An arXiv preprint establishes that the Cayley 4-form is not the only 4-form that can define a non-degenerate Riemannian metric through a formula associated with Karigiannis. The result broadens the range of algebraic objects that can play that role, but it does not identify every such form or settle the larger conjecture examined in the paper.
The work is a formal study of 4-forms on an 8-dimensional real vector space. In this setting, a 4-form is an algebraic object evaluated on four vectors, and the paper investigates what geometric structures can be recovered from it. There are no participants, measurements or empirical dataset behind the result.
A family that separates the possibilities
The clearest evidence for the non-uniqueness comes from an explicit one-parameter family called βt. The family is stabilized by SU(4), a symmetry group used to describe the construction. Within the family, the parameter value t = 1 meets the Spin(7) orbit associated with the Cayley case, while t = 0 has stabilizer Sp(4, R) and is not metric.
The same family supplies other parameter values that define non-degenerate Riemannian metrics. The examples therefore place the Cayley form alongside other metric 4-forms within a structured algebraic family, while also including a parameter value that does not define a metric.
The result is an existence statement, not a catalogue. The preprint gives an explicit family supporting metric non-uniqueness, but it does not classify all 4-forms that define metrics or state a complete general condition for metricity.
Two tests for non-degeneracy are not the same
A central part of the note concerns how to decide whether a 4-form is non-degenerate. The paper uses a triple-based definition: for every linearly independent triple of vectors u, v and w, there must be some vector x for which α(u, v, w, x) is not zero. In ordinary language, no independent three-direction slice is allowed to make the form vanish against every possible fourth direction.
The paper proves that this definition implies multisymplectic non-degeneracy, another condition used for differential forms. But the implication goes only one way in the examples. A form called ξ is multisymplectically non-degenerate while still being degenerate under the triple-based definition, so the two tests cannot simply be treated as interchangeable.
The note also gives equivalent ways to check the triple-based condition. For any independent pair of vectors, contracting the 4-form with that pair must produce a symplectic form on the quotient by the span of those two vectors. The equivalent test can also be written as a contraction-and-wedge expression that must be nonzero.
The tensor machinery behind the argument
To connect the form with metric questions, the paper introduces an area metric: a symmetric bilinear map whose inputs are pairs of bivectors. The construction produces an area metric Gα from the 4-form α.
The resulting map has a striking structural property: Gα is always indefinite. Its non-degeneracy is characterized by a determinant condition written in terms of the quasi-Hodge dual of α, the companion object used in this calculation. The condition is that det(ρ*(α)) must not vanish.
A further theorem decomposes Gα into a normalized term built from α minus a second tensor, Hα. This decomposition forms part of the tensor framework used to compare the area metric with the stronger non-degeneracy condition.
The paper reports that calculations involving q were obtained with Mathematica. The study otherwise works with formal definitions, algebraic identities, tensor manipulations, lemmas, theorems and proofs rather than statistical estimation.
The main conjecture remains open
The question left hanging is the Salamon–Walpuski conjecture considered in the note: whether α ∧ α = 0 implies that α is degenerate. The self-wedge α ∧ α combines the 4-form with itself; the conjecture proposes that a vanishing result signals degeneracy under the paper’s definition.
The paper distinguishes ordinary non-degeneracy from a stronger condition built from the area-metric framework. It proves that strong non-degeneracy implies ordinary non-degeneracy and also implies α ∧ α is not zero. At the same time, strong non-degeneracy does not imply that the 4-form is metric, even though metric 4-forms require a nonzero self-wedge.
What the paper does not show is precisely the step needed to close the argument: strong non-degeneracy is not proved equivalent to ordinary non-degeneracy. The note reports no counterexample and presents that equivalence as a conjecture, so the Salamon–Walpuski question remains unsettled within this work.
That leaves related questions open, including whether metricity implies strong non-degeneracy and what the precise general condition for a 4-form to be metric should be. The displayed examples clarify the landscape, but they do not resolve those broader relationships.
A formal result with a narrow scope
The document is an arXiv preprint, version 1, listed in mathematical differential geometry as arXiv:2608.20200v1 and dated 20 August 2026.
Its conclusions apply to formal constructions on an 8-dimensional real vector space. They do not provide empirical participants or a dataset, and the paper does not prove the Salamon–Walpuski conjecture for all 4-forms or give a complete classification of metric 4-forms.
The acknowledgements state that S.C. is supported by the Engineering and Physical Sciences Research Council through grant EP/W524402/1. The preprint also reports that Mathematica was used for calculations involving q.
Paper data and sources
Original title: A note on 4-forms in 8-dimensions
Authors: Sam Close
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text