On the Curvature of Regge Metrics
Evan Gawlik, Jack McKee · 2025
Read On the Curvature of Regge Metrics by Evan Gawlik, Jack McKee (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-2510-25027.txt.
Introduction
Evan S Gawlik and Jack McKee set a bend rule for Regge metrics, a class of smooth in parts measures of space. Such measures are used when space is cut into chunks, each smooth on its own. A bend rule for them should fit both the core frame rules and the gauge rule for a change of frame.
With move frame tools, they show their bend is a spread out mass that meets both needs at once. It matches past takes on spread bend for such measures. In the flat two way case, they prove the known Gauss and Bonnet claim for these measures: total bend ties to shape type. That seals the rule in the case most used.
Catalog record
| CREATOR | Evan Gawlik, Jack McKee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 125 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-25027
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the Curvature of Regge Metrics Evan Gawlik, Jack McKee Abstract We use moving frame techniques to derive a notion of curvature for a class of piecewise-smooth Riemannian metrics called Regge metrics, showing that it is a measure that simultaneously satisfies the (weak) Cartan structure equations and the appropriate gauge transformation law. It turns out that this distributional curvature is equivalent to existing notions of densitized distributional curvature. We also investigate more closely the n = 2 n=2 case, where we prove the Gauss-Bonnet theorem for Regge metrics. 1 Introduction Since Tullio Regge’s introduction of what are today called Regge metrics—discontinuous metrics with tangential-tangential continuity—they have found widespread use in numerical models of general relativity [ 17 , 12 ] , continuum mechanics [ 15 , 12 , 16 ] , and more. In all of these applications, one of the key features of a Regge metric is that a coherent notion of curvature can be defined. Regge’s original paper [ 17 ] discussed piecewise-constant metrics on simplicial meshes, where the scalar curvature is given simply by the angle defect around codimension-2 interfaces. Further investigations by Cheeger, Müller, and Schrader [ 3 ] proved that a broader class of curvatures, called the Lipschitz-Killing curvatures, converges in measure to their smooth counterparts in the piecewise-flat setting. Later developments involved proving convergence results for Gauss, scalar, and Einstein curvature measures on higher-order Regge metrics, where the metric is piecewise-smooth rather than only piecewise-flat but retains the jump conditions on simplex boundaries [ 5 , 2 , 6 , 7 ] . The angle defect remains part of all of these curvature measures, but extra terms involving the curvature on element interiors and the jump in mean curvature/second fundamental form over element boundaries also appear. Only recently have convergence results been proved for a distributional version of the full Riemann curvature tensor in arbitrary dimension [ 8 ] . In this paper we will focus not on the numerical convergence of distributional curvature, but rather on deriving the “correct” definition of distributional curvature from first principles. The formulas ( 1 ) and ( 22 ) are equivalent to that for the densitized distributional Riemann curvature described in [ 8 ] . Our starting point is the structure equations on the orthonormal frame bundle, about which some background is provided in the following section. We take them to be the “ground truth” that define the distributional curvature. The question is, then, which moving frames f can be used to obtain a distributional curvature functional f^{*}\Omega_{\mathrm{dist}} that is actually a measure (or more precisely an order-0 current), and that has the correct gauge transformation law? We argue that the vectors comprising such a frame f need to possess single-valued normal and tangential components on codimension-1 interfaces, forcing them to be discontinuous at (generic) codimension-2 interfaces. A frame f with this property will be called “compatible” if it satisfies a few regularity hypotheses and topological constraints detailed in Definition 1 . We argue that compatible frames are the correct generalization of smooth orthonormal frames. Section 2.1 is a discussion motivating the definition of compatible frames. Sections 2.2 - 2.3 then use this definition to derive an expression for the distributional curvature functional f^{*}\Omega_{\mathrm{dist}} . Importantly, a core part of the definition of a compatible frame involves blow-ups of polytopes, essentially to ensure that the frame is regular enough to permit integration by parts, even though it is discontinuous. The idea of using blow-ups to define geometric invariants of Regge metrics is not new, appearing in [ 1 , p. 2] for much the same reason why we use them here. In Section 2 , after deriving properties that such a frame must have, we use them to obtain an expression for f^{*}\Omega_{\mathrm{dist}} that behaves as it should. We use blow-ups to help compute f^{*}\Omega_{\mathrm{dist}} in Sections 2.2 - 2.3 , and we also use them in Section 2.5 to derive a frame-independent expression for the distributional curvature, which is more practical for real computations, especially on manifolds that are not necessarily parallelizable. Our main results are stated in Theorems 1 and 2 . Roughly speaking, they say that if an n -dimensional manifold is equipped with a (curvilinear) polyhedral mesh, a Regge metric g , and a compatible frame f , then the structure equations for the connection one-form and curvature two-form can be given meaning in a distributional sense. Furthermore, the distributional curvature two-form f^{*}\Omega_{\mathrm{dist}} , when reinterpreted as an \mathrm{End}(TM) -valued two-form, is a functional \hat{R}_{\mathrm{dist}} that acts on any \mathrm{End}(TM) -valued (n-2) -form \hat{\phi} with sufficient regularity via \left\langle\left\langle\hat{R}_{\mathrm{dist}},\hat{\phi}\right\rangle\right\rangle=\sum_{T\subset M}\int_{\mathring{T}}\left\langle\hat{R}\wedge\hat{\phi}\right\rangle+\sum_{\mathring{e}\subset\mathring{M}}\int_{\mathring{e}}\left[\!\left[\left\langle\hat{\mathrm{I\!I}}_{e}\wedge i_{e}^{*}\hat{\phi}\right\rangle\right]\!\right]+\sum_{\mathring{p}\subset\mathring{M}}\int_{\mathring{p}}\left\langle\hat{\Theta}_{p}\wedge i_{p}^{*}\hat{\phi}\right\rangle. Here, the sums over T , e , and p are sums over polytopes of codimension 0, 1, and 2, respectively, and the maps i_{e}^{*} and i_{p}^{*} are pullbacks under the inclusions e\hookrightarrow M and p\hookrightarrow M . The notation [\![\cdot]\!] specifies the jump of a multi-valued quantity over the submanifold e=T\cap T^{\prime} , which in this case is simply the sum \left\langle\hat{\mathrm{I\!I}}_{e}^{T}\wedge\hat{\phi}^{T}\right\rangle+\left\langle\hat{\mathrm{I\!I}}_{e}^{T^{\prime}}\wedge\hat{\phi}^{T^{\prime}}\right\rangle . The quantities \hat{R} , \hat{\mathrm{I\!I}}_{e}^{T} , and \hat{\Theta}_{p} are \mathrm{End}(TM) -valued 2 -forms, 1 -forms, and 0 -forms that encode the curvature tensor, second fundamental form, and angle defect, respectively. The operation \left\langle\cdot\wedge\cdot\right\rangle takes a pair of \mathrm{End}(TM) -valued forms and wedges their form parts and applies a nondegenerate pairing to their endomorphism parts. Constructing compatible frames is nontrivial, and we provide an existence proof in Section 3 . Lastly in Section 4 we investigate the specialization to two dimensions, where the Gauss curvature measure can be defined independently of the frame, and we prove that there is a suitable generalization of the Gauss-Bonnet theorem. Our results are stated for extremely general oriented parallelizable manifolds and meshes. This is partly because the presentation is actually not much more complicated for a general polyhedral mesh, since many delicate analytical conditions for Stokes’ theorem must be dealt with even in the simplest cases, and existing literature treating this subject has already been developed to a very high level of generality. One caveat is that we rely on the existence of blow-ups of polytopes, but we have only been able to locate an existence theorem for blow-ups of simplices in the literature. An explanation of the types of manifolds we use in this paper, and references to relevant literature, can be found in the appendix. Some opportunities for extensions of our results are immediately apparent. The method of moving frames generalizes quite simply to indefinite (also called pseudo-Riemannian) metrics and to more general geometries as well. Lemma 3 in Section 3 is already stated for arbitrary pseudo-Riemannian metrics. While the conditions we set for compatible frames make key use of the particular properties of the structure group O(n) , it is clear where the dependence lies and what would constitute removing it. Essentially the difficulty will lie in the jump conditions at codimension-2 polytopes, and correspondingly in the angle defect terms of the distributional curvature equation. These must be phrased in terms of another one-parameter group adapted to the geometry. We use the integration theory of differential forms whenever possible, avoiding metric dependence. 1.1 Background: Geometry in a Moving Frame In the method of moving frames [ 9 ] ,