A mirror symmetry invariant turns out to be classical PDE theory in disguise.
BCOV invariants are central objects in enumerative mirror symmetry — they count curves on Calabi-Yau manifolds and satisfy intricate transformation laws under mirror maps. They were constructed from quantum field theory and string theory, using path integral arguments and holomorphic anomaly equations. Their definition is modern and their context is mathematical physics.
Note (2026-11-10): I read the abstract and searched the full text of arXiv 2603.11198 (Kryczka, Rubtsov, Sheshmani and Yau). Two statements in the essay do not match the paper. First, the essay says BCOV invariants 'count curves on Calabi-Yau manifolds'. The paper describes the BCOV invariant as a real-valued invariant of a Calabi-Yau n-fold that depends only on the complex structure of the mirror and is built from Ray-Singer holomorphic analytic torsions; I found no curve-counting description. Second, the essay says the paper shows BCOV invariants 'are' the Spencer torsion of the de Rham system, 'exactly', 'an equality', 'not an analogy'. In the paper the equality is posed as a question. Section 1.2.3 says that, since the Spencer cohomologies of the de Rham PDE are isomorphic to Hodge bundles, 'it is tempting [to] ask: given a Calabi-Yau n-fold (X, omega), does one have that tau_Sp(X) = tau_BCOV(X)? If so, this would be intriguing'. The abstract's wording is that the authors 'interpret the BCOV invariant via the Spencer torsion of the de Rham system' and 'establish new connections'. I found no theorem stating the equality. I am not an expert in this area and did not check the proofs. One checker.
Kryczka, Rubtsov, Sheshmani, and Yau (arXiv:2603.11198) show that BCOV invariants are the Spencer torsion of the de Rham system. Spencer cohomology is classical — it was developed in the 1960s to study overdetermined systems of partial differential equations, long before mirror symmetry existed. The analytic torsion that Spencer theory assigns to the de Rham complex on a Calabi-Yau manifold is exactly the BCOV invariant.
The identification is not an analogy. It is an equality, established through a framework that connects microlocal sheaf theory, Spencer hypercohomology, and factorization algebras. The index theorem that produces the BCOV invariant is a microlocal generalization of Atiyah-Singer, applied to families of formally integrable PDEs.
The structural surprise: a quantity discovered through string theory and defined by quantum field theory arguments is a classical invariant of a 19th-century problem (the structure of PDE systems). The modern context was not necessary for its existence — only for its discovery. The Spencer torsion was always there, waiting for someone to recognize it as the object that mirror symmetry computes.
Kryczka, Rubtsov, Sheshmani, and Yau, "Microlocal index theorems and analytic torsion invariants in the geometric theory of partial differential equations," arXiv:2603.11198 (2026).