Debian Codegraph @

include/deal.II/fe/fe_simplex_p_bubbles.h

raw
1 // ------------------------------------------------------------------------
2 //
3 // SPDX-License-Identifier: LGPL-2.1-or-later
4 // Copyright (C) 2021 - 2023 by the deal.II authors
5 //
6 // This file is part of the deal.II library.
7 //
8 // Part of the source code is dual licensed under Apache-2.0 WITH
9 // LLVM-exception OR LGPL-2.1-or-later. Detailed license information
10 // governing the source code and code contributions can be found in
11 // LICENSE.md and CONTRIBUTING.md at the top level directory of deal.II.
12 //
13 // ------------------------------------------------------------------------
14
15 #ifndef dealii_fe_fe_p_bubbles_h
16 #define dealii_fe_fe_p_bubbles_h
17
18 #include <deal.II/base/config.h>
19
20 #include <deal.II/base/polynomials_barycentric.h>
21
22 #include <deal.II/fe/fe_simplex_p.h>
23
24 DEAL_II_NAMESPACE_OPEN
25
26 /**
27 * @brief Enriched version of FE_SimplexP that can be used with nodal
28 * quadrature.
29 *
30 * Many explicit time integration schemes require solving a @ref GlossMassMatrix "mass matrix" at
31 * each time step. There are various ways around this requirement - for
32 * example, step-48 replaces the mass matrix with a diagonal approximation,
33 * which makes the solution step trivial. In step-48, and also commonly for
34 * tensor-product elements, this is done by computing the mass matrix with a
35 * lower-order quadrature point based on the nodes of the finite element
36 * (i.e., the nodal quadrature rule one obtains by using the shape functions
37 * as an interpolatory basis).
38 *
39 * A major drawback of standard simplex-based finite elements is that they
40 * cannot be used with nodal quadrature since some of the quadrature weights
41 * end up being either zero or negative, resulting in either an unsolvable or
42 * unstable approximation to the mass matrix. For example: the shape functions
43 * of FE_SimplexP<2>(2) with support points at vertices have mean values of
44 * zero so that element cannot be used with mass lumping.
45 *
46 * This element avoids this issue by replacing the shape functions of
47 * FE_SimplexP with an augmented space amenable to the construction of nodal
48 * quadrature rules. For example, on the triangle a single basis function is
49 * added corresponding to interpolation at the centroid (and all other basis
50 * functions are updated to preserve the partition of unity property). This
51 * results in shape functions with positive means (i.e., a valid nodal
52 * quadrature formula). Similarly, in 3d, the polynomial space of
53 * FE_SimplexP<3>(2) is enriched with five additional degrees of freedom (where
54 * four have support points at face centroids and one has a support point at the
55 * centroid) to enable construction of valid nodal quadrature rule.
56 *
57 * Since this FE space includes bubbles (i.e., extra functions which are
58 * nonzero only on element interiors), the polynomial degrees of the component
59 * basis functions are higher than the actual approximation degree of the
60 * element. For example, with a constructor argument <code>degree = 2</code>
61 * in 3d, the polynomials are in fact cubic (degree 3) but the order of the
62 * approximation is the same as if we were using quadratic (degree 2) finite
63 * elements.
64 *
65 * The 2d quadratic element was first described in @cite fried1975finite. The
66 * 3d quadratic element implemented here was first described in
67 * @cite Geevers_2018. Higher degree elements amendable to lumping exist but
68 * are not yet implemented in this class.
69 */
70 template <int dim, int spacedim = dim>
71 class FE_SimplexP_Bubbles : public FE_SimplexPoly<dim, spacedim>
72 {
73 public:
74 /**
75 * Constructor, taking the approximation degree as an argument. The
76 * polynomial space is typically one degree higher than the approximation
77 * space for this element: see the general documentation of this class for
78 * more information.
79 *
80 * @note For <code>degree == 1</code> this element is equivalent to
81 * FE_SimplexP(1).
82 */
83 FE_SimplexP_Bubbles(const unsigned int degree);
84
85 /**
86 * @copydoc dealii::FiniteElement::clone()
87 */
88 virtual std::unique_ptr<FiniteElement<dim, spacedim>>
89 clone() const override;
90
91 /**
92 * Return a string that uniquely identifies a finite element. This class
93 * returns <tt>FE_SimplexP_Bubbles<dim,spacedim>(degree)</tt>, with
94 * @p dim, @p spacedim, and @p degree replaced by appropriate values. As
95 * usual, @p spacedim is omitted in the codimension zero case.
96 */
97 virtual std::string
98 get_name() const override;
99
100 protected:
101 /**
102 * Degree of the approximation (i.e., the constructor argument).
103 */
104 unsigned int approximation_degree;
105 };
106
107 DEAL_II_NAMESPACE_CLOSE
108
109 #endif