Physicist Challenges Claim That Cosmic Walls Could Pin Down Torsion’s Hidden Strength

A quiet but pointed dispute has broken out in the pages of The European Physical Journal C over one of the most elusive quantities in modern gravitational physics: the coupling between spinning fermions and spacetime torsion. In a comment published on 7 October 2026, Luis B. Castro of the Federal University of Maranhão in Brazil argues that a recent paper claiming to derive a numerical bound on the axial torsion coupling is undermined by algebraic and dimensional errors that cascade through every major result. The target of the critique, a study of fermion condensates trapped between double domain walls in Einstein–Cartan and teleparallel gravity, had suggested that the coupling could be pinned down to roughly one ten-thousandth. Castro’s recalculation says that number was never actually established.

To understand what is at stake, it helps to recall what torsion is and why anyone cares. In Einstein’s general relativity, spacetime geometry is described entirely by curvature, and the connection that tells parallel vectors how to move is symmetric. Einstein–Cartan theory relaxes that assumption, allowing spacetime to carry torsion, a twisting of the geometry sourced by the intrinsic spin of matter. Because ordinary matter carries very little net spin density, torsion is expected to be fantastically weak, and no experiment has yet measured its coupling to fermions directly. Any theoretical scheme that promises a concrete numerical bound therefore attracts attention, which is precisely why the original paper, authored by L. C. Garcia de Andrade and published in July 2026, seemed exciting.

The original argument ran roughly as follows. Consider a pair of domain walls, thin sheets of concentrated energy, in a spacetime whose geometry is described by a tetrad, the set of local orthonormal frames that generalizes the metric. In the teleparallel formulation, the spin connection is set to zero, so the torsion two-forms are simply the exterior derivatives of the tetrads. Garcia de Andrade computed these torsion components, derived equations for a wall-separation function h(t,z), solved them, and then studied Dirac fermions living on the walls. The effective interaction entering the Dirac equation combined the electromagnetic coupling eB with the torsion coupling g_T multiplied by an axial torsion field S. Setting that combination to zero in a thin-wall limit, the paper extracted g_T as a ratio of the magnetic field to the torsion field, arriving at a bound near ten to the minus four.

Castro’s comment dismantles this chain at its first link. Recomputing the torsion two-forms directly from the tetrad ansatz with vanishing spin connection, he obtains expressions that differ from the published ones by factors of two and by misplaced coefficients. For example, where the original paper reports a coefficient of minus four pi sigma times z h times the time derivative of h in one torsion component, the direct calculation gives minus eight pi sigma times the same combination. Since every downstream equation for the wall dynamics is built on these two-forms, the discrepancies are not cosmetic. They propagate into the differential equations governing h(t,z) and ultimately into the claimed wave equation and its effective propagation velocity, which had been interpreted as a signature of gravitational radiation from the walls.

A second, independent problem involves a subtle but crucial distinction between coordinate components and tetrad components. The original paper writes the torsion components in a form that requires reading the coefficients in the tetrad basis rather than directly in the coordinate basis of time and spatial differentials. When Castro performs the conversion carefully, the equation for the wall-separation function integrates to a logarithmic expression, h squared equals z squared plus an inverse constant times the logarithm of a linear function of z squared and t. This is not the polynomial solution printed in the original paper. Even taking the original’s own scalar equations at face value, Castro shows that their exact integration yields a logarithm, not the cubic polynomial reported, unless an unstated perturbative expansion is smuggled in. The small-h approximation invoked in the original work, he demonstrates, does not rescue the polynomial form either; it merely imposes extra consistency conditions on integration constants.

The consequences ripple outward. The wave equation for the wall separation, and with it the interpretation of an effective velocity and the possibility of gravitational waves emitted by the coalescing walls, does not follow from the corrected calculation. Castro is careful about the limits of his claim: he does not prove that no gravitational radiation or propagating torsional mode can arise in some corrected model. He shows only that the specific derivation in the original paper is invalid as written. He also notes that even within the original’s own reduced flatness constraint, the solution h squared equals two At plus B means the dramatic coalescence of the walls at time zero occurs only for a special choice of the integration constant B, not as a forced consequence of the equations.

The fermionic sector fares no better in Castro’s assessment. Solving the zero-mode Dirac equations with the effective combination A equal to eB plus g_T times S, he finds Gaussian solutions whose signs differ from those published. For positive A, one chiral component is Gaussian-damped and normalizable while the other grows without bound; for negative A, the roles reverse. This is the familiar chiral selection pattern of localized zero modes, and it matters enormously for the condensate. Castro further shows that the scalar Dirac bilinear used to construct the condensate was computed incorrectly: with the beta matrix as defined in the original paper, the bilinear mixes left- and right-handed components and, in any normalizable chiral sector, actually vanishes, even though the positive probability density remains localized. The original analysis, he argues, conflated a scalar condensate with a positive density, two mathematically distinct objects.

Then comes the dimensional problem, which may be the most damning point of all. The original bound rested on identifying the combination eB plus g_T S with a wall thickness or separation, a quantity with dimensions of length. But in natural units a magnetic field has dimensions of mass squared, while an axial torsion vector carries dimensions of mass, so the sum cannot equal a length without additional dimensional parameters. In Castro’s corrected reading, the combination A instead fixes a localization length, roughly one over the square root of the absolute value of A. A thin, tightly localized fermion mode corresponds to a large A, not to A equal to zero. Imposing the cancellation condition, as the original paper did, actually makes the localization length diverge, the opposite of a thin wall. Estimating g_T therefore requires specifying an independent physical scale, and the magnetic field and torsion field alone cannot determine the coupling.

The numbers illustrate how far the original estimate strays. Using the original paper’s own magnetic field of ten to the minus nine gauss and torsion scale of ten to the minus thirty gigaelectronvolts, Castro finds that the cancellation condition yields g_T of order minus six gigaelectronvolts, not ten to the minus four. With a localization length of one centimeter, the corrected formula gives a coupling of order hundreds of gigaelectronvolts; with one meter, the result stays near the cancellation value. The magnetic length associated with the field is about eight centimeters. No unique number emerges without an independently specified scale, which is exactly the kind of hidden assumption that can make a striking bound evaporate under scrutiny.

Castro closes with a measured summary of eight points and a constructive caveat: none of this rules out the possibility that a corrected model of magnetized, torsion-carrying double walls could support localized fermion modes or eventually yield genuine phenomenological constraints on axial torsion. But the specific conclusions and the numerical bound in the original paper, he concludes, require a full rederivation from the ground up. For a field where torsion remains one of gravity’s best-hidden secrets, the episode is a reminder that the path to measuring spacetime’s twist runs through careful bookkeeping as much as through bold ideas, and that in theoretical physics, the factors of two are never just details.

Subject of Research: Corrections to the derivation of axial torsion coupling bounds from fermion condensates in Einstein–Cartan double domain walls

Article Title: Comment on “Einstein–Cartan fermion condensates trapped in double walls induce axial torsion coupling bounds [Eur. Phys. J. C (2026) 86: 785]”

Article References: Castro, L. B. (2026). Comment on “Einstein–Cartan fermion condensates trapped in double walls induce axial torsion coupling bounds [Eur. Phys. J. C (2026) 86: 785]”. The European Physical Journal C, 86(10), Article 1148. https://doi.org/10.1140/epjc/s10052-026-16388-3

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16388-3

Keywords: Einstein–Cartan gravity, spacetime torsion, teleparallel gravity, fermion condensates, domain walls, Dirac zero modes, axial torsion coupling, gravitational waves, tetrad formalism, dimensional analysis, theoretical particle physics, general relativity