Does Pure Spinor Supersymmetry Forbid Runaway Curvature Couplings?

Gravity theories that add extra couplings between matter and curvature, or extra derivatives in the action, often hide modes that grow without bound. Two versions of this problem are well documented. A field can sit on a hilltop of its effective potential, which is a tachyonic instability: small perturbations grow exponentially. Alternatively, an action with more than two time derivatives propagates an extra mode with negative energy, which is the Ostrogradsky ghost. Recent work by Hell and collaborators shows how fragile non-supersymmetric models are on both fronts. Couplings of a massive two-form to the Riemann tensor cause runaway modes and drive gravitational waves into strong coupling far below the Planck scale, and no local field redefinition is known to remove them.

We ask whether supersymmetry, in its most rigid covariant form, already rules these couplings out. The pure spinor formalism describes ten-dimensional super-Yang–Mills and eleven-dimensional supergravity with a single operator $Q=\lambda^\alpha D_\alpha$. Here $\lambda^\alpha$ is a constrained bosonic spinor and $D_\alpha$ is the superspace derivative. Physical fields are the solutions of $Q\Psi=0$ modulo $Q$-exact shifts, and $Q^2=0$ holds off-shell. The proposal turns that one condition into a stability filter. We list every curvature coupling and higher-derivative correction that is consistent with $Q$. We then check whether any of the survivors produces a tachyonic, runaway, or ghost mode, both on exact curved supergravity backgrounds and in amplitudes computed with the first-quantized pure spinor worldline. The worldline is a path integral over a single particle trajectory.

Core Claim

Our hypothesis is that, at four derivatives, eleven-dimensional pure spinor BRST cohomology admits no deformation of the quadratic fluctuation action that couples the three-form field strength $G=dC$ to the Riemann tensor, and no curvature-squared term, beyond what field redefinitions can remove. ("BRST cohomology" here simply means the set of $Q$-closed, non-exact objects.) If that holds, the Riemann-type couplings Hell and Obata found fatal (On the Kalb-Ramond field with non-minimal coupling to gravity) and four-derivative Ostrogradsky poles are both excluded by the same algebraic condition. The first allowed corrections would then appear only at higher derivative order, where they are suppressed and must be treated as effective-field-theory terms.

The ten-dimensional super-Yang–Mills analogue is a calibration case. There, the method should find that the lowest field-strength correction allowed by supersymmetry is quartic, with no cubic term. That is the structure expected from the zero-slope limit of superstring amplitudes, which The pure spinor superparticle and 10D super-Yang-Mills amplitudes already matches at lowest order.

The second claim is specific to the worldline. In that formalism the propagator is $b/H$, where the composite $b$-ghost satisfies $\{Q,b\}=H=P^2/2$. Any change to $H$ breaks the decoupling of $Q$-exact states unless matching contact vertices are added. We will test whether every compensating package that restores decoupling is equivalent to a field redefinition, which introduces no new pole, or whether some packages carry a genuine extra pole with a wrong-sign residue. This gives a tree-level, diagram-free criterion for perturbative ghost-freedom.

Research Program

The first stage is a classification of allowed deformations. We start from the ghost-number-three superfield of eleven-dimensional supergravity. Its cohomology reproduces linearized supergravity (Equivalence of the 11D pure spinor and Brink-Schwarz-like superparticle cohomologies). We then enumerate candidate deformations of Cederwall's quadratic and cubic pure spinor actions, organized by derivative order.

We use the decomposition of Berkovits and Guillen (Equations of motion from Cederwall's pure spinor superspace actions) to separate the dependence on the extra "non-minimal" variables from the physical fields. That paper stops at leading order in the coupling, and its authors flag pole restrictions on $(\bar\lambda\lambda)^{-p}$ as the obstruction at higher orders. Our classification needs only the first subleading derivative order, so it sits just beyond their result rather than requiring the all-orders problem to be solved. We will also use the simplified eleven-dimensional $b$-ghost (Taming the 11D pure spinor b-ghost) to build the ghost-number-two and ghost-number-three vertices needed for the next stages.

The output of this stage is a list of tensor structures. For each one we record its coefficient as allowed or forbidden, mapped onto the Hell–Obata coupling labels: $\alpha$ for Ricci scalar couplings, $\beta,\sigma$ for Ricci tensor couplings, and $\gamma,\lambda$ for Riemann tensor couplings. We obtain this map by reducing on a circle, where $C_{\mu\nu 10}$ becomes the ten-dimensional Kalb–Ramond two-form.

The second stage tests stability on curved backgrounds. BRST nilpotency only makes sense on backgrounds that solve supergravity, and eleven-dimensional supergravity has no de Sitter vacuum. We therefore do not use de Sitter. We use two classes of exact solutions instead.

The first class is Ricci-flat plane waves. These have nonzero Riemann tensor but vanishing Ricci tensor, so they isolate exactly the $\gamma,\lambda$ couplings that local disformal transformations cannot cure. A disformal transformation is a field-dependent rescaling of the metric along a preferred direction. For plane waves, a single background-graviton insertion on the worldline is not an approximation, because these backgrounds are linear in a suitable gauge. The second class is AdS$_4\times S^7$. There, Ricci-tensor couplings reduce to mass shifts, so the relevant test is the Breitenlohner–Freedman bound: in anti-de Sitter space a mode may have negative mass-squared and still be stable, provided it stays above this threshold. Goon, Hinterbichler and Trodden (Symmetries for Galileons and DBI scalars on curved space) found that bosonic brane theories on curved space trade shift symmetry against stability around thresholds of this kind.

On both classes of background we derive the dispersion relation $\omega^2(k)$ for every surviving deformation. We look for tachyonic signs at small $k$ and gradient instabilities at large $k$, and we track how many degrees of freedom propagate.

The third stage covers amplitudes and massive forms. Eleven-dimensional massless amplitudes whose momenta have a component along a compact circle are ten-dimensional amplitudes of massive Kaluza–Klein modes. The Kaluza–Klein tower therefore contains massive forms coupled to gravity with no free couplings. That is a concrete realization of the massive Kalb–Ramond setting.

We compute two-graviton, two-massive-form amplitudes from the four-point worldline prescription (Tree-level 11D supergravity amplitudes from the pure spinor worldline) and fit their high-energy growth to the Hell–Obata effective field theory. The question is whether the strong-coupling scale behaves like $(\beta/M_{\rm pl}^2m^2)^{1/3}$, as found for the Proca field in Unveiling the Inconsistency of the Proca Theory with Nonminimal Coupling to Gravity, or whether the tower cancels the growth. If it cancels, we identify which effective couplings remain after the heavier modes are integrated out, and whether they fall only in the benign $\alpha$ class. Running the worldline Ward-identity test from the Core Claim on these same amplitudes covers the Ostrogradsky side.

Baselines

The naive reference point is the generic non-supersymmetric effective theory. All $\alpha,\beta,\sigma,\gamma,\lambda$ couplings are free there, and Hell and Obata show that the Ricci tensor and Riemann tensor classes produce runaway vector modes and tensor strong coupling. The proposal beats this baseline only if it proves, rather than assumes, that supersymmetry sets the $\gamma,\lambda$ coefficients to zero or to a stable combination.

The strongest existing method for curing such couplings is the bosonic disformal field redefinition, $g_{\mu\nu}\to g_{\mu\nu}-\beta A_\mu A_\nu/M_{\rm pl}^2$, from Hell's Proca paper. It removes Ricci tensor pathologies but not Riemann ones. We will run it on exactly the reduced actions produced in the first stage, with identical background solutions and perturbation variables. Beating it means stabilizing a case it cannot: a Riemann-coupled two-form on a Ricci-flat plane wave.

For the amplitude stage, the strongest competitor is the perturbiner computation that the eleven-dimensional worldline paper already matches at four points. We will compute the same Kaluza–Klein kinematics both ways, and agreement is required, not optional. The worldline's advantage must show up elsewhere, in the BRST Ward-identity test and in five-point checks where it is competitive. On Lacuna we found no numerical benchmark for any of these baselines, only analytic scaling statements, so success is judged by exact agreement and by the exponents of energy growth.

Two ablations isolate the key ingredient. The first is a bosonic truncation: we reduce only the bosonic two-derivative eleven-dimensional action. At tree level with only bosonic external states, this must give the same Kaluza–Klein amplitudes. Any Kaluza–Klein cure that it also exhibits is therefore credited to higher-dimensional locality, not to supersymmetry. The second replaces the BRST filter of the first stage with the bare requirement of gauge invariance. This shows which forbidden couplings are excluded by supersymmetry specifically.

Why not just…?

Why not just use Feynman diagrams in curved space? For a single coupling on a single background, they suffice, and we use them as the check in the second stage. They cannot answer the classification question, because they take the action as given. They also become unwieldy for multi-graviton vertices in eleven dimensions, and perturbation theory breaks down at exactly the strong-coupling scales under study.

Why not just apply disformal transformations and the brane/Lovelock constructions? These are the right tool where they work. The brane/Lovelock constructions derive ghost-free four-dimensional theories from higher-dimensional curvature combinations whose field equations stay second order (New Class of Effective Field Theories from Embedded Branes). But Hell and Obata explicitly report failure for Riemann couplings. Separately, Hell, Lüst and Zoupanos (On the ghost problem of conformal gravity) find that boundary-condition prescriptions do not remove the ghost of conformal gravity. The plane-wave test is designed to sit exactly where these tools stop.

Why not just reduce bosonic eleven-dimensional supergravity, or bootstrap on-shell three-point amplitudes, without pure spinors? For the Kaluza–Klein amplitude question this is a real threat. At four points the bosonic reduction is equivalent, which is why it appears as an ablation rather than being argued away. Flat-space on-shell bootstraps can likewise classify three-point structures. Neither approach is off-shell, and neither works on curved backgrounds. Only the curvature-coupled dispersion relations of the second stage, and the off-shell BRST classification of the first stage, require pure spinors. If those two stages add nothing beyond the bosonic reduction and the flat-space bootstrap, the honest conclusion is that pure spinors are unnecessary for this question.

Decisive Checks

Several checks validate the machinery before any new claim is made. The method must reproduce the linearized eleven-dimensional supergravity field content. It must recover, from the Kaluza–Klein tower, the degree-of-freedom counts Hell reports: five for Proca coupled to gravity in Minkowski space, and three for a massive Kalb–Ramond field. It must agree with the perturbiner four-point amplitude. And in ten-dimensional super-Yang–Mills it must rule out cubic field-strength corrections.

The central claim then rests on two checks, each with a clear failure mode. In the first stage, the claim is falsified if a BRST-consistent four-derivative term of the form $RGG$, or a curvature-squared term, survives that no field redefinition removes. In that case its plane-wave dispersion relation becomes the main result, stable or not. In the third stage, Hell-type strong coupling at the scaling $(\beta/M_{\rm pl}^2m^2)^{1/3}$ in Kaluza–Klein amplitudes would show that the tower does not cure the problem. Conversely, growth no faster than $E^2/M_{\rm pl}^2$, where $E$ is the scattering energy, together with induced couplings confined to the $\alpha$ class, would show that supersymmetric higher-dimensional origin selects exactly the stable corner of the Hell–Obata parameter space.

All computations are symbolic and run in standard computer algebra. The physics datasets and question-answering resources listed on Lacuna, and the BES integral-equation benchmark, have no bearing on this question and are not used.