Voros Coefficient in Exact WKB Analysis

• The Voros coefficient is a regularized WKB integral that subtracts the divergent leading term to compare different normalizations of formal solutions on spectral curves.
• It is explicitly formulated via Bernoulli series and Gamma-function expansions in models such as Painlevé, hypergeometric, and Heun-type equations.
• It governs transseries parameters and Stokes phenomena, serving as a critical normalization datum in quantum curves and exact WKB analysis.

A Voros coefficient is a regularized WKB integral attached to a path or cycle on a spectral curve. In exact WKB analysis it is typically defined by subtracting the divergent leading contribution from a Riccati or WKB one-form and integrating the remainder between a turning point and a singular endpoint, or around a closed cycle; it measures the discrepancy between natural normalizations of formal solutions and controls Stokes or parametric Stokes phenomena (Iwaki, 2013, Iwaki et al., 2021). In quantum-curve formulations its exponentials are the Voros symbols, and in several hypergeometric, Painlevé, and Heun-type problems these quantities admit explicit Bernoulli- or Gamma-function descriptions (Iwaki et al., 2018, Aoki et al., 2021, Iwaki et al., 7 May 2026).

1. Definition in exact WKB theory

For the third Painlevé equation of type $D_6$, a Voros coefficient is defined as a regularized integral of the odd WKB term with the divergent leading part subtracted. For a path $\Gamma(\tau,\infty)$ from a turning point or simple pole $\tau$ to $\infty$,

$$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$

and similarly for paths ending at the double-pole branches $0_{c_\ast}$ (Iwaki, 2013). Here $R_{\rm odd}$ is the odd part of the formal Riccati solution, and the subtraction of $\eta R_{-1}$ is the regularization that makes the integral convergent at the singular endpoint (Iwaki, 2013).

In the quantum-curve and topological-recursion setting, the same object is formulated as an integral of the odd TR/WKB one-form. For cycles $\gamma\in H_1(\Sigma,\mathbb Z)$ and relative paths $\beta\in H_1(\Sigma,D_\infty,\mathbb Z)$,

$\Gamma(\tau,\infty)$0

where $\Gamma(\tau,\infty)$1 is obtained from $\Gamma(\tau,\infty)$2 by subtracting the $\Gamma(\tau,\infty)$3 and $\Gamma(\tau,\infty)$4 endpoint singularities (Iwaki et al., 2021). This formulation makes explicit the distinction between cycle Voros coefficients and path Voros coefficients.

A closely related convention appears for second-order and higher-order hypergeometric equations, where the coefficient is written using the full Riccati solution $\Gamma(\tau,\infty)$5 and regularized by subtracting the non-integrable terms $\Gamma(\tau,\infty)$6: $\Gamma(\tau,\infty)$7 This is the convention used for the Weber equation and for the confluent family of Gauss hypergeometric equations (Iwaki et al., 2018, Iwaki et al., 2018).

The Airy equation provides the local model in which the coefficient is effectively trivial after canonical normalization. In that case the exact connection formula is

$\Gamma(\tau,\infty)$8

with no extra factor of the form $\Gamma(\tau,\infty)$9; the connection data are exhausted by the universal Stokes constant $\tau$0 (Aoki et al., 2022). This suggests that nontrivial Voros coefficients arise from genuinely global normalization data rather than from the universal local simple-turning-point model.

2. Normalization, transseries, and Stokes phenomena

The fundamental role of a Voros coefficient is to compare different normalizations of the same formal or resummed solution. For $\tau$1, if $\tau$2 denotes the transseries normalized at a turning point $\tau$3 and $\tau$4 the one normalized at $\tau$5, then

$\tau$6

Thus the Voros coefficient is the multiplicative renormalization of the instanton parameter between two normalizations (Iwaki, 2013).

This normalization role becomes decisive when Stokes geometry degenerates. For $\tau$7, triangle-type and loop-type degenerations occur when certain parameter combinations become purely imaginary, and the corresponding Bernoulli-series expressions for the Voros coefficients cease to be Borel summable (Iwaki, 2013). The Borel sums of the basic building blocks satisfy jump formulas

$\tau$8

and these factors become the connection coefficients for the transseries parameter (Iwaki, 2013).

In exact WKB on marked bordered surfaces, the exponentials of Voros coefficients are the Voros symbols. For a cycle $\tau$9 and a path $\infty$0,

$\infty$1

and the corresponding symbols are $\infty$2 and $\infty$3 (Allegretti, 2018). The Borel sums of the cycle symbols associated with arcs of the WKB triangulation are identified with Fock–Goncharov coordinates of framed $\infty$4-local systems, and these Borel sums admit multivalued meromorphic continuation to all of $\infty$5, branched only at the origin (Allegretti, 2018). This places the Voros coefficient within the monodromy and cluster-geometric structure of exact WKB.

A more recent application appears in exact-WKB quantization on time-dependent backgrounds. There the Voros coefficient is

$\infty$6

and it relates turning-point-normalized exact WKB solutions to asymptotic-point-normalized ones by

$\infty$7

The paper states that without this factor the exact WKB solutions generally do not match the asymptotic vacuum normalization and the resulting mode functions can fail quantization (Namba et al., 23 Sep 2025). This suggests that the Voros coefficient is not merely a higher-order correction but a normalization datum required for global physical consistency.

3. Explicit evaluation and Bernoulli structures

A major feature of Voros coefficients is that, in many integrable examples, they collapse to universal Bernoulli series. For $\infty$8, all Voros coefficients are expressed in terms of

$\infty$9

with $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$0 and $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$1 (Iwaki, 2013). For example,

$$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$2

while the double-pole coefficients involve linear combinations of $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$3 and $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$4 (Iwaki, 2013). These formulas are derived from difference equations induced by Bäcklund transformations rather than by direct integration.

For the Weber equation, the Voros coefficient has an explicit Bernoulli-polynomial expansion,

$$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$5

and its regularized version is a finite difference of the topological-recursion free energy: $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$6 The same paper gives

$$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$7

showing that the coefficient is controlled by the same Bernoulli structure as the free energy (Iwaki et al., 2018).

The second part of that program extends the same pattern to the confluent family of Gauss hypergeometric equations. For each singular point $$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$8,

$$W_{\infty}({\bf c},\eta)=\int_{\Gamma(\tau,\infty)}\bigl(R_{\rm odd}(t,{\bf c},\eta)-\eta R_{-1}(t,{\bf c})\bigr)\,dt,$$9

and the coefficients $0_{c_\ast}$0 are given explicitly by Bernoulli polynomials evaluated at half-shifted combinations of the $0_{c_\ast}$1-parameters (Iwaki et al., 2018). The paper emphasizes that different Voros coefficients of the same differential equation arise from different half-$0_{c_\ast}$2 parameter shifts of a single free energy (Iwaki et al., 2018).

For the generalized hypergeometric equation $0_{c_\ast}$3 with a large parameter, Voros coefficients are defined separately at $0_{c_\ast}$4 and $0_{c_\ast}$5, for a pair of characteristic sheets $0_{c_\ast}$6: $0_{c_\ast}$7 The explicit formulas are finite sums of Bernoulli-polynomial terms, and their Borel summability is determined by the sign of the real parts of the relevant parameter differences (Aoki et al., 2021). The corresponding Borel sums are given by Gamma-function expressions, which is a standard exact-WKB signature of regularized normalization factors (Aoki et al., 2021).

4. Quantum curves, topological recursion, and BPS structures

In the topological-recursion framework, Voros coefficients become structural rather than merely auxiliary. For hypergeometric-type spectral curves, cycle coefficients satisfy the exact formula

$0_{c_\ast}$8

while path coefficients admit the BPS-sum expansion

$0_{c_\ast}$9

Here the Bernoulli-polynomial weights $R_{\rm odd}$0 depend on $R_{\rm odd}$1, and the sum is organized by the BPS spectrum $R_{\rm odd}$2 (Iwaki et al., 2021).

The exponentials of the Borel-resummed Voros coefficients satisfy the same jump formula as the BPS automorphism. The paper proves that the Borel sums of cycle and path Voros symbols solve the almost-doubled BPS Riemann–Hilbert problem, with solution

$R_{\rm odd}$3

This identifies the Voros symbols with the meromorphic functions required by Bridgeland’s formalism (Iwaki et al., 2021).

A further structural statement is that the path coefficients define a closed one-form on parameter space. If $R_{\rm odd}$4 is the natural basis, then

$R_{\rm odd}$5

and there exists a Voros potential $R_{\rm odd}$6 such that

$R_{\rm odd}$7

The BPS $R_{\rm odd}$8-function is then

$R_{\rm odd}$9

and at a special quantization parameter this agrees, up to a simple factor, with the Borel sum of the topological recursion partition function $\eta R_{-1}$0 (Iwaki et al., 2021). This suggests a broad reinterpretation of Voros coefficients as differential-geometric and wall-crossing data on parameter space.

5. Higher-order equations, Heun-type systems, and physical applications

Voros coefficients persist beyond second-order Schrödinger form. For third-order scalar equations associated with degenerations of the 2-dimensional Garnier system, the coefficient is defined by the regularized integral

$\eta R_{-1}$1

For the $\eta R_{-1}$2 quantum curve, the paper proves

$\eta R_{-1}$3

and derives the explicit Bernoulli-polynomial expansion

$\eta R_{-1}$4

For the $\eta R_{-1}$5 curve, by contrast, the Voros coefficient vanishes identically (Takei, 2020). The coexistence of a nontrivial and a zero example within the same framework shows that nontriviality is a global property of the quantum curve rather than a formal inevitability.

For the Heun equation and all of its confluent equations, the relevant exact-WKB quantity is formulated as a Voros period rather than a Voros coefficient: $\eta R_{-1}$6 The paper imposes

$\eta R_{-1}$7

or equivalently

$\eta R_{-1}$8

and uses this condition to determine formal series expansions of the accessory parameter for Heun and every confluent Heun equation in its table (Iwaki et al., 7 May 2026). The spectral curves in these cases have genus $\eta R_{-1}$9, and the paper gives a detailed prescription for choosing the vanishing cycle that matches the classical regular or irregular conformal block through the accessory parameter (Iwaki et al., 7 May 2026). This suggests that, for genus-one problems, the global period is the appropriate analogue of the more local Voros coefficient of genus-zero hypergeometric systems.

The 2025 paper on time-dependent backgrounds supplies a distinct application. There the Voros coefficient renormalizes exact WKB solutions from turning-point normalization to asymptotic singular-point normalization,

$\gamma\in H_1(\Sigma,\mathbb Z)$0

and enters the evolution matrix for mode functions (Namba et al., 23 Sep 2025). The paper states that if one were to use the turning-point-normalized solutions directly, the amplitudes would differ by the values of the Voros coefficients compared to the WKB approximation and quantization would fail (Namba et al., 23 Sep 2025). In this application the coefficient is part of the exact normalization of the quantum state rather than only a monodromy invariant.

6. Terminology and common confusions

The term “Voros” is used in several mathematically unrelated senses. The following distinctions are explicit in the cited literature.

Usage of “Voros”	Object studied	Relation to Voros coefficient
Exact WKB / quantum curves	Regularized WKB integral or its exponential	The standard setting
Noncommutative geometry	Voros star-product	Terminological mismatch
Analytic number theory	Voros criterion for RH	Different object

The paper “Noncommutative inspired Schwarzschild black hole, Voros product and Komar energy” is not about a Voros coefficient in the exact-WKB sense. It studies the Voros star-product and states explicitly that there is no introduction, definition, computation, or use of any “Voros coefficient” in the paper (Gangopadhyay, 2012). The same terminological mismatch holds for “Spinors and Voros star-product for Group Field Theory: First Contact,” which studies the Voros star-product on the noncommutative $\gamma\in H_1(\Sigma,\mathbb Z)$1 dual to $\gamma\in H_1(\Sigma,\mathbb Z)$2, not an exact-WKB coefficient (Dupuis et al., 2011).

A different mismatch occurs in the number-theoretic paper “Analysis of Voros criterion,” which studies a Voros-type criterion for the Riemann hypothesis in terms of zero sums and derivatives of $\gamma\in H_1(\Sigma,\mathbb Z)$3 at $\gamma\in H_1(\Sigma,\mathbb Z)$4, and explicitly does not introduce a standard standalone object named “the Voros coefficient” in the WKB or spectral sense (Sergey, 2014).

Even within exact WKB, terminology varies. Some papers reserve “Voros coefficient” for open-path normalization integrals and use “Voros period” for closed-cycle integrals, especially on higher-genus spectral curves (Iwaki et al., 7 May 2026). Others package both path and cycle cases under the common language of Voros coefficients and distinguish them by the underlying homology class (Iwaki et al., 2021). The most stable invariant across these usages is not the name but the function: a Voros coefficient or Voros period is a regularized WKB integral encoding normalization, monodromy, and Stokes data.