Modularity from $q$-series

Technical Summary: Modularity from q-Series

Problem Statement
The paper addresses a challenge posed by G. E. Andrews, which in turn references a problem by L. Ehrenpreis: to devise a direct method for proving the modularity of $q$-series summation forms without relying on pre-existing modular input (such as the modularity of Dedekind's $\eta$-function, Jacobi theta functions, or Eisenstein series). Many $q$-series arising in combinatorics, representation theory, and physics exhibit "mysterious" modularity that is not visible from their summation formulas. The primary test case for this challenge is the Rogers–Ramanujan pair, whose modularity is classically known via infinite product identities but lacks a proof derived solely from their series definitions.

Methodology
The author, Ken Ono, develops a criterion for vector-valued modularity based entirely on $q$-series algebra, first-order $q$-differential systems, and analytic continuation. The methodology proceeds through the following steps:

1. Local Differential Systems: The paper treats a vector of holomorphic $q$-series, $f = (f_1, \dots, f_r)^T$, as a solution to a first-order linear system on the upper-half plane. By examining local expansions at cusps (specifically $\infty$ and $0$), the author constructs local $q$-differential systems of the form $\theta_\gamma f^{(\gamma)} = A_\gamma(q_\gamma) f^{(\gamma)}$, where $\theta = q \frac{d}{dq}$ and $A_\gamma$ is a matrix of meromorphic functions.
2. The "Good" Condition: A vector $f$ is defined as "good" for a finite orbit datum $O$ if it satisfies three conditions:
   • Local Fundamental Matrix: At each cusp, there exists a matrix $\Phi$ formed by finitely many $SL_2(\mathbb{Z})$-translates of $f$ that is invertible in a punctured neighborhood of the cusp.
   • Regularity: The logarithmic derivative matrix $A = (\theta \Phi)\Phi^{-1}$ is meromorphic on the punctured disk, extends holomorphically to the cusp, and has a limiting exponent matrix conjugate to a diagonal matrix of rational numbers.
   • Finite Continuation Closure: The local systems are closed under analytic continuation and $T$-shifts (translations) up to constant gauge transformations. This ensures that the local patches can be glued together globally.
3. Gluing and Monodromy: Using standard results from the theory of linear differential equations, the paper demonstrates that if the "good" conditions are met, the local systems glue to form a global meromorphic system. The monodromy of this global system yields a representation $\rho: SL_2(\mathbb{Z}) \to GL_r(\mathbb{C})$, thereby certifying that $f$ is a vector-valued modular function.

Key Contributions and Results

• Theorem 2 (Modularity from $q$-Series): This is the central theoretical result. It establishes an equivalence between a vector of $q$-series being a weakly holomorphic vector-valued modular function and the vector being "good" for a finite orbit datum. This theorem provides a purely intrinsic, $q$-series-based criterion for modularity, independent of external modular forms.
• Theorem 1 (Rogers–Ramanujan Modularity): The paper applies Theorem 2 to the Rogers–Ramanujan pair $G(q)$ and $H(q)$.
  • It utilizes Schur's classical finite recurrences and Andrews's finite product comparisons to derive explicit first-order $q$-differential systems for the Rogers–Ramanujan series at the cusps $\infty$ and $0$.
  • The coefficients of these systems are expressed in terms of Lambert series. Crucially, the proof does not assume these Lambert series are Eisenstein series; it only uses their $q$-expansions and the resulting differential systems.
  • The author verifies that the Rogers–Ramanujan pair satisfies the "good" conditions, thereby proving their modularity and explicitly deriving the multiplier matrices $\rho(T)$ and $\rho(S)$ without invoking modular input.
• Theorem 3: This theorem provides the constructive converse, showing that any weakly holomorphic vector-valued modular function for a congruence subgroup naturally satisfies the "good" conditions, thus validating the equivalence.
• Extension to Andrews–Gordon Series: The paper sketches the application of this method to the Andrews–Gordon $q$-series (a generalization of Rogers–Ramanujan). It notes that finite polynomial identities in the Andrews–Gordon literature yield the necessary differential systems to certify modularity for these higher-rank families.

Significance and Claims
The paper claims to solve the Ehrenpreis-Andrews problem by providing a "direct proof" of modularity that relies solely on the internal structure of the $q$-series (algebraic recurrences and analytic properties) rather than on the modularity of other known functions.

• Intrinsic Nature: The significance lies in the removal of "modular input." The proof for the Rogers–Ramanujan case does not use the transformation laws of $\eta$ or theta functions, nor does it rely on the fact that the associated Lambert series are weight 2 Eisenstein series.
• Conceptual Framework: The work reframes the detection of modularity as a problem of analytic continuation and monodromy of $q$-differential systems. It draws a parallel to the Lewis–Zagier correspondence between period functions and Maass cusp forms, positioning the $q$-differential systems as the $q$-analogue of period function equations.
• Relation to Nahm's Conjecture: The paper briefly contrasts its approach with Nahm's conjecture, which relates modularity to torsion in the Bloch group. The author posits that the "good for $O$" condition offers a different route to modularity and suggests that clarifying the relationship between Bloch group torsion and the existence of finite orbit data is an interesting open direction, though the paper does not resolve this relationship.

In summary, the paper offers a rigorous, self-contained mechanism to certify the modularity of $q$-series vectors by verifying a finite set of algebraic and analytic conditions at the cusps, successfully demonstrating this on the foundational Rogers–Ramanujan case.