A Small Taste from My New Book: Season 2 Episode 11

 


The Airy integral.


Take the differential equation ()

Ansatz: Consider a Laplace-type integral

for some function  and contour  independent of .

Then

Plug into the ODE:

We want this to vanish for all , so we try to make the integrand a total derivative in . We’d like

for some . Compute:

Thus,

Matching terms gives

so

Solving,

Hence,

Now we check:

Then

But

so

The ODE is satisfied provided the term

vanishes —or, preferably, decays—at both ends of .

Set

so

For large , the dominant term is . Its behavior at infinity is determined by .

We have decay when

Let . Then . From  we get

Define three sectors:

Sector 1 (around , sector’s bisector is ):

Sector 2 (around , sector’s bisector is ):

Sector 3 (around , sector’s bisector is ):

Each sector has width .



To find the saddle points we differentiate:

Thus, there are two saddle points. Writing   and solving  we obtain 

For , this gives

Hence,

We seek for constant phase curves, i.e., . On those curves we must have decay as  goes to infinity.


Two curves pass through each saddle point, but only one enters sectors 1, 2 and 3 at the endpoints.

Let  where  . Do a Taylor expansion:

Now compute

Locally

Write . So:

Along a steepest-descent path, the condition  must hold. Hence, for the path passing through  (so  ), the steepest descent curve satisfies

From this

The condition implies

Hence

Since the directions repeat every , the distinct cases correspond to  . The resulting directions are

These are approximate: take  for  and  for .

The decay contours for  are shown in the following plot as  and . Therefore,

solves the differential equation for  or .

 

As the following plot suggests, reflecting  yields a third contour  (yellow) running from sector 1 below the x‑axis to sector 3 below its bisector.

Thus, the integrals

for  all solve the differential equation

Since this is second order, only two solutions are linearly independent, so the integrals must be related.

There is no singularity of the integrand within , so Cauchy’s theorem yields

Thus,

We define

This is the Airy integral which was studied by Airy in relation to diffraction near caustic surface.

Diffraction near a caustic surface occurs when light rays “pile up’’ along an envelope of rays, producing intense, rapidly varying brightness patterns that cannot be described by simple geometric optics alone. It is exactly in this regime that the Airy integral appears: it is the universal mathematical model for diffraction near a fold caustic. A fold caustic is a stable, bright line pattern formed by light rays converging and reflecting or refracting off a curved, irregular surface. It represents the boundary or "envelope" where light rays become tangent, creating a high-intensity, "folded" edge in the, for example, light patterns seen on the bottom of a swimming pool. 

In optics and mathematics, an envelope is a curve or surface that is tangent to every member of a family of other curves.

If we put , i.e., we reduce the path to the imaginary axis,

It still converges.

Indeed, let

then

Take . Split

The first term is finite so it suffices to study the tail.

Integrating by parts on ,

Notice that for large , , hence

Also,

so for

Thus,

converges as an improper oscillatory integral, and therefore so does its real part,

 

We have established that the oscillatory integral defining the Airy function is indeed convergent, providing a solid analytical foundation for its integral representation. This result allows us to treat the Airy function with confidence and opens the door to a deeper exploration of its properties.

In the next episode—the final chapter of Season2we move beyond convergence and uncover the asymptotic behavior of Airy functions, extracting precise formulas that reveal their behavior for large arguments.

And if you’re enjoying this journey through analysis, consider picking up The Riemann Hypothesis Revealed for a broader perspective on the fascinating world of modern mathematics.

 

Comments