Bessel Functions

⚡ Key Takeaways

Bessel functions are the canonical solutions $J_\nu(x)$ and $Y_\nu(x)$ of Bessel's differential equation $x^2 y'' + x y' + (x^2 - \nu^2)y = 0$ . They are the "cylindrical analogues" of sine and cosine, governing oscillations and fields with circular or cylindrical symmetry. Applies to: vibrating drumheads, heat flow in cylinders, electromagnetic waves in fibres, planetary motion, frequency modulation. Origin: Daniel Bernoulli (1730s), formalized by Friedrich Wilhelm Bessel (1817–1824).

💡 The Core Insight

Separate the wave or Laplace equation in cylindrical coordinates and the radial part is forced to obey Bessel's equation. Bessel functions are therefore the radial eigenfunctions of the Laplacian on a disk — the natural "sines and cosines of the round world." Their oscillation comes from the $+x^2$ term (a restoring force); their amplitude decay like $x^{-1/2}$ comes from energy spreading over an ever-larger circumference.

DefinitionBessel's Differential Equation

For a parameter $\nu \ge 0$ (the order), Bessel's equation is

1

\quad x^2 \frac{d^2 y}{dx^2} + x \frac{dy}{dx} + (x^2 - \nu^2)\, y = 0.

Derivation hint: separate $\nabla^2 u + k^2 u = 0$ in cylindrical coordinates; the radial factor $R(r)$ with $x = kr$ obeys this equation, with $\nu$ the angular separation constant. Here $x$ is the (dimensionless) radial variable and $\nu$ is the order, typically an integer $n$ or half-integer for physical problems.

DefinitionBessel Function of the First Kind

The solution bounded at $x=0$ is the Bessel function of the first kind, defined by its series

2

\quad J_\nu(x) = \sum_{m=0}^{\infty} \frac{(-1)^m}{m!\,\Gamma(m+\nu+1)} \left(\frac{x}{2}\right)^{2m+\nu}.

Derivation hint: substitute the Frobenius ansatz $y = \sum_m a_m x^{m+r}$ into [1]; the indicial equation gives $r = \pm\nu$ , and the recurrence $a_m = -a_{m-2}/[m(m+2\nu)]$ yields the coefficients with $\Gamma$ extending the factorial. Here $\Gamma$ is the Gamma function, $\Gamma(n+1)=n!$ for integers. For integer order, $J_{-n}(x) = (-1)^n J_n(x)$ .

DefinitionBessel Function of the Second Kind

The second, linearly independent solution (singular at the origin) is the Bessel function of the second kind (Weber's function):

3

\quad Y_\nu(x) = \frac{J_\nu(x)\cos(\nu\pi) - J_{-\nu}(x)}{\sin(\nu\pi)},

Derivation hint: for non-integer $\nu$ , $J_\nu$ and $J_{-\nu}$ are independent; this combination is constructed so the limit $\nu \to n$ exists (via L'Hôpital) and produces a logarithmic second solution. For integer $n$ , $Y_n(x)$ contains a $\ln x$ term and diverges like $-\tfrac{2}{\pi}\ln(2/x)$ as $x\to 0^+$ .

TheoremGeneral Solution

Every solution of Bessel's equation [1] is a linear combination

4

\quad y(x) = A\, J_\nu(x) + B\, Y_\nu(x), \qquad A, B \in \mathbb{C}.

Derivation hint: [1] is a second-order linear ODE, so its solution space is two-dimensional; $J_\nu$ and $Y_\nu$ have nonzero Wronskian $W = 2/(\pi x)$ , hence are independent.

ProofFrobenius Construction of

J_\nu

Step 1. The point $x=0$ is a regular singular point of [1], so seek $y = \sum_{m\ge 0} a_m x^{m+r}$ with $a_0 \neq 0$ .

Step 2. Substitute into [1]. The lowest-power term $x^r$ gives the indicial equation $r^2 - \nu^2 = 0$ , so $r = \nu$ or $r = -\nu$ .

Step 3. Equating the coefficient of $x^{m+r}$ to zero gives the recurrence

5

\quad a_m\big[(m+r)^2 - \nu^2\big] + a_{m-2} = 0 \;\Rightarrow\; a_m = \frac{-a_{m-2}}{m(m+2r)} \;\;(\text{taking } r=\nu).

Derivation hint: collect like powers after differentiating the series twice and multiplying by $x^2$ , $x$ as in [1].

Step 4. All odd-index $a_m$ vanish (since $a_{-1}=0$ ). For even $m=2k$ , iterate:

6

\quad a_{2k} = \frac{(-1)^k\, a_0}{2^{2k}\, k!\,(\nu+1)(\nu+2)\cdots(\nu+k)}.

Derivation hint: repeatedly apply [5] with $r=\nu$ ; the denominator telescopes into a rising factorial.

Step 5. Choose the conventional normalization $a_0 = 1/[2^\nu \Gamma(\nu+1)]$ . Using $\Gamma(\nu+k+1) = (\nu+k)\cdots(\nu+1)\Gamma(\nu+1)$ , the coefficients collapse exactly into the series [2]. $\blacksquare$

∎

FW

Friedrich Wilhelm Bessel

1784–1846

A self-taught German astronomer and mathematician, Bessel was director of the Königsberg Observatory. He encountered these functions in 1817 while studying Kepler's equation for planetary perturbations — the Fourier coefficients of the eccentric anomaly turned out to be exactly $J_n$ . His 1824 memoir gave the first systematic treatment, including recurrence relations and the integral representation. He is also famous for the first reliable measurement of stellar parallax (61 Cygni, 1838) — the same precision instincts that drove his function theory.

🧠 Intuition

Think of a circular drumhead. A radial slice must vibrate, but unlike a guitar string, the "length" of string available at radius $r$ shrinks toward the center and the energy is spread around a circle of circumference $2\pi r$ . So as the wave travels inward it gets compressed (no clean nodes near the center) and outward it gets stretched out (amplitude decays). That is exactly $J_0$ : an oscillation whose wavelength is roughly constant but whose amplitude falls like $1/\sqrt{r}$ . The asymptotic form makes this precise:

7

\quad J_\nu(x) \sim \sqrt{\frac{2}{\pi x}}\,\cos\!\left(x - \frac{\nu\pi}{2} - \frac{\pi}{4}\right), \qquad x \to \infty.

Derivation hint: apply the method of stationary phase or the WKB approximation to the integral representation; the $x^{-1/2}$ prefactor reflects conservation of energy over an expanding wavefront. So far from the origin, $J_\nu$ is just a phase-shifted, amplitude-damped cosine — the cylindrical version of a plane wave.

Figure

J_0

,

J_1

,

J_2

: oscillation with decaying amplitude and unequal zeros

ExampleFundamental Frequency of a Circular Drum

Given: A circular drumhead of radius $a = 0.30\text{ m}$ , wave speed $c = 100\text{ m/s}$ (set by tension and density). The lowest mode is axisymmetric, $u \propto J_0(kr)\cos(\omega t)$ .

Step 1. Boundary condition: the rim is fixed, $u(a)=0$ , so $J_0(ka)=0$ . The smallest positive root is $j_{0,1} = 2.40483$ .

Step 2. Thus $ka = j_{0,1} \Rightarrow k = 2.40483/0.30 = 8.016\text{ m}^{-1}$ .

Step 3. Frequency: $\omega = ck = 100 \times 8.016 = 801.6\text{ rad/s}$ , so

f = \frac{\omega}{2\pi} = \frac{801.6}{6.2832}.

\boxed{f \approx 127.6\ \text{Hz}}

Note: unlike a string, the overtones (next roots $j_{0,2}=5.520$ , $j_{0,3}=8.654$ ) are not integer multiples — which is why drums sound inharmonic rather than pitched.

ExampleSeries Evaluation of

J_0

at a Point

Given: Compute $J_0(1)$ to four decimals using the series [2] with $\nu = 0$ .

Step 1. With $\nu=0$ , $\Gamma(m+1)=m!$ , so

J_0(x) = \sum_{m=0}^\infty \frac{(-1)^m}{(m!)^2}\left(\frac{x}{2}\right)^{2m}.

Step 2. At $x=1$ , $(x/2)^{2m} = (1/4)^m$ . Tabulate terms:

$m=0:\ 1$
$m=1:\ -\tfrac{1}{(1)^2}(0.25) = -0.25$
$m=2:\ +\tfrac{1}{(2)^2}(0.0625) = +0.015625$
$m=3:\ -\tfrac{1}{(6)^2}(0.015625) = -0.000434$
$m=4:\ +\tfrac{1}{(24)^2}(0.003906) = +0.0000068$

Step 3. Sum: $1 - 0.25 + 0.015625 - 0.000434 + 0.0000068 = 0.7651978$ .

\boxed{J_0(1) \approx 0.7652}

The rapid factorial-squared decay means just five terms give full four-digit accuracy.

⚠️ Common mistake

$J_\nu$ is NOT periodic, and its zeros are NOT equally spaced. Students often treat $J_0$ like a cosine. The zeros $j_{\nu,k}$ approach equal spacing $\pi$ only asymptotically (from [7]); near the origin they are noticeably farther apart. Likewise the amplitude is not constant — it decays as $x^{-1/2}$ . Treating Bessel functions as exact sinusoids will give wrong eigenvalues and wrong resonant frequencies.

Counterexample

"Both solutions are usable on a full disk." Consider heat flow on a solid disk $0\le r\le a$ . One might keep $Y_0$ in the general solution [4]. But $Y_0(r)\to -\infty$ as $r\to 0$ , which is physically impossible for a finite temperature at the center. Hence $B=0$ is forced: only $J_0$ survives. $Y_\nu$ is essential only on annular domains (a washer shape) where $r=0$ is excluded — there both kinds are needed to satisfy boundary conditions at the inner and outer radii.

connection

Bessel functions sit at a crossroads of analysis and physics. They are a special case of the Hypergeometric Function (confluent type), and the modified Bessel functions $I_\nu, K_\nu$ solve the same equation with $x^2 \to -x^2$ , appearing in diffusion and decaying fields. They are eigenfunctions in Sturm–Liouville Theory, giving rise to Fourier–Bessel series that generalize Fourier Series to circular domains. The half-integer orders reduce to elementary spherical Bessel functions $j_n(x)$ (e.g. $j_0 = \sin x / x$ ), central to Spherical Harmonics and quantum scattering. Their generating function $\exp\!\big[\tfrac{x}{2}(t - 1/t)\big] = \sum_n J_n(x)t^n$ links them directly to the Laplace Equation and to Generating Functions.

🔬 Where this appears

⏳ History

- 1732 — Daniel Bernoulli encounters

J_0

studying the oscillation of a hanging chain (the first appearance).

- 1764 — Leonhard Euler analyzes the vibrating circular membrane, deriving the functions for integer order.

- 1770 — Lagrange uses series of this type in his theory of planetary orbits.

- 1817–1824 — Friedrich Bessel, working on Kepler's equation, systematizes the functions, deriving recurrences and integral forms; the 1824 Berlin memoir is the definitive founding paper.

- 1867 — Carl Neumann and Hermann Hankel develop the second kind and the Hankel functions

H^{(1,2)}_\nu = J_\nu \pm i Y_\nu

for travelling cylindrical waves.

- 1878 — Lord Rayleigh applies Bessel functions extensively in The Theory of Sound.

- 1879 — Heinrich Weber gives the modern definition of

Y_\nu

used in [3].

- 1922 — G. N. Watson publishes A Treatise on the Theory of Bessel Functions, still the encyclopedic reference.

- 1964 — Abramowitz & Stegun's Handbook tabulates zeros and values, standardizing computation.

open-problem

Several deep questions remain active. The distribution of Bessel zeros $j_{\nu,k}$ — their precise spacing and their dependence on $\nu$ — connects to random matrix theory and the still-open conjectures about eigenvalue statistics. The transcendence and irrationality of individual zeros and of values like $J_0(1)$ is largely unknown; only partial results exist (e.g. positive zeros of $J_\nu$ are transcendental for rational $\nu$ , by Siegel). Efficient uniform asymptotics valid across the transition region near $x \approx \nu$ (where the function changes from monotone to oscillatory) require Airy-function corrections and remain a subtle numerical challenge for large order. Finally, the analytic continuation of $\nu$ into the complex plane underlies open problems in the theory of integral transforms.

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also