Moonshine Uniqueness Conjecture (Strongly Rational Holomorphic VOA Form)

Moonshine Uniqueness Conjecture (Strongly Rational Holomorphic VOA Form)

Definitions and mathematical statement

A vertex operator algebra has a graded vector space, vacuum and conformal vectors, and vertex operators $Y(a,z)=\sum_{m\in{\mathbb{Z}}}a_mz^{-m-1}$ satisfying the usual vacuum, translation and Jacobi axioms. Here CFT type means $V=\bigoplus_{n\ge0}V_n$ and $V_0={\mathbb{C}}\mathbf1$. The subspace $C_2(V)$ is spanned by $a_{-2}b$; $C_2$-cofiniteness means $V/C_2(V)$ is finite-dimensional. Strong rationality also requires simplicity, self-contragredience and complete reducibility of admissible modules. Holomorphic means $V$ is its only irreducible module. The target is \[ c(V)=24,\quad V_1=0\quad\Longrightarrow\quad V\cong V^{\natural} \] for every VOA satisfying those hypotheses, where $V^{\natural}$ is the standard moonshine VOA. Matching its character alone is not the requested isomorphism.

Formulation and concept references: [S1].

Short English statement

Is the moonshine vertex operator algebra uniquely determined by strong rationality, holomorphicity, central charge twenty-four, and the absence of weight-one states?

Research attempt: character rigidity versus product rigidity

GPT-6 Astra Ultra, 27 September 2026. Status: UNRESOLVED.

There are two distinctly different reconstruction questions. The hypotheses force the graded dimensions, but the conclusion identifies every vertex product. This attempt derives the former, extracts the algebraic structure in weight two, and formulates a concrete sufficient extra condition using a known uniqueness theorem. It does not prove that every VOA in the question satisfies that extra condition.

The character is forced.

Let \[ Z_V(\tau)=\operatorname{Tr}_V q^{L_0-1} =q^{-1}+\dim(V_1)+\dim(V_2)q+\cdots, \qquad q=e^{2\pi i\tau}. \] The modular-invariance theorem for rational $C_2$-cofinite VOAs supplies the modular transformation law for irreducible characters [E1]. Here there is only one irreducible module. The $T$ transformation has multiplier $e^{-2\pi i c/24}=1$. The $S$ transformation is scalar; evaluation at $\tau=i$, where the convergent character is positive, makes that scalar $1$. Hence $Z_V$ is an invariant modular function, holomorphic on the upper half-plane, with its only possible pole at the cusp and with principal part $q^{-1}$.

Let $j=q^{-1}+744+196884q+\cdots$. The difference $Z_V-j$ extends holomorphically to the compact modular curve, so is constant. Since $V_1=0$, comparison of constant terms gives \[ Z_V=j-744=q^{-1}+196884q+21493760q^2+\cdots. \tag{1} \] Thus $\dim V_2=196884$ and $\dim V_3=21493760$, and all further dimensions agree with moonshine. This uses a stated modularity theorem, not a derivation of modularity from the formal power series alone. Without the analytic transformation law, arbitrary nonnegative coefficient series with the same first terms would remain possible.

Equation $1$ is an exact result, but it only counts vectors. It contains no structure constants for $a_m b$. In particular, recognizing the coefficient $196884$ does not identify the algebra structure on a vector space of that dimension.

What the weight-two operations really give.

On $V_2$ define $a\cdot b=a_1b$. The grading rule puts this in $V_2$. Skew symmetry of vertex operators implies commutativity here. More explicitly, in the coefficient comparison the first correction involves $L(-1)b_2a$, which vanishes because $b_2a\in V_1=0$. The next possible correction is a multiple of $L(-1)^2b_3a$, which vanishes because $b_3a\in V_0=\mathbb C\mathbf1$ and $L(-1)\mathbf1=0$. Terms with higher modes have negative weight and vanish. Thus $a_1b=b_1a$.

The conformal vector obeys $\omega_1=L(0)$, so $\omega/2$ is an identity for this product on $V_2$. A normalized invariant bilinear form exists under the stated self-contragredience assumptions. Its restriction to $V_2$ is nondegenerate: invariance makes $L(0)$ self-adjoint, different weights are orthogonal, and nondegeneracy of the full graded form then implies nondegeneracy on each finite-dimensional weight space. The product and form obey \[ \langle a\cdot b,c\rangle=\langle b,a\cdot c\rangle. \tag{2} \] Also $a_3b$ is a scalar multiple of $\mathbf1$, with that scalar given by the standard normalized weight-two pairing. These conclusions produce a commutative algebra with invariant form, the general setting of a Griess algebra. They do not prove associativity, and associativity is not expected in the moonshine example.

A simple algebraic countertest shows how much information is still missing. Even in dimension two there are nonisomorphic commutative unital algebras with nondegenerate invariant symmetric bilinear forms. One is $\mathbb C\oplus\mathbb C$, with the sum pairing. Another is $\mathbb C[\epsilon]/(\epsilon^2)$, with pairing $\langle a,b\rangle=\ell(ab)$ where $\ell(1)=0$ and $\ell(\epsilon)=1$. In the basis $1,\epsilon$ its Gram matrix is \[ \begin{pmatrix}0&1\\1&0\end{pmatrix}, \] which is nondegenerate. Both pairings satisfy $2$, but the second algebra has a nonzero nilpotent and the first does not. These are not candidate VOAs and do not contradict any uniqueness theorem. They demonstrate the precise logical insufficiency of dimension, a unit, commutativity and an invariant form as a stand-alone recognition argument.

A genuine reduction using weight two.

Dong--Griess--Lam prove a uniqueness theorem with the additional hypothesis that $V_2$, with its first product, is isomorphic to the Griess algebra of the moonshine VOA [E2]. The strong rationality and holomorphicity assumptions here supply the relevant finiteness and module hypotheses in their theorem. Consequently the following route is sufficient: \[ \text{the stated hypotheses} \ \Longrightarrow\ (V_2,a_1b)\cong (V^\natural_2,a_1b) \ \Longrightarrow\ V\cong V^\natural. \tag{3} \] The second arrow is a cited theorem; the first is the unproved part of this route. It would be circular to identify the first algebra as "the Griess algebra" merely because its dimension is $196884$.

This reduction also prevents an unjustified generation argument. A positive-energy VOA has infinitely many modes and weight spaces. Knowing an abstract product on $V_2$ does not, just by linear algebra, identify all products in higher weights or show that weight two generates the whole VOA. The theorem in [E2] supplies a nontrivial reconstruction under its hypotheses. If that theorem is invoked, its algebra-isomorphism hypothesis must actually be obtained; if it is not invoked, the required generation and compatibility arguments must be proved independently.

A second sufficient condition: a Virasoro frame.

There is another precise conditional completion. Suppose that $V$ admits a Virasoro frame, namely $48$ mutually commuting simple Virasoro subVOAs of central charge $1/2$, with the sum of their conformal vectors equal to $\omega$. Lam--Yamauchi prove that a holomorphic framed VOA of central charge $24$ with $V_1=0$ is moonshine [E3]. Thus proving the existence of such a frame in every $V$ under consideration would also finish this problem.

The number $48$ is forced by central charge addition: $48/2=24$. But that numerical identity does not construct even one Ising vector, let alone a commuting family with the correct sum. If $e$ is the conformal vector of a central-charge-$1/2$ Virasoro subVOA, its modes must satisfy, in the usual normalization, \[ e_1e=2e,\qquad e_2e=0,\qquad e_3e=\tfrac14\mathbf1. \tag{4} \] These are necessary low-weight equations. The assertion that the generated Virasoro VOA is the simple $c=1/2$ model also involves its relations, and the compatibility of $48$ such subVOAs involves mixed products. Solving only $e\cdot e=2e$ in a large commutative algebra would not establish a frame.

One can phrase this as a finite algebraic search after the weight-two structure constants are known. Equations $4$, orthogonality and the required sum impose polynomial constraints on coordinates of proposed frame vectors. But those structure constants are exactly part of the unknown VOA. A search performed in the known moonshine algebra verifies its own frame, not the existence of a frame in every algebra allowed by the original hypotheses. This distinction matters even if the search is exhaustive for that one input.

Verification and first unresolved step.

The accompanying exact calculation forms $E_4^3/\Delta-744$ from truncated integral $q$-series and checks the initial coefficients in $1$. It also checks the invariant-form identity and nilpotent obstruction in the two toy algebras. These computations verify normalization and the proposed countertest; they do not classify $196884$-dimensional algebras or construct a new VOA.

The mathematical output is therefore a forced character, explicitly derived weight-two constraints, and two sound conditional reductions. The first missing step is either recognition of the complete weight-two algebra in $3$, or the existence of a Virasoro frame satisfying all its defining conditions. Neither follows from the character calculation given here. The recent source [S1] discusses VOA equivalences and orbifolds; its presence does not license assuming the frame or an orbifold presentation for an arbitrary input VOA. The requested uniqueness statement remains unresolved by this attempt.

Sources

• S1 Equivalence relations on vertex operator algebras II, arXiv:2410.18166v2. https://arxiv.org/pdf/2410.18166. Formulation source.
• E1 Y. Zhu, Modular invariance of characters of vertex operator algebras, Journal of the American Mathematical Society 9 $1996$, 237--302. Used for the transformation law, not as a uniqueness theorem. https://doi.org/10.1090/S0894-0347-96-00182-8.
• E2 C. Dong, R. L. Griess Jr. and C. H. Lam, On the uniqueness of the moonshine vertex operator algebra. The additional weight-two Griess-algebra hypothesis is essential to the theorem used here. https://arxiv.org/abs/math/0506321.
• E3 C. H. Lam and H. Yamauchi, A characterization of the moonshine vertex operator algebra by means of Virasoro frames. This proves the framed case. https://arxiv.org/abs/math/0609718.