Laplace's Equation
Leave a room's radiator on all night, or leave a metal rod's hot end burning for long enough, and
eventually the temperature everywhere stops changing. Every wisp of transient wobble has died away,
and what's left is the
steady state —
the permanent, unmoving temperature pattern the system settles into for good. Setting the time
derivative to zero in the heat equation u_t = \alpha(u_{xx} + u_{yy})
leaves exactly that permanent pattern behind:
\nabla^2 u = u_{xx} + u_{yy} = 0.
This is Laplace's equation, and \nabla^2 (the
Laplacian) is arguably the single most important operator in all of mathematical
physics. Despite having no time variable, no rate of change, and no motion in sight, it governs an
enormous range of physical situations: steady heat flow through a solid, the electrostatic
potential around a charged object, the velocity potential of an incompressible fluid, gravitational
potential in empty space — completely different physics, all obeying the very same equation. Its
solutions are called harmonic functions.
The mean-value property
The Laplacian measures how a value compares with the average of its surroundings. Setting it to
zero forces every point to equal the average of its neighbours. Made precise, this is the
mean-value property: the value of a harmonic function at the centre of any circle
equals the average of its values around that circle.
Two consequences follow at once. A harmonic function can have no interior bumps or dips — no local
maxima or minima inside the region (that is the
maximum principle).
And it is completely determined by its values on the boundary: fix the edge, and the smooth
"membrane" stretched across the inside is forced. A simple harmonic function is
u = x^2 - y^2, since u_{xx} + u_{yy} = 2 - 2 = 0.
Worth checking the mean-value property with actual numbers, because it looks almost too generous to
be true. Take that same u = x^2 - y^2 and the circle of radius
1 centred at the origin, parametrised by
(\cos\theta, \sin\theta). Around that circle,
u = \cos^2\theta - \sin^2\theta = \cos(2\theta), and averaging
\cos(2\theta) all the way round from 0 to
2\pi gives exactly 0 — precisely
u(0, 0) = 0^2 - 0^2 = 0, the value at the centre. The positive bulge
where \cos(2\theta) > 0 is cancelled, point for point, by the negative
dip where it's less than zero. That perfect cancellation, for every circle you could draw,
is what "harmonic" really buys you.
- \nabla^2 u = u_{xx} + u_{yy} = 0 — the steady state of diffusion; an elliptic PDE.
- Solutions (harmonic functions) satisfy the mean-value property: each value is the average over any surrounding circle.
- No interior maxima or minima; the solution is fixed entirely by its boundary values.
Worked example: a metal plate settling down
Take a flat, square sheet of metal. Clamp the left edge at a scorching 100^\circ
and the right edge in an ice bath at 0^\circ; hold the top and bottom
edges wherever you like. Turn on the heat and wait. At first the interior temperature sloshes
around, warming unevenly as heat diffuses in from the hot edge — that's the transient part,
governed by the full heat equation. But leave it long enough and the interior temperature stops
changing altogether: it has reached its steady state, and that steady, permanent distribution is
exactly the solution of \nabla^2 u = 0 with the given edge temperatures
as boundary data.
Here's the payoff of the mean-value property: no point inside the plate can ever end up
hotter than 100^\circ or colder than 0^\circ,
no matter how oddly shaped the plate is or how the top and bottom edges are held. Every interior
value is trapped between the boundary's hottest and coldest extremes, because it's forced to equal
an average of its neighbours, which is itself forced to equal an average of the boundary. Heat
can't manufacture a hot spot out of nowhere in a steady state — it can only pass through.
A harmonic saddle
A heatmap of u = x^2 - y^2: warm where u > 0,
cool where u < 0, pale near zero. It rises along the
x-axis and falls along the y-axis — a saddle.
Notice there is no isolated hottest or coldest spot in the interior; the extremes all sit on the
edge, the hallmark of a harmonic function.
Compare this with the metal plate from the worked example above: there, every boundary temperature
was non-negative, so the whole interior stayed non-negative too — a single hump sagging smoothly
down to the edges. Here the boundary data changes sign (rising along one axis, falling along the
other), so the interior is forced into a genuine saddle instead of a single hump. Either way, the
rule holds without exception: whatever extremes appear anywhere on the picture, they appear only on
the boundary, never buried in the middle.
It isn't a coincidence of vocabulary. A vibrating guitar string settles into pure tones called
harmonics — standing-wave patterns that are the natural "modes" of the string.
Solutions of Laplace's equation on a disk or a sphere turn out to be built from an entirely
analogous family of natural modes, called (unsurprisingly) spherical harmonics:
the smoothest possible patterns a function can trace out on a sphere, ranked from simplest to most
wrinkled, in exactly the way a string's fundamental note is followed by its overtones. The two
ideas — a musical string's pure tones and a sphere's smoothest possible temperature patterns —
emerge from the very same piece of mathematics, which is why 19th-century physicists reached for
the same word for both. Spherical harmonics later turned out to be exactly the shapes of an atom's
electron orbitals — the same mathematics, yet again, three uses later.
Every other PDE you've met so far — the heat equation, the wave equation — is an
initial-value problem: you're handed a starting snapshot at
t = 0 and asked to run it forward in time. Laplace's equation has
no time variable at all, so there is no "starting configuration" to speak of and nothing
to run forward. Instead it is a pure boundary-value problem: you specify
conditions all the way around the edge of a region, once, and the equation fills in the entire
interior in one shot. Don't go hunting for an initial condition here — there isn't one, and asking
"what happens after that?" doesn't even make sense, because Laplace's equation describes a
situation with no "after" left in it.
Solving Laplace's equation is exactly how physicists and engineers compute the electric field and
potential around charged conductors of any shape — a wire, a charged sphere, the metal plates
inside a capacitor, the intricate electrodes of a particle accelerator. Wherever there's no charge
sitting in the empty space itself, the electrostatic potential is forced to be harmonic, obeying
precisely the same \nabla^2 u = 0 as the metal plate above. That's the
quietly beautiful part: the mathematics that tells you how heat settles in a metal sheet is
identical, symbol for symbol, to the mathematics of a charged conductor's field and to the
mathematics of steady, swirl-free fluid flow. Learn to solve it once, and you've solved three
different branches of physics at the same time.
The simplest possible case makes the point vividly. Between two large parallel charged plates — one
held at potential 0, the other at V_0, a gap
d apart — the potential in between is just the straight ramp
u(x) = V_0 x / d. Check it: u_{xx} = 0, so it's
harmonic, and sure enough it obeys the "no interior extremes" rule perfectly — the potential rises
in a dead-straight line from one plate to the other, with no bump or dip anywhere in between, and
the field between the plates (its slope) comes out perfectly uniform, exactly as every physics
textbook's capacitor diagram promises.