Holonomic Modular q-Difference Equations

• Holonomic systems of modular difference equations are defined by q-holonomic modules over the q-Weyl algebra satisfying a monic q-difference equation.
• The theory unifies analytic, algebraic, and modular properties through algorithmic methods that yield explicit monodromy matrices and modular transformation laws.
• Applications in quantum topology and complex Chern–Simons theory are demonstrated by examples such as generalized q-hypergeometric equations and knot invariants.

A holonomic system of modular difference equations, in the context of $q$-difference equations, is a mathematical structure that unifies analytic, algebraic, and modular properties of $q$–holonomic modules. The theory provides essential tools for understanding quantum invariants in knot theory, quantum topology, and complex Chern–Simons theory. Modular $q$-holonomic modules are defined by their cocycle structure under the action of $SL_2(\mathbb{Z})$ and their enhanced analyticity—allowing systematic, algorithmic solution of their associated linear $q$–difference equations. Their canonical examples include generalised $q$–hypergeometric equations and key $q$–holonomic modules encountered in quantum topology (Garoufalidis et al., 2022).

1. Definition and Algebraic Structure

A $q$–holonomic module is a finitely generated left module over the $q$–Weyl algebra $$\mathcal{D}_q = (\mathbb{C}(q)[t^{\pm1}])\langle \sigma \mid \sigma t = q t \sigma \rangle$$. It is of dimension $1$ over the commutative subalgebra $\mathbb{C}(q)[t^{\pm1}]$ and generated by a cyclic vector $f$ obeying a monic $q$–difference equation

$$L f = \sum_{j=0}^{r} a_j(t, q) \sigma^j f = 0, \quad a_r \neq 0,$$

where $\sigma$ acts as $\sigma f(t) = f(q t)$. The local solution space near $t=0$ forms a fundamental solution matrix $U(t, q)$ with columns given by Frobenius or $q$–Borel–Laplace–resummed solutions. Analogously, solutions near $t = \infty$ yield $V(t, q)$. The monodromy matrix $M(t, q) = V(t, q)^{-1} U(t, q)$ is an elliptic function on the variable $t$, reflecting the invariance under $t \mapsto q t$ up to conjugation.

A $q$–difference equation, or equivalently its associated $q$–holonomic module, is termed modular if for every $\gamma \in SL_2(\mathbb{Z})$ the cocycle

$$\Omega_{U, \gamma}(z, \tau) = (U|_{\kappa_U}\gamma)(z, \tau) U(z, \tau)^{-1}$$

extends meromorphically to the cut-plane $\tau \in \mathbb{C} \setminus \{-d/c\}$, $z \in S_\gamma + \mathbb{Z} + \tau \mathbb{Z}$ (for a finite $S_\gamma \subset \mathbb{C}$), and obeys the modular-slash functional equations with prescribed weights (Garoufalidis et al., 2022).

2. Modular Cocycle Structure and $SL_2(\mathbb{Z})$ Action

The modular properties are encoded by the $SL_2(\mathbb{Z})$–cocycle: $$\Omega_{U, \gamma\gamma'}(z, \tau) = \Omega_{U, \gamma}(\gamma'(z, \tau)) \Omega_{U, \gamma'}(z, \tau),$$ where the slash operator for weight $\kappa$ is

$$(F|_\kappa \gamma)(z, \tau) = (c\tau + d)^{-\kappa} F\left( \frac{z}{c\tau+d}, \frac{a\tau+b}{c\tau+d} \right),\quad \gamma = \begin{pmatrix} a & b \ c & d \end{pmatrix}.$$

The extension criterion for modularity reduces, by Theorem 1.5, to the cocycle’s extension for $\gamma=S$ (with the $T$–cocycle always trivial, $\Omega_T=1$). Thus, the required functional equations (for $S$ and $T$) are direct $q$–analogues of those satisfied by classical modular objects such as the Dedekind $\eta$–function and the Jacobi $\theta$–function.

If $U$ and $V$ are two fundamental matrices at $t=0$ and $t=\infty$ of weights $\kappa_U$ and $\kappa_V$, respectively, and if the monodromy matrix $M$ satisfies $M|_{\kappa_U} \gamma = \Delta_{\kappa_V, \gamma} M$ for a suitable diagonal factor $\Delta_{\kappa_V, \gamma}$, then $\Omega_{U, \gamma} = \Omega_{V, \gamma}$ for all $\gamma \in SL_2(\mathbb{Z})$. By the Gauss reduction of $SL_2(\mathbb{Z})$, extension for $S$ (and $T$) suffices for full modularity (Garoufalidis et al., 2022).

3. Generalised $q$–Hypergeometric Equations

A prototypical modular $q$–holonomic module is provided by the generalised $q$–hypergeometric equation. For parameters $a = (a_1, \ldots, a_r)$, $b = (b_0, \ldots, b_{r-1})$ with $b_0 = q$, define the operator

$$L = \prod_{j=0}^{r-1}(1 - q^{-1} b_j \sigma_t) - t \prod_{j=1}^{r}(1 - a_j \sigma_t), \quad \sigma_t f(t) = f(q t).$$

This exhibits a singularity structure visible in its Newton polygon, with solutions at $t=0$ given by

$$f^{(q^{-1}b_j)}(t) = B_j(a, b, q) \frac{\theta(q^{-1} b_j t; q)}{\theta(t; q)} {}_r\phi_{r-1}\left( q a / b_j ;\; q b / b_j ; q, t \right),$$

and at $t=\infty$ by

$$g^{(a_i^{-1})}(t) = A_i(a, b, q) \frac{\theta(q^{-1} a_i t; q)}{\theta(q^{-1} t; q)} {}_r\phi_{r-1}\left( q a_i / b ;\; q a_i / a ; q, q^r b_1 \cdots b_{r-1} a_1^{-1} \cdots a_r^{-1} t^{-1} \right).$$

The monodromy matrix $M = V^{-1} U$ is constructed using Heine's formula, with entries

$$M_{ij}(t) = \frac{(q;q)_\infty^3 \theta(b_j t; q) \theta(a_i / b_j t; q) \theta(a_i; q)}{\theta(t; q) \theta(a_i t; q) \theta(b_j; q) \theta(a_i / b_j; q)}$$

modulo normalization. Under the $S$ and $T$ actions, $M$ satisfies the appropriate modular transformation laws, confirming the modularity of all $_r\phi_{r-1}$–modules (Garoufalidis et al., 2022).

4. Examples from Complex Chern–Simons Theory

Several central $q$–holonomic systems arising in quantum topology and Chern–Simons theory are modular, with explicit construction of fundamental matrices and monodromies:

• $q$–Pochhammer Symbol: The equation $(1 - q t) f(q t, q) - f(t, q) = 0$ has fundamental solutions $f^{(0)}(t, q) = (q t; q)_\infty$, $g^{(1)}(t, q) = \theta(t; q)/(q; q)_\infty$, with trivial monodromy $M=1$. The $S$–cocycle computes as the inverse of the Faddeev quantum dilogarithm, which extends meromorphically to $\tau\in\mathbb{C}\setminus\mathbb{R}_{\le0}$.
• Appell–Lerch System: The system $(\sigma^2 + (q t - 1)\sigma - t)f = 0$ and its shifts yield a $2\times 2$ system with unipotent monodromy and $S$–cocycle comprised of the Mordell integral and theta–quotients, independent of auxiliary parameters.
• $4_1$–Knot Equation: The equation $$q t f(q t) + (1 - 2 t) f(t) + q^{-1} t f(q^{-1} t) = 0$$ leads to solutions at $t=0$ and $t=\infty$ in terms of $_2\phi_1$–series and theta–quotients. The monodromy function, with explicitly known elliptic entries, and the $S$–cocycle (the Andersen–Kashaev state–integral) both extend to the cut–plane, confirming modularity (Garoufalidis et al., 2022).

5. Algorithmic Solution Procedures

Modular linear $q$–difference equations are solvable via a robust five-step methodology:

1. Choice of Cyclic Vector: Represent the module with a single operator $L = \sum_{j=0}^r a_j(t, q) \sigma^j$, forming the companion matrix $A(t, q)$ such that $U(q t) = A(t) U(t)$.
2. Newton Polygon & Frobenius Solutions: The Newton polygon of $L$ reveals slopes whose corresponding prefactors $\theta(t; q)^\kappa$ and indicial polynomials lead to a basis of formal Frobenius solutions.
3. $q$–Borel Transform: For divergent formal power series, apply the $q$–Borel transform of order $\kappa$:

$$B_\kappa\left(\sum_{n\ge 0} a_n t^n\right)(\xi) = \sum_{n\ge 0} (-1)^n q^{\kappa n(n+1)/2} a_n \xi^n,$$

rendering them analytic in $\xi$.

1. $q$–Laplace Transform: Invert with the $q$–Laplace transform,

$$L_\kappa[f](t, \lambda) = \frac{1}{\theta(\lambda; q^\kappa)} \sum_{m\in\mathbb{Z}} (-1)^m q^{\kappa m(m+1)/2} \lambda^m f\left(q^{\kappa m}\lambda t\right),\quad (\kappa > 0),$$

or a contour-integral variant for $\kappa < 0$, recovering analytic solutions in $t$.

2. Monodromy via Poles & Residues: The matrices $U$ and $V$ attain meromorphic status in $t$. Their ratio $M=V^{-1}U$ is an elliptic function uniquely determined by its pole structure, principal parts, and normalization at $t=0$ or $t=1$.

In the modular case, the extension of $\Omega_S(z, \tau)$ to the cut-plane guarantees, through the cocycle relations, full modular analyticity for all $\gamma \in SL_2(\mathbb{Z})$ without further analyses (Garoufalidis et al., 2022).

6. Significance and Applications

The modularity property provides a conceptual explanation for several structural features of quantum invariants of knots and 3-manifolds as they appear in both exact and perturbative Chern–Simons theory. In particular, the theory underpins the closed-form evaluation of monodromy matrices for a wide class of $q$–holonomic systems—enabling explicit residue-factorizations and theta–quotient expressions. The methods facilitate effective computation and structural understanding of quantum invariants, exemplified by the specific cases of the $q$–Pochhammer symbol, Appell–Lerch system, and $4_1$–knot $q$–difference equations (Garoufalidis et al., 2022).