This paper shows that a natural notion of total energy for spacetime initial data is positive when the space dimension n is at least four. The main technical conclusion is that a constant α that appears in the metric far away from the matter must be positive. In the language of the field, α is the coefficient in the usual asymptotic expansion of the metric at infinity, so proving α>0 is the spacetime positive energy theorem for these data.

The authors work with an initial data set (M,g,q). Here g is a Riemannian metric on an n‑dimensional manifold M (n≥4) and q is a symmetric tensor that encodes the extrinsic curvature of M in spacetime. They define a scalar µ from the scalar curvature of g and q, and a one‑form J from derivatives of q. The hypotheses include that M is asymptotically flat (outside a compact set it looks like the complement of a ball in Euclidean space) with precise decay rates for g and q, and a pointwise inequality µ−|J|_g>0 everywhere. They also assume a stronger inequality near infinity of the form µ−|J|_g ≥ κ r^{3−n}|q|_g for some κ>0. Under these assumptions the theorem asserts α>0.

To prove this, the authors use a regularized version of the Jang equation. The Jang equation is a geometric partial differential equation introduced by P. S. Jang and used by Schoen and Yau to turn a spacetime positivity problem into a Riemannian one. The regularization they use adds a small “capillary” term to control the behavior of solutions. They build a global solution u of that modified Jang equation, obtain decay and derivative estimates for u at infinity, and then form a new metric tg = g + du⊗du and a one‑form Ξ built from u and q.

A key analytic identity, originally from Schoen and Yau, relates the scalar curvature of the new metric tg and the one‑form Ξ to the original energy quantities µ and J and to the Jang equation itself. By combining that identity with the assumed positivity of µ−|J|_g, and by using a shielding principle from Lesourd–Unger–Yau to localize the argument, the authors obtain an integral inequality. They then argue by contradiction: assuming α≤0 they use these constructions and a dimension‑descent argument from their earlier Riemannian positive mass result to manufacture an (n−1)‑dimensional dataset with zero mass. That contradicts their Riemannian positive mass theorem, so α must be positive.