Minimal-Surface and Spinorial Proof Strategies

The Schoen–Yau and Witten proofs establish the same positivity theorem by turning the dominant energy condition into different nonnegative geometric objects. The first route uses minimal hypersurfaces and, for non-time-symmetric data, the Jang deformation. The second uses a Dirac equation and a Weitzenböck identity. Their hypotheses and equality arguments are not interchangeable: dimension controls singularities in the minimal-surface method, while spin structure and spinor boundary conditions control the spinorial method.

Required background. Analysis for Mathematical Relativity supplies elliptic estimates, Hypersurfaces and Foliations supplies minimal and initial hypersurfaces, Curvature supplies Gauss identities, and Einstein Equations supplies the constraints.

Helpful background. Causality and Global Structure clarifies the future-causal ADM vector, while Partial Differential Equations and Green Functions supplies elliptic existence tools.

A common theorem with different engines

Take a complete three-dimensional asymptotically flat initial data set $(\Sigma,h,K)$ with sufficient regularity and decay, no inadmissible inner boundary, and

$$K_{ij}=-h_i{}^\mu h_j{}^\nu\nabla_\mu n_\nu.$$

Assume $\rho\ge|j|_h$. Both methods prove $E_{\mathrm{ADM}}\ge|P|$ and analyze equality. In the time-symmetric case $K=0$, the condition becomes $R(h)\ge0$. The routes diverge in the device that converts this local inequality into a boundary statement at infinity.

For the geometric route, negative mass is assumed and the end is deformed to a convenient asymptotic form without changing its sign. One then constructs a stable area-minimizing hypersurface. Its stability inequality, the Gauss equation, and Gauss–Bonnet force a curvature sign incompatible with the asymptotic geometry. The original three-dimensional proof is organized in Schoen and Yau 1979, §§2–4, pp. 50–70.

For the spinorial route, one solves a first-order elliptic Dirac-type equation for a spinor tending to a constant at infinity. Integration by parts converts the asymptotic spinor boundary term into $E-P_iv^i$ and the interior into nonnegative covariant-derivative and matter terms. This argument is given in Witten 1981, §§2–3, pp. 384–391.

Minimal surfaces and the Jang deformation

In the Riemannian case, stability of a minimal surface $S$ gives, for test function $f$ with compact support (or any smooth $f$ if $S$ is closed),

$$\int_S\left(|\nabla_S f|^2+\frac12R_S f^2\right)dA \ge \frac12\int_S\left(R(h)+|A|^2\right)f^2dA.$$

Here $A$ is the second fundamental form of the two-sided surface, $R_S$ is its intrinsic scalar curvature, and the normal is globally chosen. To see the origin of the inequality, the second variation first gives $\int_S(|\nabla_Sf|^2-(\operatorname{Ric}_h(\nu,\nu)+|A|^2)f^2)dA\ge0$. For a minimal surface the Gauss equation is $R_S=R(h)-2\operatorname{Ric}_h(\nu,\nu)-|A|^2$; eliminating the ambient normal Ricci term gives the displayed formula. Thus the sign is fixed by a second variation, rather than guessed from the desired conclusion.

With $R(h)\ge0$, topology and asymptotics constrain the possible stable surface. The contradiction is geometric and dimension-sensitive: in higher dimensions, minimizing hypersurfaces can acquire singular sets, so the simple smooth argument requires additional regularity theory and dimension bounds.

When $K\ne0$, nonnegative scalar curvature of $h$ no longer follows from the dominant energy condition. The Jang graph in $\Sigma\times\mathbb R$ is chosen so that its scalar curvature satisfies a favorable identity containing $\rho-|j|$ plus divergence terms. Apparent horizons can make the Jang solution blow up; that behavior is useful in boundary variants but requires separate analysis. Thus “apply the time-symmetric proof” is not a valid proof for general data.

Spinors and the dominant energy condition

On a spin initial manifold, solve the Sen–Witten equation $\mathcal D\psi=0$ with a constant asymptotic value. The Weitzenböck identity has a boundary term equal to an ADM contraction and a bulk structure

$$\int_\Sigma \left(|\nabla^{\mathrm{SW}}\psi|^2 +4\pi G\,\langle\psi,(\rho+j_i\Gamma^i)\psi\rangle \right)d\mu_h.$$

Here the Hermitian matrices $\Gamma^i$ obey $\Gamma^i\Gamma^j+\Gamma^j\Gamma^i=2h^{ij}\mathbf1$ in an orthonormal frame. Consequently $(j_i\Gamma^i)^2=|j|_h^2\mathbf1$: the eigenvalues of $\rho+j_i\Gamma^i$ are $\rho\pm|j|_h$. This proves precisely why the matter term is nonnegative when $\rho\ge|j|$. Choosing the asymptotic spinor varies the unit vector in $E-P_iv^i$, producing the full energy–momentum inequality. The spin assumption supplies the bundle and global Dirac operator; elliptic existence supplies a spinor with the required asymptotics.

The method is compact, but not hypothesis-free. Inner boundaries require a compatible chirality or horizon boundary condition. Zeros and regularity of the spinor must be handled in rigidity. In dimensions where the manifold is not automatically spin, Witten’s route does not cover the same class as a minimal-hypersurface theorem.

Rigidity enters differently

In the geometric route, equality forces the stable-surface and deformation inequalities to saturate; one traces that saturation back to Euclidean geometry or a Minkowski embedding. This requires ruling out hidden singular or boundary contributions. In the spinorial route, zero boundary energy forces the nonnegative bulk integral to vanish for every asymptotic spinor, since $E=0$ also implies $P=0$. A basis of asymptotic spinors yields a full parallel family; its integrability and the global equality argument give the Minkowskian data. One parallel spinor alone does not establish flatness of a Lorentzian geometry.

Neither equality argument is simply continuity from positive mass. A manifold with a singular interior is a counterexample to omitting completeness from either route. A nonspin higher-dimensional manifold is a counterexample to claiming that the spinorial proof applies universally, while a dimension with singular minimizing hypersurfaces blocks the naive smooth geometric proof.

Comparison and scope

The minimal-surface route exposes topology, stable hypersurfaces, and the geometry later reused in Penrose inequalities. The spinorial route exposes the Lorentzian energy–momentum vector and generalizes elegantly when spin structure is available. Both need asymptotic control and a global elliptic solution; both use the dominant energy condition; both require a separate rigidity analysis.

This comparison does not claim that the original papers alone contain every modern dimensional extension. It states the three-dimensional asymptotically flat theorem and the mechanisms that survive in later variants. Boundaries, low regularity, extra dimensions, and modified matter assumptions must be matched to a theorem that explicitly handles them.

There is also a pedagogical difference. The geometric proof makes a negative mass assumption visible by producing an impossible hypersurface; the spinor proof makes positivity visible directly as a sum of squares. That does not make one argument more physical or more general. Their independent agreement is valuable precisely because normalization, energy sign, and the Minkowski equality case can be checked through two mathematically different mechanisms.

Check the hypotheses

Why can the time-symmetric Schoen–Yau argument start from $R(h)\ge0$, while a general initial data set satisfying $\rho\ge|j|$ needs an additional deformation or a spacetime-spinor connection?

Solution

For $K=0$, the Hamiltonian constraint is $R(h)=16\pi G\rho$, so the energy condition directly gives nonnegative scalar curvature. For $K\ne0$, $R(h)=16\pi G\rho-K^2+|K|^2$ has no definite sign. The Jang deformation repackages the constraint into a favorable scalar-curvature identity, while the Sen–Witten connection inserts both $\rho$ and $j$ directly into its Weitzenböck formula. Omitting that step would use an inequality the constraints do not provide.

References

• Schoen, Richard, and Shing-Tung Yau. “On the Proof of the Positive Mass Conjecture in General Relativity.” Communications in Mathematical Physics 65 (1979): 45–76. DOI.
• Witten, Edward. “A New Proof of the Positive Energy Theorem.” Communications in Mathematical Physics 80 (1981): 381–402. DOI.