---
title: Sub-static Heintze-Karcher Inequality
url: https://www.emergentmind.com/topics/sub-static-heintze-karcher-inequality
type: topic
---

# Sub-static Heintze-Karcher Inequality

The sub-static Heintze–Karcher inequality is a weighted mean-curvature inequality on a Riemannian triple \((M,g,V)\) satisfying a sub-static tensor condition of the form
\[
\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\ge 0,
\]
or, in the notation used in later rigidity work,
\[
Q:=f\operatorname{Ric}-\nabla\nabla f+\Delta f\,g\ge 0.
\]
Its basic form compares a weighted boundary integral \(\int_{\Sigma}\frac{V}{H}\) with a weighted bulk term \(\int_{\Omega}V\), and in the presence of horizon components it includes additional boundary contributions. The subject emerged from weighted Reilly formulas in model spaces [1405.4518], was formulated invariantly for general sub-static manifolds by Li–Xia [1603.02201], was sharpened in the horizon setting and analyzed in its equality case [2307.04253], and has since developed toward shifted-curvature versions in warped products [2504.15109], static special cases proved by spinorial methods [1806.01120], and applications to inverse-mean-curvature-flow monotonicity and uniqueness in asymptotically locally hyperbolic static geometry [2509.18026].

## 1. Sub-static structure and weighted curvature tensors

Li–Xia define a Riemannian triple \((M,g,V)\) to be **static** if
\[
\Delta V\, g-\nabla^2 V + V\,\mathrm{Ric}=0,
\]
and **sub-static** if
\[
\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\ge 0.
\]
Equivalently, the tensor
\[
Q=\frac{1}{V}\big(\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\big)
\]
is nonnegative definite wherever \(V>0\). In the notation of the equality-case paper, the same structure is written as
\[
Q=f\operatorname{Ric}-\nabla\nabla f+\Delta f\,g\ge 0,
\]
and the paper explicitly refers to \(Q\) as the “substatic Ricci tensor” [1603.02201; 2307.04253].

This framework is the weighted analogue of the role played by \(\mathrm{Ric}\ge 0\) in the classical Reilly–Ros theory. In particular, the weighted potential \(V\) is not auxiliary: it is part of the ambient geometry, and the combination \(\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\) is the tensor that governs the sign of the interior term in generalized Reilly identities. Qiu–Xia’s weighted Reilly formula already contains this tensorial structure, although the paper does not use the later terminology “sub-static” [1405.4518].

The ambient boundary structure is equally important. Li–Xia allow
\[
\partial\Omega=\Sigma\cup\Big(\bigcup_{l=1}^T N_l\Big),
\]
where \(\Sigma\) is the outermost boundary hypersurface and the \(N_l\) are inner boundary components. Their inner boundary condition requires that each \(N_l\) be minimal, that \(V>0\) in \(\Omega\) and \(V=0\) on \(N_l\), and that
\[
\frac{\Delta V\,g-\nabla^2V}{V}(\nu,\nu)>0
\]
on each \(N_l\). In later language these \(N_l\) play the role of horizons [1603.02201].

## 2. Core inequalities and the generalized Reilly mechanism

The basic Li–Xia Heintze–Karcher inequality states that if \((M^n,g,V)\) is an \(n\)-dimensional sub-static Riemannian triple, \(\Omega\subset M\) is bounded, \(\partial\Omega=\Sigma\cup(\bigcup_{l=1}^T N_l)\) satisfies the inner boundary condition, \(\Sigma\) is mean convex, and \(V>0\) in \(\Omega\), then
\[
n\left(\int_\Omega V\,d\Omega+\sum_{l=1}^T c_l\int_{N_l} V_{,\nu}\,dA\right) \le (n-1)\int_\Sigma \frac{V}{H}\,dA.
\]
If \(T=0\), this reduces to
\[
n\int_\Omega V\,d\Omega \le (n-1)\int_\Sigma \frac{V}{H}\,dA.
\]
Under a local warped-product structure near each inner boundary component,
\[
g=dr^2+\lambda(r)^2 g_{N_l},\qquad V=\lambda'(r),\qquad \lambda''(0)\neq 0,
\]
the horizon term becomes explicit:
\[
n\int_\Omega V\,d\Omega +\sum_{l=1}^T \lambda(0)^n \operatorname{Vol}(N_l,g_{N_l}) \le (n-1)\int_\Sigma \frac{V}{H}\,dA.
\]
If equality holds, then \(\Sigma\) is umbilical [1603.02201].

The analytic engine is Li–Xia’s generalized Reilly formula:
\[
\begin{aligned}
&\int_\Omega V\left((\Delta f-\tfrac{\Delta V}{V}f)^2-\left|\nabla^2f-\tfrac{\nabla^2V}{V}f\right|^2\right)\,d\Omega \\
={}& \int_{\partial\Omega} \Big( V h(\nabla z,\nabla z) +2V u\,\Delta z +V H u^2 +V_{,\nu}|\nabla z|^2 +2z\,\nabla^2V(\nabla z,\nu) \Big)\,dA \\
&\quad +\int_{\partial\Omega} \Big( -2zu(\Delta V+H V_{,\nu}) -z^2 \frac{\nabla_\nu V}{V} \Big)\,dA \\
&\quad +\int_\Omega V\,Q\!\left(\nabla f-\frac{\nabla V}{V}f,\, \nabla f-\frac{\nabla V}{V}f\right)\,d\Omega .
\end{aligned}
\]
Here \(z=f|_{\partial\Omega}\), \(u=f_{,\nu}\), and \(VQ=\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\). The proof of the Heintze–Karcher inequality solves
\[
\begin{cases}
\Delta f-\dfrac{\Delta V}{V}f = 1 & \text{in }\Omega,\\[1ex]
f=0 & \text{on }\Sigma,\\
f=c_l & \text{on }N_l,
\end{cases}
\]
then combines the nonnegativity of \(Q\) with the pointwise tensor inequality
\[
\left|\nabla^2f-\frac{\nabla^2V}{V}f\right|^2 \ge \frac1n \left(\Delta f-\frac{\Delta V}{V}f\right)^2.
\]
An integration-by-parts identity,
\[
\int_\Sigma V u\,dA - \sum_{l=1}^T c_l\int_{N_l} V_{,\nu}\,dA = \int_\Omega V\,d\Omega,
\]
and a final Hölder inequality on \(\Sigma\) then produce the boundary-to-volume estimate [1603.02201].

Qiu–Xia’s earlier weighted Reilly formula plays the same structural role in constant-curvature models. In \(\mathbb H^n\) and \(\mathbb S^n_+\), with \(V=\cosh r\) or \(V=\cos r\), they rederive the weighted inequalities
\[
\int_M \frac{V}{H}\,dA \ge n\int_\Omega V\,d\Omega,
\]
while for \(\mathrm{Sect}\ge -1\) they prove
\[
\int_M \frac{V}{H}\,dA \ge \int_M V_\nu\,dA=\int_\Omega \Delta V\,d\Omega.
\]
This precursor is not yet a general sub-static theorem, but the tensor \((\Delta V)g-\nabla^2V+V\,\mathrm{Ric}\) is already the decisive quantity [1405.4518].

## 3. Equality, rigidity, and warped-product splitting

The equality case is the main subject of the rigidity paper. For a substatic manifold with connected horizon boundary \(\partial M=\{f=0\}\), assuming
\[
\frac{\nabla \nabla f}{f} \in C^{0, \alpha}(M \cup \partial M),
\]
and a connected, smooth, strictly mean-convex hypersurface \(\Sigma\) homologous to \(\partial M\), equality in
\[
\frac{n-1}{n}\int_\Sigma \frac{f}{H}\, d \sigma
\geq \int_\Omega f\, d \mu + c_{\partial M} \int_{\partial M} \abs{\nabla f}\, d\sigma
\]
holds if and only if the region \(\Omega\) enclosed by \(\Sigma\) and \(\partial M\) is isometric to
\[
\left([s_0, \overline{s}] \times \partial M, \frac{ds \otimes ds}{f(s)^2} + s^2 g_{\partial M}\right).
\]
The more general disconnected statement yields a corresponding warped-product description on each component [2307.04253].

The proof starts from the sharpened remainder identity
\[
\begin{split}
\frac{n-1}{n} \int_{\Sigma}\frac{f}{H}\, d\sigma -\int_{\Omega} f\, d\mu
- \sum_{j } c_j \int_{N_j} \abs{\nabla f}\, d\sigma
\ge
\frac{n}{n-1}\int_\Omega
&\bigg|\nabla \nabla u - \frac{\Delta u}{n} g \\
&\qquad - u\left(\frac{\nabla \nabla f}{f}  - \frac{\Delta f}{nf} g\right)\bigg\vert^2 \\
&+ Q\left(\nabla u - \frac{u}{f} \nabla f, \nabla u - \frac{u}{f} \nabla f\right) d\mu,
\end{split}
\]
where \(u\) solves the associated Dirichlet problem. Equality forces
\[
\nabla \nabla u - \frac{\Delta u}{n}g - u \left(\frac{\nabla \nabla f}{f} - \frac{\Delta f}{nf} g\right) = 0
\]
and
\[
Q\left(\nabla u - \frac{u}{f}\nabla f, \nabla u - \frac{u}{f}\nabla f\right) = 0.
\]
With the conformal change \(\tilde g=f^{-2}g\) and \(\varphi=u/f\), the first identity becomes
\[
\nabla\nabla_{\tilde g}\varphi - \frac{\Delta_{\tilde g}\varphi}{n}\tilde g = 0,
\]
and the proof then uses a Tashiro-type splitting argument and the vanishing of the substatic tensor in a distinguished direction to show that the enclosed region is warped [2307.04253].

In substatic warped products
\[
\left([s_0, \overline{s}) \times N, \frac{ds \otimes ds}{f(s)^2} + s^2 g_{N}\right),
\]
the same paper proves a stronger statement: if equality holds, then
\[
\Sigma=\{s=c\}
\]
for some \(c\in (s_0,\overline{s})\). Since Brendle had already observed that constant-mean-curvature hypersurfaces saturate the Heintze–Karcher inequality via a Minkowski identity, this removes assumption (H4) from Brendle’s Alexandrov theorem in warped products [2307.04253].

## 4. Shifted mean curvature and new warped-product inequalities

A major recent development is the introduction of a shifted denominator \(p_1(\kappa)-\varepsilon\) and the matching numerator \(\lambda'(r)-\varepsilon u\) in sub-static warped products
\[
M^{n+1}=[0,\bar r)\times N,\qquad \bar g=dr^2+\lambda(r)^2 g_N,
\]
with conformal vector field \(V=\lambda(r)\partial_r\), support function
\[
u=\langle \lambda\partial_r,\nu\rangle,
\]
and potential \(\lambda'(r)\). The warped product is called static or sub-static when \((M^{n+1},\bar g,\lambda'(r))\) is static or sub-static [2504.15109].

The geometric reason for the shifted numerator is the Hessian identity
\[
\nabla_j\nabla_i\Phi=\lambda' g_{ij}-u h_{ij},
\]
which can be rewritten as
\[
\nabla_j\nabla_i\Phi = (\lambda'-\varepsilon u)g_{ij} -u(h_{ij}-\varepsilon g_{ij}).
\]
This leads to the first shifted Minkowski formula
\[
\int_\Sigma (\lambda'-\varepsilon u)\,d\mu = \int_\Sigma u\,(p_1(\kappa)-\varepsilon)\,d\mu.
\]

Under **static-convexity**
\[
h_{ij}\ge \frac{\bar\nabla_\nu\lambda'}{\lambda'}\, g_{ij},
\]
the general shifted Heintze–Karcher inequality is
\[
\int_\Sigma \frac{\lambda'-\varepsilon u}{p_1(\kappa)-\varepsilon}\,d\mu \ge (n+1)\int_\Omega \lambda'\,dv,
\]
and, in the presence of a horizon boundary \(\partial M=\{0\}\times N\),
\[
\int_\Sigma \frac{\lambda'-\varepsilon u}{p_1(\kappa)-\varepsilon}\,d\mu
\ge
(n+1)\int_\Omega \lambda'\,dv + \lambda(0)^{n+1}\operatorname{Vol}(N,g_N).
\]
If static-convexity is strict at some point, equality implies that \(\Sigma\) is umbilic and has constant mean curvature; in the horizon case, equality forces \(\Sigma=\{r\}\times N\) [2504.15109].

The special hyperbolic case
\[
M=\mathbb H^{n+1},\qquad \lambda'=\cosh r,\qquad \varepsilon=-1
\]
gives a genuinely new non-mean-convex theorem. If \(\Omega\) is a bounded domain with smooth boundary \(\Sigma\) in \(\mathbb H^{n+1}\) and
\[
p_1(\kappa)>-1 \quad \text{on }\Sigma,
\]
then
\[
\int_\Sigma \frac{\lambda'+u}{p_1(\kappa)+1}\,d\mu \ge (n+1)\int_\Omega \lambda'\,dv,
\]
with equality if and only if \(\Sigma\) is umbilic, hence a geodesic sphere. This is stronger than the mean-convex regime \(p_1>0\), and the proof is by a special monotonicity argument along inward unit normal flow rather than by the general integral argument [2504.15109].

## 5. Static realizations, spinorial special cases, and current applications

The sub-static Heintze–Karcher inequality includes several rigid static subclasses. A notable example is the pseudo-hyperbolic setting treated by spinorial methods. If a connected \((n+1)\)-dimensional Riemannian spin manifold carries a nontrivial imaginary Killing spinor \(\psi\), with
\[
V=|\psi|^2,\qquad \nabla_X\psi=\frac{i}{2}\gamma(X)\psi,
\]
then
\[
\operatorname{Hess}V=Vg,\qquad \Delta V=(n+1)V.
\]
For a hypersurface \(\Sigma\) bounding a compact domain \(\Omega\), with inward unit normal \(N\) and \(H>0\), the paper proves
\[
\int_\Sigma \frac{V}{H}\,d\Sigma+\int_\Sigma \langle \nabla V,N\rangle\,d\Sigma \ge 0.
\]
In pseudo-hyperbolic manifolds \(M=\mathbb R\times_{\exp}P\), this becomes
\[
\int_\Sigma \frac{V}{H}\,d\Sigma \ge (n+1)\int_\Omega V\,d\mathrm{vol},
\]
with equality if and only if \(\Sigma\) is totally umbilical. The paper does not use the term “sub-static,” but the identity \(\operatorname{Hess}V=Vg\) places the ambient manifold in the static, hence sub-static, regime [1806.01120].

In three-dimensional asymptotically locally hyperbolic static vacuum geometry, the sub-static Heintze–Karcher inequality is used as a structural input rather than as an endpoint theorem. For an ALH static system \((M^3,g,V)\) with horizon boundary \(\partial M=\bigsqcup_{j=1}^J\partial_jM\), the static equations are
\[
\nabla_g^2 V = V\,\Ric + 3V\,g,\qquad \Delta_g V = 3V.
\]
The paper quotes the multi-horizon sub-static Heintze–Karcher inequality
\[
\int_\Sigma \frac{V}{H}\,d\sigma
\ge \frac{n}{n-1}\int_\Omega V\,d\Omega
+\sum_{j=1}^L c_j \int_{\partial_jM}\frac{\partial V}{\partial \nu}\,d\sigma,
\]
and in the three-dimensional static vacuum case the coefficient simplifies to
\[
c_j=\frac{|\partial_jM|}{3|\partial_jM|+2\pi\chi(\partial_jM)}.
\]
Using these optimal coefficients, the paper defines the IMCF quantity
\[
Q(t)=|\Sigma_t|^{-1/2}
\left(
\int_{\Sigma_t}VH\,d\sigma
-6\int_{\Omega_t}V\,d\Omega
+4\sum_{j=1}^K
\frac{2\pi\chi(\partial_jM)}
{3|\partial_jM|+2\pi\chi(\partial_jM)}
\kappa_j |\partial_jM|
\right),
\]
proves it is monotone nonincreasing, and derives Minkowski inequalities, surface-gravity bounds, a reverse Penrose inequality, and rigidity/uniqueness results for Kottler black holes. The same paper also establishes a weak-regularity version of the Heintze–Karcher inequality for \(C^1\) hypersurfaces with bounded, nonnegative, integrable weak mean curvature [2509.18026].

## 6. Neighboring theories and non-sub-static analogues

Several recent developments are Heintze–Karcher-type inequalities in settings that are structurally related to the sub-static theory but not formulated in terms of a static potential. The free-boundary/capillary paper in the Euclidean half-space proves
\[
(n+1)\big(|\Omega|-|\Lambda|\big)\le\int_{M}\frac{n}{H_M}
\quad\text{in the hydrophobic case},
\qquad
(n+1)\big(|\Omega|+|\Lambda|\big)\le\int_{M}\frac{n}{H_M}
\quad\text{in the hydrophilic case},
\]
with equality exactly for spherical caps. The correction term \(|\Lambda|\) is geometric rather than a static weight, and the proof uses a Montiel–Ros map and a torsion/Reilly framework with substrate terms [2210.16376].

Capillary and anisotropic analogues go further. In the Euclidean unit ball, a Zermelo-navigation Randers metric converts a capillary boundary condition into a free-boundary condition in a Finsler metric, yielding
\[
\int_\Sigma \frac{x_{n+1}+\cos\theta_0\,\langle \nu,E_{n+1}\rangle}{H}\, dA
\ge
\frac{n+1}{n}\int_\Omega x_{n+1}\, dV,
\]
with equality if and only if \(\Sigma\) is a \(\theta_0\)-capillary spherical cap [2401.08450]. In wedges and convex domains, anisotropic free-boundary and capillary inequalities take the forms
\[
\int_\Sigma \frac{F(\nu)}{H^F}\,dA\ge (n+1)|\Omega|
\]
and
\[
\int_\Sigma \frac{F(\nu)-\langle \nu,\mathbf k^F\rangle}{H^F}\,dA \ge (n+1)|\Omega|,
\]
with rigidity by truncated Wulff shapes [2403.19815]. Related sharp anisotropic free-boundary results in convex domains and half-space capillarity likewise replace a static potential by an anisotropic density \(F(\nu)\) and prove Alexandrov-type rigidity via Wulff geometry [2311.01162; 2211.02913].

A different neighboring direction is PDE-based Heintze–Karcher theory from \(p\)-Laplacian overdetermined problems. On complete noncompact manifolds with \(\mathrm{Ric}\ge0\), the identity
\[
\frac{n^2}{(p-1)(n-1)}\int_\Omega L_uP\,dv
+\int_{\partial\Omega}\frac{(1+nH|u_\nu|^{p-2}u_\nu)^2}{H}\,ds
=
\int_{\partial\Omega}\frac{ds}{H}-n|\Omega|
\]
yields
\[
\int_{\partial\Omega}\frac{ds}{H}\ge n|\Omega|,
\]
with equality only for Euclidean balls in \(\mathbb R^n\) [2305.03492]. On compact manifolds with \(\operatorname{Ric}\ge (n-1)K>0\), a new \(P\)-function produces
\[
\int_{\partial M} \frac{1}{nH}\,ds \ge |M|+\frac{2(p-1)}{p}\lambda\int_{M} u^{p-1}\,dv,
\]
which is Heintze–Karcher-type but solution-dependent rather than ambient-potential-driven [2512.19329].

Finally, normal-Jacobian comparison for submanifolds furnishes a Heintze–Karcher-type comparison principle outside the static framework. The comparison theorem
\[
|\det \exp^\perp_{*(x,t\xi)}|
\le
\left(\frac{\mathbf s_\delta(t)}{t}\right)^{m-1}
\Bigl(\mathbf c_\delta(t)-\mathbf s_\delta(t)\langle \underline{\mathbf H}(\underline x),\underline\xi\rangle\Bigr)^n
\]
under \(\Ric_n^M\ge n\delta\) is explicitly described as weakening the lower sectional-curvature assumption in classical Heintze–Karcher comparison, and it leads to sharp Fenchel–Borsuk–Chern–Lashof-type and Willmore–Chen-type inequalities. This suggests a broader tube-volume comparison perspective parallel to, but distinct from, the sub-static weighted-Reilly theory [2605.06074].

The modern sub-static Heintze–Karcher inequality is therefore best understood as a weighted boundary-volume principle governed by the tensor \(\Delta V\,g-\nabla^2V+V\,\mathrm{Ric}\), with a mature theory of horizon terms, equality rigidity, warped-product specializations, and geometric applications. Its neighboring capillary, anisotropic, \(p\)-Laplacian, and normal-Jacobian theories do not replace that framework, but they clarify which parts of Heintze–Karcher theory are specific to static potentials and which belong to a larger family of curvature-reciprocal inequalities.

Source: https://www.emergentmind.com/topics/sub-static-heintze-karcher-inequality