Automorphisms of Stokes multipliers in higher-order WKBJ theory

Abstract: We consider the Stokes phenomenon and higher-order Stokes phenomenon (HOSP) of formal asymptotic transseries arising in the WKBJ analysis of linear differential equations and integral problems. We introduce a framework of automorphisms that act on the Stokes constants of the divergent expansion, explained via late-late-term expansions and parametric Alien calculus, to capture this phenomenon. Our method is applied to a paradigmatic example: we obtain the full Stokes line structure and automorphisms for the Swallowtail problem from catastrophe theory, which contains four WKBJ components. We demonstrate that, in a system with four or more WKBJ components, the automorphism associated with the HOSP can itself change value across another higher-order Stokes line, which occurs when different higher-order Stokes lines intersect. We then argue that no additional special behaviour emerges for transseries with five or more WKBJ components.

• The paper develops a unified Alien-calculus framework in which ordinary Stokes automorphisms transform transseries parameters while higher-order automorphisms transform Stokes constants.
• The paper proves that four WKBJ components realize all qualitatively distinct higher-order Stokes switching, while systems with five or more components introduce no new mechanism.
• The paper completely analyzes the four-component swallowtail integral, finding intersecting higher-order Stokes lines that can simultaneously activate multiple Stokes multipliers and affect exact-WKB quantization.

Overview

This paper develops a framework for describing the Stokes phenomenon and the higher-order Stokes phenomenon (HOSP) in divergent asymptotic transseries using automorphisms, and applies it to a complete analysis of the swallowtail integral from catastrophe theory (2603.13332). The central objects are two families of maps: the ordinary Stokes automorphism $\mathcal{S}_{i>j}$, which acts on transseries parameters $\sigma_i$ across Stokes lines, and the higher-order Stokes automorphism $\mathcal{T}_{i>k;j}$, which acts on the Stokes constants $S_{ij}$ themselves across higher-order Stokes lines. Both are derived from Ecalle's second bridge equation in parametric Alien calculus.

The paper establishes two structural results. First, for transseries containing four or more WKBJ components, the higher-order Stokes automorphism can itself change value along an existing higher-order Stokes line, at points where distinct higher-order Stokes lines intersect. Second, no qualitatively new behaviour arises for five or more WKBJ components: all changes to higher-order automorphisms remain governed by three-singulant interactions. Consequently, the full generality of Stokes phenomenon is realised already at four WKBJ components.

Automorphisms from late terms and Alien calculus

The starting point is a Gevrey-1 asymptotic transseries of the form

$$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$

whose late-term coefficients diverge factorial-over-power as $n\to\infty$, with contributions indexed by singulant differences $\chi_j-\chi_i$. Introducing the dotted Alien derivative $$\dot{\Delta}_\chi = \mathrm{e}^{-\chi(z)/\epsilon}\Delta_\chi$$, whose key property is commutation with the singularly perturbed derivative, the authors show that when $\psi$ is a formal solution of a linear ODE, each Alien derivative is proportional to another transseries component. This yields Ecalle's second bridge equation,

$$\dot{\Delta}_{\chi_j-\chi_i}\,\psi = \frac{S_{ij}\sigma_i}{2\pi\mathrm{i}}\frac{\partial\psi}{\partial\sigma_j},$$

so that crossing a Stokes line $\sigma_i$0 (where $\sigma_i$1 is real and positive) induces the linear map $\sigma_i$2. The Stokes constants $\sigma_i$3 are analytic invariants of the ODE, independent of boundary data; the transseries parameters encode boundary conditions.

The HOSP arises by reexpanding the late terms themselves: since $\sigma_i$4 diverges as $\sigma_i$5 with contributions proportional to $\sigma_i$6, one obtains iterated Alien derivatives satisfying $\sigma_i$7. Resumming the optimally truncated $\sigma_i$8-summation shows that $\sigma_i$9 jumps by $\mathcal{T}_{i>k;j}$0 across curves where $\mathcal{T}_{i>k;j}$1 is real and positive — the higher-order Stokes line $\mathcal{T}_{i>k;j}$2. A crucial point is that $\mathcal{T}_{i>k;j}$3 need not equal the iterated composition, so the HOSP genuinely modifies the Stokes data rather than reproducing an existing constant.

Geometrically, the framework packages the total Stokes phenomenon along a path $\mathcal{T}_{i>k;j}$4 as a word in generators $\mathcal{T}_{i>k;j}$5 and $\mathcal{T}_{i>k;j}$6, inducing a $\mathcal{T}_{i>k;j}$7 bundle over $\mathcal{T}_{i>k;j}$8 with parallel transport on the transseries parameters. Activity of both line types is encoded in adjacency graphs local to each region of the plane: a Stokes line $\mathcal{T}_{i>k;j}$9 is active iff $S_{ij}$0, and a higher-order Stokes line $S_{ij}$1 is active iff $S_{ij}$2.

Why four components suffice

The paper argues that no new phenomenon appears beyond four WKBJ components. Iterating the reexpansion of amplitude functions into their own late terms ("late-late-late-terms") merely regenerates the same three-index structure with relabelled indices. With four or more singulants, however, the factors $S_{ij}$3 and $S_{ij}$4 appearing in $S_{ij}$5 can themselves change value where $S_{ij}$6 intersects $S_{ij}$7 or $S_{ij}$8 — requiring at least four distinct singulants. This effect was previously observed only implicitly in a fifth-order PDE study by Body, King and Tew; the present work identifies it explicitly as a change of the higher-order automorphism across pre-existing higher-order Stokes lines. For five or more components, the only new feature is a more frequent occurrence of such crossings, not any new type of switching.

The swallowtail problem

The swallowtail integral

$S_{ij}$9

satisfies a fourth-order ODE and hence possesses four WKBJ components, making it the minimal example exhibiting every variant of Stokes phenomenon. Under the far-field rescaling $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$0, $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$1, $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$2, the governing equation becomes singularly perturbed, and the four singulants are obtained exactly via the Hamilton–Jacobi system associated with the Borel operator, yielding $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$3 with $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$4 solving a quartic. The expansion has three singular points in general; special parameter values reduce this to two (Pearcey-like) or one (Airy-like) turning points.

Stokes constants are computed at a reference point ($$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$5) by deforming steepest-descent contours in the integral formulation, following Berry and Howls' hyperasymptotic adjacency method. For this normalisation the constants take values in $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$6 and satisfy $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$7. Applying the higher-order automorphisms across the full network of higher-order Stokes lines then determines $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$8 throughout $$\psi(z,\epsilon) \sim \sum_{i=1}^{N} \epsilon^{-\alpha_i} \sigma_i \mathrm{e}^{-\chi_i(z)/\epsilon}\left[\psi^{(i)}_0(z)+\epsilon\psi^{(i)}_1(z)+\cdots\right],$$9.

The resulting structure exhibits several notable features:

• Virtual turning points: higher-order Stokes lines emanate from points where $n\to\infty$0 but the corresponding Stokes constant vanishes nearby, rendering these lines inactive.
• Crossing higher-order Stokes lines: for $n\to\infty$1, four higher-order Stokes lines intersect at a single point. Across this intersection, the constants $n\to\infty$2, $n\to\infty$3, and $n\to\infty$4 change simultaneously from zero to non-zero values, so that three Stokes lines switch activity through one crossing point.
• Stokes crossings: the active Stokes-line graph contains intersections of three and six Stokes lines, dividing the neighbourhood into five and eight sectors respectively, with the transseries parameters given explicitly in each sector as integer combinations of the initial parameters.

For the specific behavioural condition matching the swallowtail integral ($n\to\infty$5, $n\to\infty$6), many active Stokes lines are irrelevant because the inducing exponential is absent. The authors note that correctly accounting for the varying higher-order automorphisms is necessary to enforce quantization conditions on subdominant components — a misassignment of Stokes constants would propagate errors into any exact-WKB quantization built on this transseries.

Limitations and open questions

The analysis carries explicit assumptions. The method presumes that higher-order Stokes lines do not coincide exactly with ordinary Stokes lines (which would modify the Stokes automorphism itself, per Nemes' work on Dingle's final main rule) nor with one another (which would produce a new form of higher-order automorphism). The authors state that computing behaviour under such coincidences depends substantially on the geometry involved and leave it as an open problem. The treatment also assumes the transseries solves a linear ODE — the bridge-equation derivation relies on this — and while the algebraic-expansion case common in applied problems fits as the special case $n\to\infty$7, nonlinear problems yield unboundedly many transseries parameters and Stokes constants, requiring careful selection of which to track. Finally, the demonstration restricts to real $n\to\infty$8 with complex $n\to\infty$9; extension to complex parameters is stated to be straightforward but is not carried out.

Conclusion

The paper reformulates regular and higher-order Stokes phenomena as a hierarchy of automorphisms — linear maps on transseries parameters generated by dotted Alien derivatives, and automorphisms of those maps acting on Stokes constants — derived consistently from Ecalle's bridge equation. Applied to the swallowtail catastrophe, the framework produces the first complete Stokes structure for the four-saddle regime, revealing that higher-order Stokes automorphisms change value across intersecting higher-order Stokes lines and that multiple Stokes lines switch activity at shared crossing points. The structural conclusion that four WKBJ components exhaust the possible variants of Stokes switching provides a classification benchmark for resurgent asymptotics of higher-order differential equations.