Convergence of Discrete Exterior Calculus for the Hodge-Dirac Operator

Radovan Dabetić, Ralf Hiptmair · 2025

Read Convergence of Discrete Exterior Calculus for the Hodge-Dirac Operator by Radovan Dabetić, Ralf Hiptmair (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-2507-19405.txt.

Introduction

Radovan Dabetić and Ralf Hiptmair give a short proof that a discrete method for a key field operator converges, in this 2025 paper. The operator, named for Hodge and Dirac, sits at the heart of field equations. Discrete exterior calculus rebuilds such operators on grids and meshes.

Users need proof that the grid version tends to the true answer as the mesh refines. The authors supply that proof in compact form. They lean on a recent framework built with generalized Whitney forms. That framework was made for close problems about the Hodge-Laplacian. A short, clean proof is easier for others to check and reuse. It strengthens the base for simulation tools in physics and engineering.

Catalog record

CREATORRadovan Dabetić, Ralf Hiptmair
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-19405

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Convergence of Discrete Exterior Calculus for the Hodge-Dirac Operator Radovan Dabetić, Ralf Hiptmair Abstract A short proof of convergence for the discretization of the Hodge-Dirac operator in the framework of discrete exterior calculus (DEC) is provided using the techniques established in [ [ Johnny Guzmán and Pratyush Potu , A Framework for Analysis of DEC Approximations to Hodge-Laplacian Problems using Generalized Whitney Forms , arXiv Preprint 2505.08934 , 2025 ] ] . 1 Introduction There are two fundamental operators associated with L^{2} -de Rham complexes, the Hodge-Laplacian and the Hodge-Dirac operator. On triangulated domains \Omega\subset\mathbb{R}^{n} of Euclidean space, there are two fundamental ways to discretize de Rham complexes. The first is a finite-element approach employing discrete differential forms known as finite element exterior calculus (FEEC). The second has the flavor of a finite-volume technique and is called discrete exterior calculus (DEC). A comprehensive a priori convergence theory for the DEC approximation of Hodge-Laplacians has recently been achieved by J. Guzmán and P. Potu in the breakthrough work [ 4 ] . The present paper harnesses the novel techniques from [ 4 ] to establish the convergence of the DEC discretization of the Hodge-Dirac operator associated with the L^{2} de Rham Hilbert complex on a bounded domain of {n} -dimensional Euclidean space. This work supplements [ 4 ] , from where we borrow the bulk of our notation, often without defining it again. We also refer to [ 4 ] for background information on FEEC and DEC and a discussion of pertinent literature. The reader is advised to study [ 4 ] before reading the present paper. 2 The Hodge-Dirac Operator Let \Omega\subset\mathbb{R}^{n} be a bounded, Lipschitz, polytopal, and topologically trivial domain and we write {\Lambda^{k}(\Omega)} for the space of smooth k -forms thereon. As in [ 4 ] the exterior derivative operators are denoted by d^{k}:{\Lambda^{k}(\Omega)}\to{\Lambda^{k+1}(\Omega)} , 0\leq k<n , the (Euclidean) Hodge star operators by \star_{k} and the codifferential operators by \delta_{k}:=(-1)^{k}\star_{k-1}^{-1}d^{n-k}\star_{k}:{\Lambda^{k}(\Omega)}\to{\Lambda^{k-1}(\Omega)},k=1,\dots,n . The Hodge star operators induce inner products on {\Lambda^{k}(\Omega)} . Definition 2.1 . The L^{2} inner product on two k -forms \omega and \mu is given by \displaystyle\left\langle\omega,\mu\right\rangle_{L^{2}{\Lambda^{k}(\Omega)}}:=\int_{\Omega}\omega\wedge\star\mu. We denote by {\Lambda(\Omega)}:=\bigoplus_{k=0}^{n}{\Lambda^{k}(\Omega)} the exterior algebra of (smooth) differential forms on \Omega and write \displaystyle\differential:=\begin{pmatrix}0&&\\ d^{0}&0&\\ &d^{1}&0&\\ &&\ddots&\ddots\end{pmatrix},\quad\delta:=\begin{pmatrix}0&\delta_{1}&\\ &0&\delta_{2}\\ &&0&\ddots\\ &&&\ddots\end{pmatrix} (1) for the exterior derivative and codifferential on {\Lambda(\Omega)} . We equip {\Lambda(\Omega)} with the natural Hilbert space structure by combining the inner products from Definition 2.1 . For \mathfrak{u}\equiv(\mathsf{u}_{0},\dots,\mathsf{u}_{n}),\mathfrak{v}\equiv(\mathsf{v}_{0},\dots,\mathsf{v}_{n})\in{\Lambda(\Omega)} we set \displaystyle\left\langle\mathfrak{u},\mathfrak{v}\right\rangle_{L^{2}\Lambda}:=\sum_{k=0}^{n}\left\langle\mathsf{u}_{k},\mathsf{v}_{k}\right\rangle_{L^{2}\Lambda}\;. Write L^{2}{\Lambda(\Omega)}:=\bigoplus_{k=0}^{n}L^{2}{\Lambda^{k}(\Omega)} , where L^{2}{\Lambda^{k}(\Omega)} is the space of square-integrable k -forms, i.e. k -forms with coefficients in L^{2}(\Omega) . Also refer to [ 2 , Section 6.2.6] , where Sobolev spaces of differential forms are introduced. Let \displaystyle H{\Lambda(\Omega)}:=\left\{\mathfrak{u}\in L^{2}{\Lambda(\Omega)}:\differential\mathfrak{u}\in L^{2}{\Lambda(\Omega)}\right\}\;, and define \mathring{V}:=\bigoplus_{k=0}^{n-1}\mathring{H}{\Lambda^{k}(\Omega)}\oplus L^{2}_{*}{\Lambda^{n}(\Omega)} , where \displaystyle L^{2}_{*}{\Lambda^{n}(\Omega)}:=\left\{\mathfrak{v}\in L^{2}{\Lambda^{n}(\Omega)}:\int_{\Omega}\mathfrak{v}=0\right\}\;, (2) and \mathring{H}{\Lambda^{k}(\Omega)} is the space of functions in H{\Lambda^{k}(\Omega)} with vanishing trace on \partial\Omega , see [ 2 , Section 6.2.6] . Also let H^{*}{\Lambda(\Omega)} be the domain of \delta , see also [ 2 , Section 6.2.6] . Definition 2.2 . The Hodge-Dirac operator is \DD:=\differential+\delta with domain of definition \mathcal{D}(\DD):=\mathring{H}\Lambda(\Omega)\cap H^{*}\Lambda(\Omega) , where the domain of \differential is \mathring{H}\Lambda(\Omega) and that of \delta is H^{*}\Lambda(\Omega) . Taking the cue from [ 6 ] we put the focus on the following boundary value problem 1 1 1 Note that the boundary conditions are included implicitly in the domain of the operator. for the Dirac operator: Given \mathfrak{f}\in L^{2}{\Lambda(\Omega)} , seek \mathfrak{u}\in\mathcal{D}(\DD)\cap(\ker\DD)^{\perp},\mathfrak{p}\in\ker\DD such that \DD\mathfrak{u}+\mathfrak{p}=\mathfrak{f}. (3) Corollary 8 of [ 6 ] tells us the well-posedness of the following weak form of ( 3 ): Given \mathfrak{f}\in L^{2}{\Lambda(\Omega)} , seek \mathfrak{u}\in\mathring{H}{\Lambda(\Omega)},\mathfrak{p}\in\ker\DD such that \displaystyle\begin{aligned} \left\langle\differential\mathfrak{u},\mathfrak{v}\right\rangle_{L^{2}\Lambda}+\left\langle\mathfrak{u},\differential\mathfrak{v}\right\rangle_{L^{2}\Lambda}+\left\langle\mathfrak{p},\mathfrak{v}\right\rangle_{L^{2}\Lambda}&=\left\langle\mathfrak{f},\mathfrak{v}\right\rangle_{L^{2}\Lambda}&&\forall\mathfrak{v}\in\mathring{H}\Lambda(\Omega)\\ \left\langle\mathfrak{u},\mathfrak{v}\right\rangle_{L^{2}\Lambda}&=0&&\forall\mathfrak{v}\in\ker\DD.\end{aligned} (4) As we are working with a domain with trivial topology, \ker\DD (the space of harmonic forms) is trivial (see [ 2 , Section 4.3] for more information) except for constant n -forms, i.e. \ker\DD|_{\mathring{V}}=\{0\} , so that we can consider the following simpler problem: Given \mathfrak{f}\in L^{2}{\Lambda(\Omega)} with \int_{\Omega}\mathsf{f}_{n}=0 2 2 2 Or given the general case, we can recover \mathfrak{p} by taking the mean of the n -form in \mathfrak{f} and then subtract the mean to get a suitable right-hand side. , seek \mathfrak{u}\in\mathring{V} such that \displaystyle\left\langle\differential\mathfrak{u},\mathfrak{v}\right\rangle_{L^{2}\Lambda}+\left\langle\mathfrak{u},\differential\mathfrak{v}\right\rangle_{L^{2}\Lambda}=\left\langle\mathfrak{f},\mathfrak{v}\right\rangle_{L^{2}\Lambda}\quad\forall\mathfrak{v}\in\mathring{V}. (5) 3 DEC Discretization of the Hodge-Dirac Operator In order to discretize ( 5 ) we rely on an oriented well-centered simplicial mesh 3 3 3 See [ 3 ] for more information on cellular complexes. \mathcal{T} of \Omega , that is, as stipulated by [ 5 , Definition 2.4.3] the circumcenter of any simplex of \mathcal{T} lies in its interior. Write \mathcal{T}^{k} for the k -cells of \mathcal{T} and C^{k}(\mathcal{T}) the k -cochains on \mathcal{T} . Furthermore, let \tilde{\mathcal{T}} designate the (orthogonal) dual mesh of \mathcal{T} and let *\sigma\in\tilde{\mathcal{T}} for \sigma\in\mathcal{T}^{k} denote the dual cell of \sigma , see [ 4 , Section 3.2] . We denote by \partial^{k}:C^{k}(\mathcal{T})\to C^{k+1}(\mathcal{T}) the coboundary operator, with respect to the coordinate basis C^{k}(\mathcal{T}) and C^{k+1}(\mathcal{T}) represented by the incidence matrix of oriented k and (k+1) -cells. Let \hbox to5.91pt{\vbox to5.65pt{\pgfpicture\makeatletter\hbox{\hskip 2.9562pt\lower-2.55202pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke} \lxSVG@begingroup@{fill} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} {{\lx@inpgf@ignorespaces}}{ {}{{}}{} {}{} {}{} {}{} {}{} {\lx@inpgf@ignorespaces}\lxSVG@begingroup@{_scopebegin} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\lxSVG@roundjoin\lxSVG@begingroup@{stroke-linejoin} \lx@inpgf@ignorespaces{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 3.94 L -2.31 -3.19 L 3.74 1.22 L -3.74 1.22 L 2.31 -3.19 Z}{fill:none} \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}_{k}:C^{k}(\mathcal{T})\to C^{n-k}(\tilde{\mathcal{T}}) , k=0,\dots,n , be the customary discrete Hodge stars in the context of discrete exterior calculus, see [ 4 , Definition 3.6] . Let \hbox to5.91pt{\vbox to5.65pt{\pgfpicture\makeatletter\hbox{\hskip 2.9562pt\lower-2.55202pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke} \lxSVG@begingroup@{fill} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} {{\lx@inpgf@ignorespaces}}{ {}{{}}{} {}{} {}{} {}{} {}{} {\lx@inpgf@ignorespaces}\lxSVG@begingroup@{_scopebegin} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\lxSVG@roundjoin\lxSVG@begingroup@{stroke-linejoin} \lx@inpgf@ignorespaces{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 3.94 L -2.31 -3.19 L 3.74 1.22 L -3.74 1.22 L 2.31 -3.19 Z}{fill:none} \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}_{k}^{-1} denote its inverse as in [ 4 , Definition 3.6] . Definition 3.1 ( [ 4 , Section 4.5] ) . For \mathsf{u},\mathsf{v}\in C^{k}(\mathcal{T}) , define the inner product \displaystyle\left\llbracket\mathsf{u},\mathsf{v}\right\rrbracket_{k}:=\sum_{\sigma\in\mathcal{T}^{k}}\frac{|*\sigma|}{|\sigma|}\left\langle\mathsf{u},\sigma\right\rangle\left\langle\mathsf{v},\sigma\right\rangle, (6) where \left\langle\mathsf{u},\sigma\right\rangle\equiv\mathsf{u}(\sigma) is the duality pairing of C^{k}(\mathcal{T}) and \mathcal{T}^{k} . We denote the norm induced by this inner product by {\left|\kern-1.07639pt\left|\kern-1.07639pt\left|\cdot\right|\kern-1.07639pt\right|\kern-1.07639pt\right|}_{k} . Let C(\mathcal{T}):=\bigoplus_{k}C^{k}(\mathcal{T}) and define an inner product thereon for \mathbf{u},\mathbf{v}\in C(\mathcal{T}) by \displaystyle\left\llbracket\mathbf{u},\mathbf{v}\right\rrbracket:=\sum_{k=0}^{n}\left\llbracket\mathsf{u}_{k},\mathsf{v}_{k}\right\rrbracket_{k}, where \mathbf{u}=(\mathsf{u}_{0},\dots,\mathsf{u}_{n}),\mathsf{u}_{k}\in C^{k}(\mathcal{T}) . For the induced norm we write {\left|\kern-1.07639pt\left|\kern-1.07639pt\left|\cdot\right|\kern-1.07639pt\right|\kern-1.07639pt\right|} . Definition 3.2 (Discrete Codifferential, [ 4 , Definition 3.7, Definition 4.3] ) . The discrete codifferential \delta^{\textnormal{DEC}}_{k+1}:C^{k+1}(\mathcal{T})\to C^{k}(\mathcal{T}) is defined as \displaystyle\delta^{\textnormal{DEC}}_{k+1}:=(-1)^{k+1}\hbox to5.91pt{\vbox to5.65pt{\pgfpicture\makeatletter\hbox{\hskip 2.9562pt\lower-2.55202pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke} \lxSVG@begingroup@{fill} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} {{\lx@inpgf@ignorespaces}}{ {}{{}}{} {}{} {}{} {}{} {}{} {\lx@inpgf@ignorespaces}\lxSVG@begingroup@{_scopebegin} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\lxSVG@roundjoin\lxSVG@begingroup@{stroke-linejoin} \lx@inpgf@ignorespaces{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 3.94 L -2.31 -3.19 L 3.74 1.22 L -3.74 1.22 L 2.31 -3.19 Z}{fill:none} \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}_{k}^{-1}\tilde{\partial}^{n-(k+1)}\hbox to5.91pt{\vbox to5.65pt{\pgfpicture\makeatletter\hbox{\hskip 2.9562pt\lower-2.55202pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke} \lxSVG@begingroup@{fill} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} {{\lx@inpgf@ignorespaces}}{ {}{{}}{} {}{} {}{} {}{} {}{} {\lx@inpgf@ignorespaces}\lxSVG@begingroup@{_scopebegin} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\lxSVG@roundjoin\lxSVG@begingroup@{stroke-linejoin} \lx@inpgf@ignorespaces{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 3.94 L -2.31 -3.19 L 3.74 1.22 L -3.74 1.22 L 2.31 -3.19 Z}{fill:none} \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}_{k+1},\quad k=0,\dots,n-1, where \tilde{\partial}^{n-(k+1)} denotes the coboundary operator on C^{n-(k+1)}(\tilde{\mathcal{T}}) . In analogy to \differential and \delta from ( 1 ) we set \displaystyle\DECd:=\begin{pmatrix}0&&\\ \partial^{0}&0&\\ &\partial^{1}&0&\\ &&\ddots&\ddots\end{pmatrix},\quad\delta^{\textnormal{DEC}}:=\begin{pmatrix}0&\delta^{\textnormal{DEC}}_{1}&\\ &0&\delta^{\textnormal{DEC}}_{2}\\ &&0&\ddots\\ &&&\ddots\end{pmatrix}, the exterior derivative and co-derivative on the exterior algebra of cochains mapping C(\mathcal{T})\to C(\mathcal{T}) . Lemma 3.1 (Discrete Adjoint) . It holds that \displaystyle\left\llbracket\delta^{\textnormal{DEC}}\mathbf{u},\mathbf{v}\right\rrbracket=\left\llbracket\mathbf{u},\DECd\mathbf{v}\right\rrbracket\quad\forall\mathbf{u},\mathbf{v}\in C(\mathcal{T}). Proof. This follows directly from [ 4 , Lemma 4.12] and the definition of \left\llbracket\cdot,\cdot\right\rrbracket . ∎ Definition 3.3 . The DEC Hodge-Dirac operator is \DECD:=\DECd+\delta^{\textnormal{DEC}}. Definition 3.4 .

Read the complete text