Helmholtz equation

The formula

∇²U + k²U = 0          k = ω/c = 2π/λ

The Laplacian of the field plus the wavenumber squared times the field is zero. U is the complex amplitude at each point in space — pressure for acoustics, a field component for EM — and k counts radians of phase per meter. One equation, one frequency, time already gone.

u(r, t) = Re{ U(r) · e^(jωt) }   substituted into   ∂²u/∂t² = c²·∇²u

Where it comes from. Assume the whole field oscillates at one frequency ω, plug that into the wave equation, and the e^(jωt) cancels off both sides. What's left is the Helmholtz equation — the wave equation with time separated out, which is why the old name is the reduced wave equation.

∇²U + k²U = −s(r)

The driven version. Put a source on the right side — a monopole, a current element, a shaker — and the solution is the steady-state field that source sets up at frequency ω.

G(r) = e^(−jkr) / (4π·r)

The free-space Green's function in 3D: what one point source radiates. Amplitude falls as 1/r, phase winds up at k radians per meter. Every antenna integral and every BEM acoustics code is built by stacking copies of this.

lim (r→∞)  r · (∂U/∂r + jkU) = 0

The Sommerfeld radiation condition. In an open domain the PDE alone has two solutions — waves going out and waves coming in from infinity. This condition throws away the incoming one. Sources must be sources, not sinks.

Where you meet it

Waveguide cutoff on the bench. Solve Helmholtz over the cross-section of WR-90 (22.86 mm broad wall) and the lowest eigenvalue says the TE₁₀ mode propagates above fc = c/(2a) = 6.56 GHz. Below that, k² − kc² goes negative and the field decays instead of traveling — at 3 GHz, WR-90 eats about 10.6 dB per centimeter. That's not loss in the resistive sense; it's the math refusing to carry the wave. It's also why a piece of below-cutoff guide makes a clean, calibrated attenuator.
EMC chamber and cavity reviews. A shielded room is a Helmholtz eigenvalue problem with conducting walls. A 4.8 × 2.4 × 2.4 m enclosure has its first resonance near 70 MHz — f = (c/2)·√((m/a)² + (n/b)² + (p/d)²) with the lowest nonzero index set. Below the first handful of modes a reverb chamber can't stir a statistically uniform field, which is why the test plan lists a lowest usable frequency and the review board asks how it was justified.
Acoustic test cells and control rooms. Same equation, air instead of RF. A room with a 6.1 m dimension has an axial mode at 343/(2·6.1) ≈ 28 Hz, and every low-frequency measurement made in it rides on top of that comb of resonances. Room-mode math is Helmholtz with rigid-wall boundary conditions and nothing more.
Frequency-domain simulation. Every FEM acoustics run, every full-wave EM solve at a spot frequency, every sonar and radar scattering prediction is a discretized Helmholtz problem. When the review asks why the model has four million elements for a one-cubic-meter box at 10 GHz, the answer is k ≈ 210 rad/m and a 30 mm wavelength that the mesh has to resolve everywhere.

How it works

The central shift is from "how does this evolve in time" to "what shapes can live here at this frequency." In a closed domain with boundary conditions, the Helmholtz equation only has nonzero solutions at discrete values of k — the eigenvalues. Those are the modes. A cavity doesn't resonate at certain frequencies because of some separate resonance physics; it resonates because those are the only k values where the boundary-value problem has an answer. Cavity modes, waveguide cutoffs, room modes, and drumhead patterns are all the same eigenvalue problem wearing different boundary conditions.

Geometry picks the special functions. Separate variables in a rectangle and you get sines and cosines; in a cylinder, Bessel functions; in a sphere, spherical Bessel functions and spherical harmonics ^[2]. That's the entire reason Bessel functions show up in coax, fiber, and circular-waveguide work — they're not a modeling choice, they're what the Helmholtz equation does in round geometry.

The sign of k² carries the physics. Positive means oscillatory solutions — propagation. Where the effective k² goes negative (below cutoff, past a total-internal-reflection boundary), solutions become growing and decaying exponentials, and the physical one decays. Evanescent fields are still solutions; they store energy near the boundary instead of carrying it away.

Three things bite people in practice. First, the time convention: EEs write e^(jωt), physicists write e^(−iωt), and the outgoing Green's function flips between e^(−jkr) and e^(+ikr) accordingly. Mix conventions between a textbook and a code and every phase in the model is conjugated — the magnitudes look fine and the phases are silently wrong. Second, the equation is single-frequency by construction. It says nothing about transients, startup, or broadband behavior; a pulse means solving at many frequencies and inverse-transforming, or going back to the time domain. Third, it assumes the medium is linear and, in the simple form, homogeneous — c constant. Graded media and flow make k a function of position and the clean modal picture starts to bend.

Numerically, Helmholtz is famously nastier than it looks. The operator isn't positive definite the way Laplace's is, so standard iterative solvers struggle. Worse, the "ten elements per wavelength" rule stops being enough as k grows: the discrete wave accumulates a phase error over many wavelengths, so the computed field drifts away from the best answer the mesh could represent. Babuška and Sauter proved this pollution effect cannot be eliminated in standard FEM in two or more dimensions ^[8] — you can only outrun it by refining faster than the wavelength rule or changing the method. If a high-frequency model was meshed by the same rule of thumb as the low-frequency one, that's the first thing to check.

History

Hermann von Helmholtz came to this equation from music, not mathematics. Trained as an army surgeon with an 1842 medical doctorate, he spent his first career in physiology before drifting toward physics, landing the Berlin physics chair in 1871 ^[5]^[6]. Along the way he built an experimental program on hearing and musical tone — the tuned resonators that carry his name were his instrument for picking individual harmonics out of a complex sound ^[6], and his major study of tone perception appeared in the early 1860s ^[5]^[6].

The equation itself comes from a 72-page paper in Crelle's Journal in 1860, "Theorie der Luftschwingungen in Röhren mit offenen Enden" — the theory of air oscillations in tubes with open ends ^[3]^[4]. The practical question was the organ pipe: what actually sets the pitch of a tube open at the ends, given that the air doesn't stop moving exactly at the mouth? Working the acoustics at a fixed frequency put the time-harmonic form ∇²U + k²U = 0 at the center of the analysis, and the equation has carried his name since ^[1].

The open-domain half of the story took another fifty years. Solving the equation in unbounded space leaves an ambiguity the PDE can't settle — mathematically, radiation converging from infinity is as valid as radiation going out. Arnold Sommerfeld closed that gap in 1912 in "Die Greensche Funktion der Schwingungsgleichung," stating the radiation condition with the line engineers still quote: the sources must be sources, not sinks, of energy ^[7]. Between Helmholtz's eigenvalue problem in the pipe and Sommerfeld's condition at infinity, the two halves of every modern antenna, sonar, and scattering solver were on the table.

Related tools

/tools/frequency-wavelength — converts between f, λ, and the k that drives everything above
/tools/sound-speed-materials — the c in k = ω/c for acoustic problems
/tools/lc-resonance — the lumped-element cousin of a cavity mode
/tools/vibration-natural-freq — same eigenvalue idea, one degree of freedom
/tools/vswr-return-loss — what standing waves in a guide look like at the connector
/tools/fresnel-zone — Helmholtz-equation diffraction applied to a link path

Sources

Written by HE in our own words from the cited sources — engineering judgment included, your stamp still required. All entries →