The Associated Discrete Laplacian in ℝ 3 \mathbb{R}^{3} and Mean Curvature with Higher–order Approximations.
Wei-Hung Liao https://orcid.org/0000-0002-9915-1030 · 2025
Read The Associated Discrete Laplacian in ℝ 3 \mathbb{R}^{3} and Mean Curvature with Higher–order Approximations. by Wei-Hung Liao https://orcid.org/0000-0002-9915-1030 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-11871.txt.
Introduction
Wei-Hung Liao proves in 2025 results about discrete Laplacians on tetrahedral meshes in three-dimensional space. The primal and dual constructions give quite different operators. He shows the dual version's associated Laplacian satisfies the Euler-Lagrange equation of the Dirichlet energy, splitting into face Laplacians plus a discrete mean curvature term. The dual construction is optimal compared with the primal one. Its mean curvature reacts more sharply to sudden angle changes, and a higher-order correction for those changes cuts the error without refining the mesh.
Catalog record
| CREATOR | Wei-Hung Liao https://orcid.org/0000-0002-9915-1030 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-11871
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Associated Discrete Laplacian in ℝ 3 \mathbb{R}^{3} and Mean Curvature with Higher–order Approximations. Wei-Hung Liao https://orcid.org/0000-0002-9915-1030 Abstract In ℝ 3 \mathbb{R}^{3} , the primal and dual constructions yield completely different discrete Laplacians for tetrahedral meshes.In this article, we prove that the discrete Laplacian satisfies the Euler-Lagrange equation of the Dirichlet energy in terms of the associated discrete Laplacian corresponding to the dual construction. Specifically, for a three–simplex immersed in ℝ 3 \mathbb{R}^{3} , the associated discrete Laplacian on the tetrahedron can be expressed as the discrete Laplacian of the faces of the tetrahedron and the associated discrete mean curvature term given by the ambient space ℝ 3 \mathbb{R}^{3} . Based on geometric foundations, we provide a mathematical proof showing that the dual construction gives a optimal Laplacian in ℝ 3 \mathbb{R}^{3} compared to the primal construction. Moreover, we show that the associated discrete mean curvature is more sensitive to the initial mesh than other state-of-the-art discrete mean curvatures when the angle changes instantaneously. Instead of improving the angular transient accuracy through mesh subdivision, we can improve the accuracy by providing a higher–order approximation of the instantaneous change in angle to reduce the solution error. MSC Classification] 53C43 (Primary); 52B70, 53A05, 47A60 keywords discrete differential geometry, discrete dual Laplacian, discrete mean curvature, higher–order approximation pacs [ 1 Introduction With the rapid development of imaging techniques and the proliferation of sensor-acquired 3D images, techniques for processing and analysing image data have become very important. In recent years, the discretization of differential operators has been an exciting area of numerical analysis, geometric processing and computational engineering. The energy function space of a manifold is determined by the geometrical characteristics of the manifold. By studying its Euler-Lagrange equations, it can reflect much of the global geometric behaviour of the manifold. One of the energy functions commonly considered is the Dirichlet energy, so the discrete Laplacian becomes very important in numerical analysis, computer graphics and geometry processing. According to the discretization of Laplacian, discrete operators will inherit different properties. For triangular surface meshes, Wardetzky et al. Wardetzky et al. (2007) classified several important properties and analysed the connection between these properties for different discretization and continuous operators. The cotangent discretization is widely used in geometric processing, and it satisfies all of the above desirable properties if the mesh has Delaunay properties, and for any fixed set of function values it also minimizes the Dirichlet energy of the piecewise linear interpolation function, lacking the maximum principle only if the mesh is not Delaunay. There are several ways to derive the cotangent Laplacian on simplicial meshes. For triangle meshes, the primal and dual constructions of the Laplacian lead to the same discrete cotangent–weight Laplacian, as shown in Pinkall and Polthier (1993) . The familiar formula for the primal construction of the Laplacian in dimensions equal to 3 first appeared in Desbrun et al. (1999) , but they omitted the important length term. Later, Gu et al. Gu et al. (2003) gave a definition of the primal discrete Laplacian on a tetrahedral mesh for the study of heat flow problems to minimise harmonic energy, and then Chao et al. Chao et al. (2010) considered the same discrete Laplacian in the context of elasticity and function interpolation. For higher dimensions, the primal discrete Laplacian is considered in Crane (2019) . In 2020, Alexa et al. Alexa et al. (2020) provided explicit formula for the lesser-known dual construction of the Laplacian on tetrahedral meshes. The primal and dual approaches produce the same discrete Laplacian in triangular meshes. However, they have completely different representations and properties in tetrahedral meshes. Definition 1.1 . Suppose \mathcal{N}^{3} is a tetrahedral mesh embedded in \mathbb{R}^{3} . A discrete Laplacian defined on \mathcal{N}^{3} is said to be optimal if the sum of the discrete Laplacian and discrete mean curvature vector defined on the two-dimensional manifold \partial\mathcal{N}^{3} converges to the discrete Laplacian on \mathcal{N}^{3} in a weak* sense. 1 1 1 Let \{f_{n}\} be a sequence of bounded linear functionals on a norm space X and there exists an f\in X^{*} such that f_{n}(x)\longrightarrow f(x) for all x\in X ; namely, f_{n}\stackrel{{\scriptstyle\ast}}{{\rightharpoonup}}f . The main contributions of this paper are in the following properties: • Based on the geometric fact of hypersurfaces in \mathbb{R}^{3} , the discrete Laplacian on a triangular mesh can define the discrete mean curvature at vertices or on edges (Section 4.1 ). With respect to the discrete Laplacian on a tetrahedral mesh, as an approach for the continuous Laplacian in \mathbb{R}^{3} , it must contain a term representing the mean curvature of the hypersurface. On the basis of the dual construction of discrete Laplacian, we define its associated discrete Laplacian and reciprocal discrete mean curvature and endow it with some notion of the geometry of surface curvature(Section 5 ). • We give a mathematical proof that the associated discrete Laplacian satisfies the Euler-Lagrange equation for the Dirichlet energy. That is, for a three–simplex immersed in \mathbb{R}^{3} , the associated discrete Laplacian on the tetrahedron can be expressed as the discrete Laplacian of the faces of the tetrahedron and the discrete mean curvature term given by the ambient space (Section 6 ). • For geometric PDEs, reducing the mesh size allows the mesh to better reflect the surface structure, thus minimizing the discretization error. Ending the computation early before solving the PDE accurately can lead to solution errors. In addition to improving the mesh quality, it is important to ensure that the approximate order of the geometries is not less than the approximate order of the mesh. In Section 7 , for the mean curvature equation, we provide higher–order approximations of the mean curvature |\vec{H}| to minimize the solution error. 2 Preliminary and overview Let \left(\mathcal{M},\gamma\right) and \left(\mathcal{N},g\right) be two connected Riemannian manifolds of dimension n and n+d with metric tensors \left(\gamma_{\alpha\beta}\right) and \left(g_{ij}\right) in the local charts x=\left(x_{1},\dots,x_{n}\right) and f=\left(f_{1},\dots,f_{n+d}\right) on \mathcal{M} and \mathcal{N} , respectively. 2.1 Euler-Lagrange equation for the Dirichlet energy Let f be a map in C^{\infty}(\mathcal{M},\mathcal{N}) . In practice, the manifold \mathcal{N} can be chosen as an isometric embedding of \mathbb{R}^{n+d} with g\equiv\delta being the Euclidean metric, then the Dirichlet (String) energy of the map f=\left(f^{1},\cdots,f^{n},\cdots,f^{n+d}\right) is defined as the integral of the square of df as follows, \displaystyle E_{D}(f)\equiv \displaystyle\frac{1}{2}\int_{\mathcal{M}}|df|^{2}d\mathrm{vol}_{\mathcal{M}}=\frac{1}{2}\int_{\mathcal{M}}\gamma^{\alpha\beta}(x)\delta_{ij}(f)\frac{\partial f^{i}}{\partial x^{\alpha}}\frac{\partial f^{j}}{\partial x^{\beta}}\sqrt{\gamma}dx. (1) where \gamma=\det\left(\gamma_{\alpha\beta}\right) and d\mathrm{vol}_{\mathcal{M}}=\sqrt{\gamma}dx is the volume element on \mathcal{M} . Given any point x\in\mathcal{M} , we can find an open ball \mathbb{B}_{r}(x)\subset\mathcal{M} such that \mathbb{B}_{r}(x) is mapped into a coordinate chart U of \mathcal{N} by a smooth map f . In these coordinate systems, f can be viewed as a map f:\mathbb{B}_{r}(x)\to U\subset\mathbb{R}^{n} and the maps f+t\phi for \phi\in C^{\infty}_{o}\left(\mathbb{B}_{r}(x),U\right) and |t| sufficiently small determine the variation of the Dirichlet energy in ( 1 ). By the calculus of variations ( Jost, 2011 , Chap 8) , we know that the Dirichlet energy in ( 1 ) achieves a critical value if and only if a critical point f satisfies the following Euler–Lagrange equation componentwisely. \displaystyle\Delta_{\mathcal{M}}f^{m}+g^{ij}\Gamma_{st}^{m}(f(x))\frac{\partial f^{s}}{\partial x^{i}}\frac{\partial f^{t}}{\partial x^{j}}=0,\quad m=1,\cdots n+d. (2) where \displaystyle\Delta_{\mathcal{M}}f^{m}:=\frac{1}{\sqrt{\gamma}}\frac{\partial}{\partial x^{\alpha}}\left(\sqrt{\gamma}\gamma^{\alpha\beta}\frac{\partial}{\partial x^{\beta}}f^{m}\right),\quad\Gamma_{st}^{m}:=\frac{1}{2}g^{ml}\left(g_{sl,t}+g_{tl,s}-g_{st,l}\right) is the Laplace-Beltrami operator on \mathcal{M} and the Christoffel symbos in \mathcal{N} , respectively. Moreover, f\in C^{2}\left(\mathcal{M},\mathcal{N}\right) are called harmonic map, if it is