Curvature-Capacity Flow Overview

• Curvature–capacity flow is a geometric evolution where interface velocity depends on mean curvature and the normal derivative of a capacity potential, generalizing classical mean curvature flow with nonlocal effects.
• The formulation couples parabolic evolution with elliptic Dirichlet problems, ensuring local well-posedness and extending to weak formulations for handling singularities and topological changes.
• Extensions to p-capacity and associated flows provide powerful tools for deriving sharp geometric inequalities and solving variational problems in complex, nonlocal settings.

Curvature-capacity flow refers to a class of geometric evolution equations in which the normal velocity of a moving interface or hypersurface is governed jointly by its curvature and the normal derivative of a capacity potential, often coupled with elliptic PDEs characterizing the capacity in the exterior domain. This framework generalizes mean curvature flows by introducing nonlocal terms arising from various notions of electrostatic or $p$-capacity and has applications to variational inequalities, sharp geometric estimates, and the analytic resolution of geometric measure problems.

1. Classical Formulation and Mean Curvature–Nonlocal Dirichlet Flow

The archetypal curvature–capacity flow, as introduced by Yu, is described for evolving sets $\Omega_t \subset \mathbb{R}^n$ whose boundaries $\Gamma_t$ move with velocity

$V_n(x,t) = H(x,t) - \partial_n U(x,t)$

where $H$ is the mean curvature of $\Gamma_t$ and $U$ solves the exterior Dirichlet problem for the Laplacian: $$\Delta_x U(x,t) = 0 \quad \text{in} \ \mathbb{R}^n \setminus \overline{\Omega_t}, \qquad U = 1 \ \text{on} \ \Gamma_t, \qquad U \to 0 \ \text{as} \ |x| \to \infty.$$ The flow reduces heuristically to an $L^2$-gradient flow of the functional

$$E[\Omega] = \operatorname{Per}(\Omega) + \int_{\mathbb{R}^n \setminus \Omega} |\nabla U|^2 \, dx,$$

with $\Omega_t \subset \mathbb{R}^n$0 reflecting the first variation of perimeter and $\Omega_t \subset \mathbb{R}^n$1 the (nonlocal) effect of capacity.

Local well-posedness is established by recasting the moving-boundary system on a fixed domain using the Hanzawa transform. The evolution is thus governed by a quasilinear parabolic PDE for the deformation function, coupled with elliptic Dirichlet problems for the pullback capacity potentials. Well-posedness, regularity, and continuous dependence on initial data follow from Banach-space analysis in Hölder scales, leveraging Schauder estimates for both elliptic and parabolic operators (Yu, 2017).

2. Weak Formulations: Flat Flows and Minimal Barrier Flows

Upon the development of singularities (e.g., neck formation, topology change), classical solutions cease to exist. Two principal weak formulations extend the flow globally in time:

1. Flat (Minimizing Movement) Flows: Time is discretized, and at each step, a variational minimization is performed for the functional

$\Omega_t \subset \mathbb{R}^n$2

with suitable constraints on admissibility. Existence, energy-dissipation, and Hölder time-regularity of the resulting flows are obtained, but uniqueness and semigroup properties generally remain open.

2. Minimal Barrier (Comparison) Flows: Based on a geometric inclusion principle analogous to the Brakke minimal barrier approach, these are defined so that the evolution always remains the smallest barrier above the initial set. This guarantees uniqueness, the semigroup property, and monotonicity under set inclusion, though there is no variational structure and regularity issues are subtle.

Both approaches coincide with the classical flow up to the first singularity (Yu, 2017).

3. Analytic and Geometric Features

A common geometric property is that stationary solutions—where the driving velocity vanishes—are precisely critical shapes for the perimeter–capacity energy. In the radially symmetric case, the flow reduces to an explicit ODE for the radius, converging monotonically to a stationary solution determined by the curvature–capacity balance. Lemmas provide regularity of the capacity potential, density bounds at the free boundary, and control of set-symmetric differences via coarea estimates.

In the planar curve case, long-time existence and convergence to a unique stationary ball are established under a strict convexity–capacity condition $\Omega_t \subset \mathbb{R}^n$3, mirroring results for classical mean curvature flow but with nonlocal effects driven by the capacity potential. The evolution is the $\Omega_t \subset \mathbb{R}^n$4-gradient flow for $\Omega_t \subset \mathbb{R}^n$5, and $\Omega_t \subset \mathbb{R}^n$6 is strictly decreasing (Caffarelli et al., 2017).

4. Curvature–Capacity Flows for $\Omega_t \subset \mathbb{R}^n$7-Capacity Functionals

A significant generalization arises in the context of the $\Omega_t \subset \mathbb{R}^n$8-capacitary Orlicz–Minkowski problem. Here, one seeks convex bodies $\Omega_t \subset \mathbb{R}^n$9 such that the support function $\Gamma_t$0 and $\Gamma_t$1-capacity equilibrium potential $\Gamma_t$2 satisfy on the sphere: $\Gamma_t$3 where $\Gamma_t$4 and $\Gamma_t$5 are prescribed, $\Gamma_t$6 is the Gauss curvature, and $\Gamma_t$7. The inverse Gauss curvature–capacity flow is employed to produce smooth solutions, with the embedding or support function evolving via a fully nonlinear parabolic PDE coupled to the $\Gamma_t$8-capacity potential (Chen et al., 2023).

Key analytic steps include:

• Short-time existence and preservation of convexity via degree $\Gamma_t$9 homogeneity in the curvatures.
• Uniform $V_n(x,t) = H(x,t) - \partial_n U(x,t)$0-$V_n(x,t) = H(x,t) - \partial_n U(x,t)$1 bounds using maximum principles and normalization identities.
• $V_n(x,t) = H(x,t) - \partial_n U(x,t)$2 bounds and higher regularity by auxiliary function maximum arguments à la Firey.
• Monotonicity of the $V_n(x,t) = H(x,t) - \partial_n U(x,t)$3-capacity along the flow, with strict increase except at stationary points.
• Convergence (via compactness and subsequences) to solutions of the Monge–Ampère form.
• Uniqueness is established when $V_n(x,t) = H(x,t) - \partial_n U(x,t)$4 satisfies a scaling inequality and $V_n(x,t) = H(x,t) - \partial_n U(x,t)$5.

Existence extends to weak solutions for general finite Borel measures, beyond the smooth category (Chen et al., 2023).

5. Capacity Inequalities and Geometric Flows

Inverse mean curvature flow (IMCF) and its anisotropic generalizations underpin the derivation of sharp geometric inequalities for capacity in both Euclidean and hyperbolic spaces, and in the presence of boundary and conical singularities (Cruz, 2017, Li et al., 2023). The IMCF is given by

$V_n(x,t) = H(x,t) - \partial_n U(x,t)$6

and, in the conical or boundary setting, has additional Neumann-type conditions. It preserves star-shapedness and mean convexity, with explicit monotonicity formulas for integrals of mean curvature and area along leaves $V_n(x,t) = H(x,t) - \partial_n U(x,t)$7 of the foliation. These estimates yield sharp and, in many cases, equality-achieving upper bounds for $V_n(x,t) = H(x,t) - \partial_n U(x,t)$8-capacities, generalizing classical isoperimetric and Alexandrov–Fenchel inequalities.

In specific contexts, capacity bounds take the form: $V_n(x,t) = H(x,t) - \partial_n U(x,t)$9 for mean-convex free boundaries in convex cones, saturated only for spherical sectors. Penrose-type inequalities, involving capacity and ADM mass in asymptotically flat manifolds with boundary, also emerge from such flows. In Euclidean space, weak IMCF yields capacity bounds with sharp constants when the boundary is a sphere or Wulff shape (Cruz, 2017, Li et al., 2023).

6. Nonlocal and Random Geometry Effects

Curvature–capacity flows are sensitive to nonlocal geometric effects. For instance, evolution by mean curvature coupled with the capacity potential can exhibit strict monotonicity of the combined energy, direct convergence to symmetric equilibria, and regularization properties preventing finite-time singularity formation under convexity-type constraints (Caffarelli et al., 2017).

In the context of flow through randomly curved manifolds, statistical correlations between mass flux reduction and average Ricci scalar display bifurcation based on the scale of curvature perturbations. The effects of “curvature-interference” are captured numerically and analytically, revealing distinct scaling behavior for flux diminution and validating the geometric sensitivity of flow in highly complex spaces (Mendoza et al., 2012).

7. Summary Table: Key Flows and Features

Flow Type	Main Evolution Law	Existence/Regularity	Limiting/Energy Property
Classical MCND Flow	$H$0	Short-time smooth	Energy decreasing (Perimeter + Dirichlet)
Inverse MC Flow (IMCF)	$H$1	Global for convex	Supplies monotonic capacity/curvature bounds
Inverse Gauss-capacity	Nonlinear PDE (support function and $H$2-cap. potential)	Global smooth for $H$3	Monotonic $H$4-capacity, convergence to MA
Barrier/Flat Weak	Variational minimization / inclusion-comparison	Global weak	Weak gradient flow / uniqueness, semigroup

The curvature–capacity flow spectrum, therefore, forms a coherent analytic and geometric theory integrating classical curvature-driven flows with nonlocal functionals rooted in variational capacity, with extended applicability to sharp geometric inequalities, flow methods for PDE/geometric measure problems, and broader dynamical contexts encompassing nonlocality and statistical geometry.