On a compact boundaryless Riemannian spin manifold with , integrating the Lichnerowicz spinor-square formula against a harmonic spinor gives . Positivity forces the spinor to be parallel. Its norm is constant on each connected component, so a nontrivial solution is nowhere zero there. Strictly positive scalar curvature at any point on such a component excludes nonzero harmonic spinors.
An orthonormal tetrad, or vierbein, is a local basis of tangent vectors together with inverse coframe , satisfying and . Capital indices label the locally orthonormal Lorentz frame; lower-case indices label spacetime coordinates. Changing the frame by a local Lorentz transformation leaves the metric unchanged. A spinor description uses a compatible local spin frame and the torsion-free spin connection.
Let fixed flat matrices satisfy . The curved spacetime gamma matrices are and . Therefore
The identity matrix is four by four for a four-dimensional Dirac spinor. The combined tensor/spin covariant derivative makes , the Clifford version of the tetrad postulate.
Fix the curvature slots by , with , and . The torsion-free spin connection then has . These choices fix the signs in the requested identities.
Write and similarly for four indices. Clifford anticommutation gives
On contraction with the Riemann curvature tensor, the four-form term vanishes by the first Bianchi identity. The first and last two-form terms vanish by pair antisymmetry; the remaining two-form terms are symmetric Ricci contractions against antisymmetric gamma matrices and vanish too. The scalar terms are zero, minus R and minus R respectively. Thus the Riemann tensor Clifford contraction is
This algebraic result is independent of Lorentz signature when the contraction convention is held fixed.
Apply the Dirac operator twice, using covariant constancy of the gamma matrices. Define the rough wave operator by , with the full connection acting on the intermediate covector-spinor derivative. Splitting the product into symmetric and antisymmetric parts gives
Consequently the Lichnerowicz spinor-square formula and the assumed massless Dirac equation imply . Torsion would modify the commutator and introduce extra terms, so the torsion-free connection is part of this argument.
The chirality matrix is covariantly constant under the metric-preserving spin connection and anticommutes with every . Hence
Thus multiplication by reverses the sign of the mass parameter. At it maps solutions into solutions of the same Dirac equation.
Use the spinor curvature identity
Here the rough Laplacian includes the affine connection on the index of the first derivative. Clifford expansion gives
In the last contraction, the four-gamma part vanishes by the first Bianchi identity, the two-gamma parts vanish by symmetry of the Ricci tensor, and the scalar contractions give . Therefore the Lichnerowicz spinor-square formula is
Multiplying the first-order equation by now proves
The explicit curvature coefficient is fixed by the minimal spinor coupling; it is not an arbitrary scalar-field curvature coupling.
For the general metric-preserving connection, define and let be its curvature. Distinguish the spin connection acting on the spinor from the full second derivative acting also on its derivative index. Their antisymmetric relation is
Thus the torsion correction to the Dirac square gives the exact squared field equation
General torsion removes the curvature symmetries used in the preceding simplification. The curvature contraction may now contain nonscalar Clifford terms, in addition to the displayed first-order torsion term; simply replacing by a torsionful scalar in the torsion-free answer is not valid.
Equivalently, write the connection as Levi-Civita plus contorsion tensor . Then , where , and the square contains . This exhibits the extra couplings through derivatives of contorsion, first-order spinor derivatives and quadratic contorsion. Metric compatibility still preserves , so the mass-sign result remains valid. Spin therefore couples directly to curvature through its connection and the universal term, and to torsion through additional spin-dependent terms. The scalar reduction in the torsion-free square does not mean the full spinor transport is insensitive to the remaining Riemann curvature.
Choose the Riemann curvature tensor convention
with positive scalar curvature for a round sphere. The connection preserves the gamma matrices. If , its square is
Here is the rough Laplacian, including the connection on the derivative index. In the product of two antisymmetric gamma pairs, the fully antisymmetric four-gamma term vanishes by the first Bianchi identity. The two-gamma contractions vanish because the Ricci tensor is symmetric, while the scalar contractions give . Thus the Lichnerowicz spinor-square formula is
Multiplying by , with constant , proves
In Riemannian signature in this positive-Clifford convention is formally anti-self-adjoint; the more usual self-adjoint Dirac operator has square . Stating this convention avoids an apparent sign conflict between the two forms of the formula.
Now let on the compact Riemannian spin manifold. Use the positive-definite spinor inner product and integrate the squared equation. Integration by parts has no boundary contribution, giving
Both terms are nonnegative. Smoothness then forces pointwise. Consequently every harmonic spinor is parallel, and . A nonzero parallel spinor has constant positive norm on its connected component, so its scalar curvature vanishes there. This is compact harmonic-spinor rigidity under nonnegative scalar curvature.
The full Ricci conclusion follows from integrability, not merely from vanishing scalar curvature. Since a parallel spinor has zero curvature for its spin connection, contract its Ricci identity with :
The three-gamma term vanishes by the Bianchi identity; the remaining contractions give the displayed Ricci term. For each fixed , put . Clifford multiplication twice gives
Positive definiteness and imply . Thus the Riemannian parallel-spinor Ricci-flatness result is
On a connected manifold a nontrivial parallel solution is nowhere zero, so this holds everywhere. On a disconnected manifold it holds on every component supporting a nonzero spinor; the stated nowhere-vanishing hypothesis ensures all components are covered. The positive-definite metric tensor is essential to the last inference.