Representations of vertex operator algebras and braided finite tensor categories

We discuss what has been achieved in the past twenty years on the construction and study of a braided finite tensor category structure on a suitable module category for a suitable vertex operator algebra. We identify the main difficult parts in the construction, discuss the methods developed to overcome these difficulties and present some further problems that still need to be solved. We also choose to discuss three among the numerous applications of the construction.

💡 Research Summary

The paper surveys two decades of progress toward endowing suitable module categories of vertex operator algebras (VOAs) with the structure of a braided finite tensor category. It begins by recalling why such a categorical framework is desirable: in two‑dimensional conformal field theory (CFT) the fusion of primary fields, modular transformations of characters, and the Verlinde formula all point to an underlying tensor‑categorical structure. The authors then identify the principal technical obstacles that have historically prevented a straightforward construction. First, many VOAs of interest are not semisimple; their module categories contain indecomposable but reducible objects, and the usual tensor product defined via intertwining operators may fail to be associative or to admit a braiding. Second, the convergence and analytic continuation of products of intertwining operators become subtle when logarithmic terms appear, as in logarithmic CFTs. Third, establishing rigidity (existence of duals) and the modularity of the resulting category requires delicate control over projective covers and the behavior of characters under the modular group.

To overcome these difficulties, the paper outlines the methodology pioneered by Huang, Lepowsky, and collaborators. Their “logarithmic tensor product theory” starts from the notion of grading‑restricted generalized modules and imposes the C₂‑cofiniteness condition (or its variants) to guarantee that the space of intertwining operators is finite‑dimensional. Under these hypotheses one can define a tensor product bifunctor by means of a universal property involving the convergence of products of intertwining operators. The authors explain how the associativity isomorphisms are constructed from the analytic continuation of three‑point functions and how the braiding isomorphisms arise from the skew‑symmetry property of VOAs. A crucial step is the proof of modular invariance for the space of one‑point functions on the torus; this yields the S‑ and T‑matrices that serve as the categorical twist and ribbon structure. Rigidity is then established by showing that every simple object admits a projective cover and that evaluation/coevaluation maps satisfy the required triangle identities, thereby turning the braided finite tensor category into a modular tensor category (MTC).

The survey proceeds to illustrate three representative applications of this categorical machinery. (1) The Verlinde formula: the fusion coefficients obtained from the tensor product coincide with the entries of the modular S‑matrix, confirming the conjectural link between fusion rules and modular transformations for C₂‑cofinite rational VOAs. (2) Logarithmic Verlinde theory: for non‑semisimple examples such as the triplet W‑algebra, a modified Verlinde formula involving pseudo‑characters and generalized S‑matrices is derived, showing that even in the logarithmic setting the categorical framework yields computable fusion data. (3) Construction of three‑dimensional topological quantum field theories (TQFTs): the obtained MTCs feed directly into the Reshetikhin‑Turaev construction, providing invariants of 3‑manifolds that are interpreted physically as partition functions of Chern‑Simons‑type theories coupled to logarithmic CFTs.

In the final section the authors list open problems that remain. A major challenge is to prove rigidity for broader classes of logarithmic VOAs without relying on C₂‑cofiniteness, which would require a deeper understanding of projective resolutions in the VOA module category. Extending the tensor‑category construction to VOAs that fail the C₂‑condition, or to higher‑genus modular functors, is another direction. Moreover, a systematic theory of “non‑semisimple modular forms” that captures the transformation properties of logarithmic characters is still lacking. The paper concludes that, while substantial progress has been made—culminating in a robust theory of braided finite tensor categories for a wide range of VOAs—significant work remains to fully integrate logarithmic phenomena and to exploit the resulting categories in both mathematics and physics.