A family of quadratic finite volume element schemes over triangular meshes for elliptic equations
Introduction
As a popular numerical tool for solving partial differential equations, the finite volume method (FVM) has a long history. It possesses the local conservation property and can deal with the complex geometry boundary, and so on. For these reasons, FVM has attracted a lot of attentions and has wide applications in scientific and engineering computations, see [1], [2], [3], [4], [5], [6], [7] for an incomplete references. Finite volume element method (FVEM) is one category of FVM. There are many researchers concentrate on linear FVEM, see [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] and the references cited therein, which makes the development of the mathematical theory of linear FVEM almost as satisfactory as linear finite element method (FEM). Recently, under a mesh assumption, a unified theoretical analysis of any order FVEM over quadrilateral mesh is established and the references can be found in [22], [23], [24], [25].
However, for high order FVEM over triangular mesh, it is still a challenging work especially to establish the coercivity result. Based on the coercivity result, some researchers proposed special high order FVEM, or studied their convergence properties. For example, Vogel, Xu and Wittum [26] constructed a high order FVEM. By introducing some orthogonal conditions (on element boundary and in the interior of element), Wang and Li [27] proposed a high order FVEM such that the error has an optimal convergence rate. Recently Wang and Li [28] analyzed the superconvergence of the quadratic FVEM proposed in [27], and this quadratic scheme has been applied to solve a nonlinear elliptic problem [29].
In order to establish the coercivity result, most existing works adopted the element analysis and required a certain minimum angle condition of the triangular mesh. For instance, by assuming that the maximum angle of each triangular element is not greater than , and the ratio of the lengths of the two sides of the maximum angle belongs to , [30] obtained a coercivity result for the first quadratic FVEM. For some special quadratic FVEM, Liebau [31] required that the geometry of the triangulation is not too extreme. Xu and Zou [21] obtained the inf–sup condition for several quadratic FVEM and improved some existing results, for example, the minimum angle should be greater than or equal to for the scheme proposed in [32], for the scheme proposed in [31] and for the scheme proposed in [30]. A general framework for construction and analysis of higher-order finite volume methods was developed in [33], where three sufficient conditions for the coercivity were proposed. For a specific scheme, these conditions can be verified by a computer program. For example, by the method in [33], the minimum angle for the quadratic FVEM in [27], [28], [29] is . Later, the relationship of the uniform ellipticity, inf–sup condition and uniform local ellipticity of high order FVEM was investigated in [34]. Recently, Zou [35] proposed an unconditionally stable quadratic FVEM, regrettably, this scheme does not ensure the optimal convergence order in error norm, at least it seems so numerically. More studies about the hybrid FVEM and the Hermite FVEM can be found in [33], [36], [37].
Based on the development of quadratic FVEM over triangular mesh, one can see that most existing coercivity analysis is scheme-wise, i.e., it must be done one by one for each specific scheme. On the other hand, we observe that the quadratic FVEM proposed in [27] owns the optimal error order and superconvergence property, whose coercivity is based on a computer program proposed in [33]. So it is worth to improve the coercivity result of this scheme. Unfortunately, it is difficult to obtain the coercivity result with a simple and analytic expression through the existing techniques.
In this work, we propose and study a family (depend on a parameter ) of quadratic finite volume element schemes over triangular meshes. Precisely, we first convert the element bilinear form to a quadratic form with respect to a 6-by-6 singular element matrix. Unlike the previous works [16], [21], [31], [33], [37], here the computation of this element matrix is decomposed into two parts: the first part is the element stiffness matrix of the standard quadratic finite element method, where the finite element stiffness matrix is symmetric and singular; the second part is the tensor product of two vectors, where one vector relies on and the other vector relies on the three interior angles. Therefore, each entry of this element matrix only depends on the three interior angles of triangle and the parameter . Then, the analysis of this element matrix can be further transformed to that of a 5-by-5 symmetric matrix. As a direct consequence, we reach a sufficient condition to ensure the existence, uniqueness and coercivity of these schemes over triangular meshes. Based on the coercivity result, we obtain the optimal error estimate.
Compared with existing works, the present work has several contributions. Firstly, we provide a uniform framework to construct and analyze a family of quadratic FVE schemes with a parameter , covering some existing quadratic schemes. Secondly, by the element analysis, we obtain a sufficient condition with a simple and analytic expression, to guarantee the existence, uniqueness and coercivity result of the finite volume element solutions. Thirdly, compared to the previous works [16], [21], [27], [31], [33], [37], the above sufficient condition depends only on the scheme parameter and the interior angles of each triangular element. Therefore, it is easy to judge on any triangular mesh. In addition, in order to understand clearly the sufficient condition, two minimum angle conditions are derived and graphically shown in Fig. 4. In particular, the minimum angle condition for the quadratic scheme in [27] is improved from to .
The rest of this paper is organized as follows. A family of quadratic FVE schemes are constructed in Section 2. In Section 3 we present the coercivity result and discuss the minimum angle conditions to guarantee the existence and uniqueness of the finite volume element solutions. The proof of the coercivity result is given in Section 4 and the optimal error estimate is obtained in Section 5. Several numerical examples are provided in Section 6 to verify the theoretical results, and some concluding remarks are given in the last section.
Section snippets
A family of quadratic FVE schemes
We focus here on the isotropic steady state diffusion problem where is a bounded polygonal domain, and is a piecewise smooth function that can be bounded above and below, i.e., there exist positive constants and , such that
Suppose that the primary mesh is a triangular partition of , where the nonempty intersection of any two closed triangles is either a common edge or a common vertex and denotes the
The main results
For simplicity of exposition, we introduce some notations. For each triangular element , let be the three interior angles of . Denote and where and are defined by (6), (7), respectively. Next, let be a linear mapping that maps to , satisfying for each vertex , and for each midpoint , We remark that the above mapping was introduced
The proof of Theorem 1
In this section, the proof of Theorem 1 is performed through the element analysis, the main idea of which is presented in the following subsection.
error estimates
Once the coercivity result is obtained, the estimate of error is a standard routine. We have the following results.
Theorem 2 Assume that is shape regular and is piecewise with respect to . Assume also that the exact solution . Then, for any , under the assumption , we have
Proof By the Green’s formula, we have It follows that Suppose is the standard Lagrange interpolation of , satisfying for each
Numerical examples
In this section, we present some numerical examples to verify the theoretical results. Example 1, Example 2 are designed for continuous and discontinuous coefficients, respectively, while Example 3 is for a singular solution. In these examples, we choose and employ four types of triangular meshes. The first mesh (Mesh I) is a uniform triangular one, see Fig. 5(a), and the coordinates of the vertices are given by where is the mesh size. The third
Conclusions
We propose and study a family of quadratic FVE schemes over triangular meshes for elliptic equations. Some existing schemes in the literature can be viewed as special members of this family of schemes. The existence and uniqueness of these FVE schemes are obtained under a certain mesh geometry assumption that has a simple and analytic expression. Based on this result, we prove the coercivity results under an even stronger assumption . As a byproduct, the coercivity result in [27] is
Acknowledgments
The authors would like to thank the reviewers for their careful readings and useful suggestions. This work was partially supported by the National Natural Science Foundation of China (No. 11871009) and CAEP Foundation, China (No. CX2019028).
References (39)
- et al.
A high-order-accurate unconstructed mesh finite-volume scheme for the advection-diffusion equation
J. Comput. Phys.
(2002) - et al.
Superconvergence of quadratic finite volume method on triangular meshes
J. Comput. Appl. Math.
(2019) - et al.
Higher-order finite volume methods II: Inf-sup condition and uniform local ellipticity
J. Comput. Appl. Math.
(2014) - et al.
Finite volume methods: Foundation and analysis
- et al.
The Finite Volume Method in Computational Fluid Dynamics: An Advanced Introduction with OpenFoam and Matlab
(2016) - et al.
On the construction and analysis of high order locally conservative finite volume-type methods for one-dimensional elliptic problems
SIAM J. Numer. Anal.
(2004) High-order finite difference and finite volume weno schemes and discontinuous galerkin methods for cfd
J. Comput. Fluid Dyn.
(2003)- et al.
An Introduction To Computational Fluid Dynamics: The Finite Volume Method
(2007) - et al.
Some error estimates for the box method
SIAM J. Numer. Anal.
(1987)
On the finite volume element method
Numer. Math.
A note on the optimal -estimate of the finite volume element method
Adv. Comput. Math.
Error estimates in , and in covolume methods for elliptic and parabolic problems: a unified approach
Math. Comp.
On the accuracy of the finite volume element method based on piecewise linear polynomials
SIAM J. Numer. Anal.
On first and second order box schemes
Computing
Coercivity results of a modified -finite volume element scheme for anisotropic diffusion problems
Adv. Comput. Math.
On the finite volume element method for general self-adjoint elliptic problems
SIAM J. Numer. Anal.
The Generalized Difference Methods for Partial Differential Equations: Numerical Analysis of Finite Volume Methods
Generalized difference methods on arbitrary quadrilateral networks
J. Comput. Math.
Cited by (14)
Biquadratic element discrete duality finite volume method for solving elliptic equations on quadrilateral mesh
2024, Journal of Computational PhysicsL<sup>2</sup> error estimates for a family of cubic finite volume methods on triangular meshes
2023, Computers and Mathematics with ApplicationsA polygonal finite volume element method for anisotropic diffusion problems[Formula presented]
2023, Computers and Mathematics with ApplicationsA family of quadratic finite volume method for solving the Stokes equation
2022, Computers and Mathematics with ApplicationsA family of quadratic finite volume element schemes for anisotropic diffusion problems on triangular meshes
2022, Journal of Computational and Applied MathematicsCitation Excerpt :In order to establish the coercivity result of the quadratic FVE schemes, the existing works only faced the difficulties generated by the mesh geometry. In this paper, we extend the result of [31] to the anisotropic diffusion problems, following the element analysis approach. Obviously, we have to overcome the difficulties caused by the mesh geometry, the diffusion tensor and both combined.
High order locally conservative finite element solutions for anisotropic diffusion problems in two dimensions
2021, Computers and Mathematics with Applications