Math Methods · Learn

Bessel Functions

Solve the wave equation, heat equation, or Laplace's equation in a circle or cylinder, and the radial direction stops behaving like sin/cos. It behaves like a Bessel function instead — and once you've met one, you start seeing them everywhere round.

▶ Play the lab📖 Learn the theory

1Where the equation comes from

Take the 2-D wave equation (a drumhead) or the heat equation on a disk, and separate variables in polar coordinates $r,\theta$. The angular part comes out as familiar sines and cosines, $\cos(n\theta)$, $\sin(n\theta)$, because angle is periodic. The radial part $R(r)$ satisfies, after the substitution $x=kr$:

$$ x^2 y'' + x y' + (x^2-n^2) y = 0 $$

This is Bessel's equation of order $n$. It is linear but has a variable coefficient, so it doesn't reduce to constant-coefficient methods — its solutions get their own name.

2The series solution and $J_n$

Using the Frobenius method (a power series times $x^n$) gives the solution that stays finite at $x=0$:

$$ J_n(x) = \sum_{k=0}^{\infty} \frac{(-1)^k}{k!\,(n+k)!}\left(\frac{x}{2}\right)^{n+2k} $$

$J_n$ — the Bessel function of the first kind — looks like a decaying, oscillating cosine: it has infinitely many zeros, but the spacing settles toward $\pi$ and the envelope decays like $1/\sqrt{x}$ rather than staying flat. The second independent solution, $Y_n(x)$ (second kind), blows up logarithmically at $x=0$, so it only enters when the domain doesn't include the center — an annulus, a pipe wall, a coaxial cable.

3Zeros are eigenvalues

A drumhead clamped at radius $a$ needs the radial solution to vanish there: $J_n(ka)=0$. So the allowed values of $k$ — and therefore the allowed vibration frequencies — are exactly the zeros of $J_n$, scaled by $1/a$. The lab numbers them for you; finding them is the same root-finding problem as the Root Finding lab, just applied to a transcendental special function instead of a polynomial.

▶ In the lab

Sweep the order $n$ and watch where $J_n(0)$ sits (1 for $n=0$, 0 for every $n\ge1$) and how far out the first zero moves as $n$ grows. Open the lab →

4Which order, and why

Order / kindUse when
$J_0$Axisymmetric problems — no angular variation. Center of a drumhead, steady radial heat flow into a solid cylinder.
$J_n,\ n\ge1$Patterns with $n$ angular nodal lines — a drum mode that isn't perfectly symmetric, a non-axisymmetric temperature or field distribution.
$Y_n$ (second kind)Domain excludes the origin — annulus, hollow pipe, coaxial geometry — where a solution singular at $r=0$ is still allowed.
Modified $I_n,K_n$Same equation with a sign flip ($x^2y''+xy'-(x^2+n^2)y=0$) — shows up in steady diffusion/heat-conduction problems instead of oscillatory ones.

5A short history

Bessel functions show up well before Bessel. Daniel Bernoulli used something equivalent in 1732 studying the oscillations of a hanging chain, and Euler met them again working out the vibrating circular membrane. But it was the German astronomer Friedrich Bessel, in 1824, who systematically defined, tabulated, and named the functions that now bear his name — not while studying vibration at all, but while computing planetary position perturbations from Kepler's equation. That three unrelated problems — a drumhead, a cylindrical heat sink, and a planet's orbit — all reduce to the exact same ODE is a good demonstration of why "special functions" earn their keep: the equation, not the application, is what's fundamental.

Key takeaways

Frequently asked questions

How do Bessel functions work?

They're the solutions to Bessel's equation, which appears whenever you separate variables in the wave, heat, or Laplace equation using polar or cylindrical coordinates. The radial part of the solution behaves like a decaying, oscillating wave instead of a plain sine or cosine, because the equation has a variable coefficient.

What are Bessel functions used for?

Vibrating circular membranes (drumheads), steady heat conduction in cylinders and pipes, electromagnetic fields in coaxial cables and waveguides — and historically, planetary orbit perturbations, which is what Friedrich Bessel was actually computing when he systematized them.

Why does J₀(0) equal 1?

J₀ is the only Bessel function of the first kind that's nonzero at the origin — every $J_n$ with $n\ge1$ vanishes at $x=0$. This falls straight out of the series solution above, where the leading term of $J_n$ is proportional to $x^n$, which is 1 only when $n=0$.

What's the difference between Bessel functions of the first and second kind?

$J_n$ stays finite at the origin, so it's the one to use whenever the domain includes the center — a solid disk or rod. $Y_n$ diverges logarithmically at the origin, so it only enters when the domain excludes the center — an annulus, a hollow pipe, a coaxial cable.

EngineeringCandy · Learn · the theory behind the playground