Borcherds' 1986 PNAS paper 'Vertex algebras, Kac-Moody algebras, and the Monster' constructs, for every Kac-Moody algebra A all of whose real roots have norm 2 and whose root lattice embeds in an even lattice R, and for every weight r in the weight lattice of R, an irreducible integrable A-module L(r), obtained from the Virasoro-physical subspaces of the lattice Fock-space vertex algebra. These modules strictly generalize the highest-weight, lowest-weight, and adjoint representations of A. The paper poses a single explicit problem: is L(r), up to isomorphism, the only A-module that is irreducible, contains weight r with multiplicity one, and has no weights s of squared norm exceeding (r,r)? The paper proves the answer is yes when r is a highest or lowest weight vector. The problem is recorded as open, with a precise surviving core: uniqueness is undecided for general A — in particular indefinite (hyperbolic or Lorentzian) A — and r not Weyl-conjugate to a highest or lowest weight vector, and no classification of irreducible integrable modules over indefinite Kac-Moody algebras exists against which the question could be checked.
Background
A Kac–Moody algebra is an infinite-dimensional Lie algebra (a vector space with an antisymmetric bracket satisfying the Jacobi identity) determined by a generalized Cartan matrix — an integer matrix with 2's on the diagonal and non-positive entries elsewhere. The definition simultaneously generalizes finite-dimensional semisimple Lie algebras (positive definite matrix), affine Lie algebras (positive semidefinite matrix; these are the central extensions of loop algebras, i.e. of maps from the circle into a semisimple Lie algebra), and indefinite Kac–Moody algebras (indefinite matrix); the hyperbolic algebras, the smallest indefinite case, are already of unbounded complexity [Feingold–Frenkel 1983]. The algebra carries a symmetric invariant bilinear form ; its roots — the nonzero weights of the adjoint action of on itself by the bracket — split into real roots, of squared length 2, and imaginary roots, and the Weyl group is the group generated by the reflections in the hyperplanes orthogonal to the real roots. A module for is a vector space on which acts by linear maps respecting the bracket. A weight of a module is a vector in the dual of the Cartan subalgebra (the abelian diagonal part of on which the root-space decomposition is a simultaneous eigenspace decomposition) such that the corresponding simultaneous eigenspace is nonzero; its multiplicity is . A module is integrable if for every real root the attached copy of (the 3-dimensional simple Lie algebra generated by the corresponding root pair) acts as a direct sum of finite-dimensional representations; a highest (resp. lowest) weight module is one generated by a vector annihilated by all positive (resp. negative) root spaces. For finite-dimensional semisimple algebras, integrable highest-weight modules exhaust the irreducibles, but for infinite-dimensional the module landscape is much wider and largely unclassified.
An even lattice is a free abelian group of finite rank with an integer-valued non-degenerate bilinear form taking even values on all elements. From Borcherds builds an explicit vertex algebra : roughly, the group algebra of (basis vectors written , ) tensored with a symmetric algebra on , carrying an infinite family of mutually compatible mode operators ; the underlying state space is the physicist's Fock space (the standard free-particle state space built from a group algebra and polynomial oscillator variables). Vertex algebras had just been introduced, motivated by the Frenkel–Lepowsky–Meurman construction of the moonshine module — the vertex algebra whose graded character is the modular function and whose automorphism group is the Monster, the largest of the 26 sporadic finite simple groups [FLM 1984]. The Virasoro algebra (the central extension of the Lie algebra of polynomial vector fields on the punctured line, with generators and central charge) acts on ; for each half-integer the physical subspace is , terminology from string theory, where such subspaces are the setting of the no-ghost theorem of Goddard–Thorn [1972] — the string-theory result that on these physical subspaces the natural indefinite Hermitian form is positive semidefinite, so quotienting out null vectors loses no information. The quotient ( a natural derivation of ) is a Lie algebra, and when contains the root lattice of a Kac–Moody algebra with all real roots of norm 2 (possibly modulo null vectors), the derived algebra of is realized in it, with at most a central kernel when is connected and not affine — this is the paper's realization theorem.
For each element of the weight lattice of (the dual lattice of all vectors in the real span of whose inner product with every element of is an integer) the paper constructs an irreducible -module : if , take the physical subspace with and quotient by its maximal graded submodule not containing the vector ; if , first rebuild the Fock space using the group algebra of the shifted lattice . The modules generalize the classical ones: if is a highest (or lowest) weight vector then is the usual integrable highest (lowest) weight module, and if is a real root then is a quotient of the adjoint representation (the action of on itself by the bracket). They satisfy: (i) is irreducible; (ii) the weight has multiplicity 1 and every weight with has multiplicity 0 — a condition that forces to be integrable, so multiplicities are constant on Weyl-group orbits; (iii) a contravariant bilinear form exists and is unique up to a scalar (positive definite when is a highest or lowest weight vector); (iv) all weight multiplicities are finite, though no closed formula for them is known; (v) the highest/lowest-weight and real-root cases reduce to known modules; (vi) Weyl-conjugate give the same module.
Beyond those known cases, is a genuinely new kind of integrable module: for indefinite no classification of irreducible integrable modules exists, and even the individual weight multiplicities of such modules are open — the paper proves only a general bound (a partition-function-type upper bound on multiplicities), and for hyperbolic algebras the multiplicity problem had been open since Feingold–Frenkel's foundational study [1983; Feingold–Frenkel–Ries 1993]. The same Fock-space machinery underlies Borcherds' later proof of the Conway–Norton moonshine conjectures (the 1979 prediction that the graded character series attached to the Monster's elements are exactly specific modular functions) [1992], where for the fake monster Lie algebra (the Lie algebra of physical states of the even unimodular Lorentzian lattice of signature , the Lie-algebraic analogue of the moonshine module) the analogous multiplicities are exactly computable by the no-ghost theorem.
Problem Statement
Quoting the primary source (Borcherds, Proc. Natl. Acad. Sci. USA 83 (1986), Section 6): "Problem: Is L(r) the only A module satisfying conditions i and ii above? (It is if r is a highest or lowest weight vector, and in this case condition ii implies condition i.)"
Unpacked. Let be an even lattice with integral non-degenerate bilinear form ; let be a Kac–Moody algebra all of whose real roots have norm 2 and whose root lattice embeds in (possibly modulo null vectors), realized inside the Lie algebra of physical states of the Fock-space vertex algebra of ; let be any element of the weight lattice of , i.e. the dual lattice of all vectors in the real span of whose inner product with every element of is an integer. Let be the irreducible -module constructed as the quotient of the physical subspace (or of the Fock space built from the shifted lattice when ) by its maximal graded submodule not containing . Question: is , up to isomorphism, the unique -module such that (i) is irreducible, and (ii) the weight occurs in with multiplicity exactly 1 and every weight of satisfies ?
The paper settles this affirmatively when is a highest or lowest weight vector (then is the usual integrable highest/lowest-weight module and condition (ii) implies condition (i)). The open core: general — in particular indefinite (hyperbolic or Lorentzian) — and not Weyl-conjugate to a highest or lowest weight vector. No classification of irreducible integrable modules over indefinite Kac-Moody algebras is known against which the question could be checked.
References
- Richard E. Borcherds, "Vertex algebras, Kac-Moody algebras, and the Monster", Proceedings of the National Academy of Sciences USA 83 (1986), no. 10, 3068-3071. DOI: 10.1073/pnas.83.10.3068, https://doi.org/10.1073/pnas.83.10.3068.
- Richard E. Borcherds, "Generalized Kac-Moody algebras", Journal of Algebra 115 (1988), no. 2, 501-512. DOI: 10.1016/0021-8693(88)90275-X, https://doi.org/10.1016/0021-8693(88)90275-X.
- Richard E. Borcherds, "Monstrous moonshine and monstrous Lie superalgebras", Inventiones Mathematicae 109 (1992), no. 1, 405-444. DOI: 10.1007/BF01232032, https://doi.org/10.1007/BF01232032.
- Richard E. Borcherds, "Problems in moonshine", in: First International Congress of Chinese Mathematicians (Beijing, 1998), AMS/IP Studies in Advanced Mathematics 20, American Mathematical Society, Providence, RI, 2001, pp. 3-10. DOI: 10.1090/amsip/020/01, https://doi.org/10.1090/amsip/020/01.
- Lisa Carbone, Walter Freyn, Kyu-Hwan Lee, "Dimensions of imaginary root spaces of hyperbolic Kac-Moody algebras", Contemporary Mathematics 623 (2014), 23-40. DOI: 10.1090/conm/623/12465, https://doi.org/10.1090/conm/623/12465.
- Vyjayanthi Chari, "Integrable representations of affine Lie-algebras", Inventiones Mathematicae 85 (1986), no. 2, 317-335. DOI: 10.1007/BF01389093, https://doi.org/10.1007/BF01389093
- Vyjayanthi Chari, Andrew Pressley, "A new family of irreducible, integrable modules for affine Lie algebras", Mathematische Annalen 277 (1987), no. 3, 543-562. DOI: 10.1007/BF01458331, https://doi.org/10.1007/BF01458331
- Vyjayanthi Chari, Andrew Pressley, "Integrable representations of twisted affine Lie algebras", Journal of Algebra 113 (1988), no. 2, 438-464. DOI: 10.1016/0021-8693(88)90171-8, https://doi.org/10.1016/0021-8693(88)90171-8.
- Kaiming Zhao, "Weight sets of irreducible integrable representations of Kac-Moody algebras of indefinite type", Communications in Algebra 24 (1996), no. 14, 4373-4383. DOI: 10.1080/00927879608825821, https://doi.org/10.1080/00927879608825821.
- Xiangqian Guo, Kaiming Zhao, "Irreducible representations of untwisted affine Kac-Moody algebras", Revista Matemática Iberoamericana 42 (2026), no. 4, 1487-1520. DOI: 10.4171/rmi/1620, https://doi.org/10.4171/rmi/1620.
- Michael Roitman, "On free conformal and vertex algebras", Journal of Algebra 217 (1999), no. 2, 496-527. DOI: 10.1006/jabr.1998.7834, https://doi.org/10.1006/jabr.1998.7834.
- Michael Roitman, "Combinatorics of free vertex algebras", Journal of Algebra 255 (2002), no. 2, 297-323. DOI: 10.1016/S0021-8693(02)00155-2, https://doi.org/10.1016/S0021-8693(02)00155-2.
- Alex J. Feingold, Igor B. Frenkel, "A hyperbolic Kac-Moody algebra and the theory of Siegel modular forms of genus 2", Mathematische Annalen 263 (1983), no. 1, 87-144. DOI: 10.1007/BF01457086, https://doi.org/10.1007/BF01457086.
- Alex J. Feingold, Igor B. Frenkel, John F. X. Ries, "Representations of hyperbolic Kac-Moody algebras", Journal of Algebra 156 (1993), no. 2, 433-453. DOI: 10.1006/jabr.1993.1083, https://doi.org/10.1006/jabr.1993.1083.
- Peter Niemann, "Some generalized Kac-Moody algebras with known root multiplicities", Memoirs of the American Mathematical Society 157 (2002), no. 746. DOI: 10.1090/memo/0746, https://doi.org/10.1090/memo/0746.
- Seok-Jin Kang, Kyu-Hwan Lee, Kyungyong Lee, "A combinatorial approach to root multiplicities of rank 2 hyperbolic Kac-Moody algebras", arXiv:1501.02026 (2015), https://arxiv.org/abs/1501.02026.
- Aidan Backus, Peter Connick, Joshua Lin, "An algorithm for computing root multiplicities in Kac-Moody algebras", arXiv:1912.04540 (2019), https://arxiv.org/abs/1912.04540.
- Igor B. Frenkel, James Lepowsky, Arne Meurman, "A natural representation of the Fischer-Griess Monster with the modular function J as character", Proceedings of the National Academy of Sciences USA 81 (1984), no. 10, 3256-3260. DOI: 10.1073/pnas.81.10.3256, https://doi.org/10.1073/pnas.81.10.3256
- Peter Goddard, Charles B. Thorn, "Compatibility of the dual Pomeron with unitarity and the absence of ghosts in the dual resonance model", Physics Letters B 40 (1972), no. 2, 235-238. DOI: 10.1016/0370-2693(72)90420-0, https://doi.org/10.1016/0370-2693(72)90420-0
- Elizabeth Jurisich, James Lepowsky, Robert L. Wilson, "Realizations of the Monster Lie algebra", arXiv:hep-th/9408037 (1994), https://arxiv.org/abs/hep-th/9408037.
- Robert McRae, "On integral forms for vertex algebras associated with affine Lie algebras and lattices", Journal of Pure and Applied Algebra 219 (2015), no. 5, 1236-1257. DOI: 10.1016/j.jpaa.2014.06.003, https://doi.org/10.1016/j.jpaa.2014.06.003.
- Scott Carnahan, "A self-dual integral form of the moonshine module", SIGMA 15 (2019), Paper 030. DOI: 10.3842/SIGMA.2019.030, https://doi.org/10.3842/SIGMA.2019.030.
- Samuel DeHority, "Affinizations of Lorentzian Kac-Moody Algebras and Hilbert Schemes of Points on K3 Surfaces", arXiv:2007.04953 (2020), https://arxiv.org/abs/2007.04953.
- Thomas Driscoll-Spittler, "Vertex operator expressions for Lie algebras of physical states", arXiv:2408.07597 (2024), https://arxiv.org/abs/2408.07597.