions, the solutions of these PDEs are high-dimensional and non-smooth. Therefore, approximation of these functions suffers from a curse of dimension. We demonstrate that through their inherent compositionality deep neural networks can resolve the characteristic flow underlying the transport equations and thereby allow approximation rates independent of the parameter dimension.

An inhomogeneous cell-based smoothed finite element method for the nonlinear transient response of functionally graded magneto-electro-elastic structures with damping factors

2018 ◽  
Vol 30 (3) ◽  
pp. 416-437 ◽  
Author(s):  
Liming Zhou ◽  
Ming Li ◽  
Bingkun Chen ◽  
Feng Li ◽  
Xiaolin Li
Keyword(s):  
Finite Element Method ◽  
Finite Element ◽  
Element Model ◽  
Micro Electro Mechanical System ◽  
Functionally Graded ◽  
Elastic Structures ◽  
Mapping Procedure ◽  
Gaussian Integration ◽  
Smoothed Finite Element Method ◽  
Element Method

In this article, an inhomogeneous cell-based smoothed finite element method (ICS-FEM) was proposed to overcome the over-stiffness of finite element method in calculating transient responses of functionally graded magneto-electro-elastic structures. The ICS-FEM equations were derived by introducing gradient smoothing technique into the standard finite element model; a close-to-exact system stiffness was also obtained. In addition, ICS-FEM could be carried out with user-defined sub-routines in the business software now available conveniently. In ICS-FEM, the parameters at Gaussian integration point were adopted directly in the creation of shape functions; the computation process is simplified, for the mapping procedure in standard finite element method is not required; this also gives permission to utilize poor quality elements and few mesh distortions during large deformation. Combining with the improved Newmark scheme, several numerical examples were used to prove the accuracy, convergence, and efficiency of ICS-FEM. Results showed that ICS-FEM could provide solutions with higher accuracy and reliability than finite element method in analyzing models with Rayleigh damping. Such method is also applied to complex structures such as typical micro-electro-mechanical system–based functionally graded magneto-electro-elastic energy harvester. Hence, ICS-FEM can be a powerful tool for transient problems of functionally graded magneto-electro-elastic models with damping which is of great value in designing intelligence structures.

A fast algorithm for the adaptive discretization of 3D parametric curves

2019 ◽  
Vol 37 (5) ◽  
pp. 1663-1682
Author(s):  
Jianming Zhang ◽  
Chuanming Ju ◽  
Baotao Chi
Keyword(s):  
Mesh Generation ◽  
Fast Algorithm ◽  
Design Methodology ◽  
Three Dimensional ◽  
Parametric Curves ◽  
Content Type ◽  
Gaussian Integration ◽  
Optimal Discretization ◽  
Adaptive Discretization ◽  
Conventional Algorithm

Purpose The purpose of this paper is to present a fast algorithm for the adaptive discretization of three-dimensional parametric curves. Design/methodology/approach The proposed algorithm computes the parametric increments of all segments to obtain the parametric coordinates of all discrete nodes. This process is recursively applied until the optimal discretization of curves is obtained. The parametric increment of a segment is inversely proportional to the number of sub-segments, which can be subdivided, and the sum of parametric increments of all segments is constant. Thus, a new expression for parametric increment of a segment can be obtained. In addition, the number of sub-segments, which a segment can be subdivided is calculated approximately, thus avoiding Gaussian integration. Findings The proposed method can use less CPU time to perform the optimal discretization of three-dimensional curves. The results of curves discretization can also meet requirements for mesh generation used in the preprocessing of numerical simulation. Originality/value Several numerical examples presented have verified the robustness and efficiency of the proposed algorithm. Compared with the conventional algorithm, the more complex the model, the more time the algorithm saves in the process of curve discretization.

Calculation of second virial coefficients for nitrogen and carbon monoxide

1980 ◽  
Vol 58 (6) ◽  
pp. 820-827 ◽  
Author(s):  
M. D. Whitmore ◽  
D. A. Goodings
Keyword(s):  
Carbon Monoxide ◽  
Good Accuracy ◽  
Power Series Expansion ◽  
Virial Coefficients ◽  
Coulomb Repulsion ◽  
Intermolecular Potential ◽  
Potential Models ◽  
Second Virial Coefficients ◽  
Gaussian Integration ◽  
Anisotropic Part

The classical second virial coefficients B(T) for nitrogen and carbon monoxide have been calculated exactly as a function of temperature for three different realistic models of the intermolecular potential. The potential models, due to Kohin, Raich and Mills, and Raich and Gillis, differ mainly, but not solely, in the way in which they represent the short-range Coulomb repulsion between molecules. As this interaction depends on the molecules' shapes, it is highly anisotropic. To ensure good accuracy in the results for B(T) the angular and radial integrals were performed by suitable Gaussian integration methods.The contributions to B(T) of various anisotropic terms are considered, and a power series expansion in terms of the anisotropic part of the potential discussed. The calculated results are compared with experiments, and some general conclusions drawn.

Sign in / Sign up

Export Citation Format

Share Document