Asymptotic and Special-Function Methods

An approximation can be accurate in one region and fail abruptly near a turning point, coalescing saddle, or change of contour. This chapter connects asymptotic expansions with canonical differential equations and integral representations so that the limit, domain, and error remain visible.

Enter this chapter

Begin with Asymptotic Scales, Remainders, Uniformity, and Optimal Truncation to interpret an expansion correctly. For a separated wave equation, Special Functions from Equations and Boundary Data offers a direct entry.

Topics

1. Asymptotic Scales, Remainders, Uniformity, and Optimal Truncation — Defines the limiting scale and distinguishes fixed-order accuracy from convergence.
2. Laplace Method and Steepest Descent — Identifies which saddles contribute along a legally deformed contour.
3. Stationary Phase, Coalescing Saddles, and Stokes Geometry — Repairs isolated-saddle approximations near degeneracy and Stokes transitions.
4. WKB and Eikonal Methods and Turning-Point Matching — Connects short-wavelength propagation across turning points.
5. Mellin Transforms and Scaling Asymptotics — Relates scaling laws and logarithms to transform singularities.
6. Special Functions from Equations and Boundary Data — Selects canonical solutions using their equations, branches, and boundary conditions.
7. Heat Kernels, Zeta Functions, and Spectral Determinants — Connects heat traces, analytic continuation, and regulated spectral products.
8. Lagrangian Submanifolds, Generating Functions, and Semiclassical Phases — Interprets semiclassical phases through symplectic geometry.

How the ideas fit together

Steepest descent approximates an integral near its relevant saddle points, while WKB approximates a differential equation through rapidly varying phases. Both fail when their local nondegeneracy assumptions fail. Coalescing saddles and simple turning points can lead to Airy-type uniform approximations: one canonical function replaces several separately divergent terms. See NIST DLMF, §§ 2.4(v) and 2.8(iii).

A more accurate local expansion does not determine which global solution is required. Contour accessibility, boundary conditions, and normalization still select the physical combination. Nor does adding terms indefinitely guarantee improvement: a divergent asymptotic series can have a useful optimal truncation.

Review the chapter

Why does a small expansion parameter alone not justify WKB everywhere?

Solution

The local error also depends on derivatives relative to the local wavelength. Near a turning point the leading wavenumber vanishes, so that ratio can become large even while the nominal parameter is small. A uniform local model and matching are then needed.

Where to go next

Continue to Differential Equations and Green Operators when selecting the boundary problem, or return there to check the exact operator behind an approximation.

References

• National Institute of Standards and Technology, Digital Library of Mathematical Functions, Chapter 2, “Asymptotic Approximations.” Maintained reference.