A heat equation wearing an i
Guide 1 of this rung put you in front of the Navier-Stokes equations, where the unknown is a velocity field you can almost see. The Schrodinger equation asks something stranger of you: the unknown is a complex-valued field psi(x,t), the wavefunction, and what it describes is not a substance but a probability. Yet the equation that governs it is, on paper, almost the most familiar PDE you know. Strip away constants and write it for a free particle on a line: i psi_t = -psi_xx. Cover the i with your thumb and you are staring at the diffusion equation u_t = -u_xx run in reverse. The whole quantum revolution hangs on that one imaginary unit.
Restoring the physics, the time-dependent Schrodinger equation for one particle reads i hbar psi_t = -(hbar^2 / 2m) psi_xx + V(x) psi, where hbar is Planck's constant divided by 2 pi, m is the mass, and V(x) is the potential energy the particle sits in. The left side is first order in time, exactly like the heat equation; the right side is the Hamiltonian acting on psi — kinetic energy as a second derivative (a piece of the Laplacian in higher dimensions) plus potential energy as multiplication by V. So structurally it is parabolic in shape, one time derivative balanced against two space derivatives. The factor i is what refuses to let it behave parabolically.
Why the i flips diffusion into waves
The cleanest way to feel the difference is to feed each equation the same single mode and watch what time does to it. Both equations are linear, so by the superposition principle we may study one wave e^(i k x) at a time and add them back later. Plug a trial solution psi = e^(i k x) e^(-i w t) into the free equation i psi_t = -psi_xx (units chosen so the constants vanish). The space derivative pulls down a factor -k^2 from psi_xx, the time derivative pulls down -i w times i = w, and matching the two sides gives w = k^2. That little relation between frequency w and wavenumber k is the dispersion relation, and everything follows from its shape.
Here is the fork in the road. For the heat equation the same calculation gives w = -i k^2, an imaginary frequency, so each mode carries a factor e^(-k^2 t) that decays — high wavenumbers (sharp features) die fastest, which is the smoothing you met long ago. For Schrodinger, w = k^2 is real, so the mode carries e^(-i k^2 t), a pure phase of magnitude exactly 1 that rotates without shrinking. Nothing decays; each mode just spins in the complex plane at its own rate. The heat equation kills amplitudes, the Schrodinger equation only turns phases. That single switch from imaginary to real frequency is the whole story of the i.
feed e^(ikx) to each equation, read off the frequency w HEAT u_t = u_xx -> w = -i k^2 -> mode ~ e^(-k^2 t) DECAYS (smooths) SCHRODGR i psi_t = -psi_xx -> w = +k^2 -> mode ~ e^(-i k^2 t) ROTATES (|.|=1, conserves) WAVE u_tt = c^2 u_xx -> w = +- c k -> mode ~ e^(-i c k t) TRAVELS at speed c w real => amplitude preserved (Schrodinger, wave) w =/= linear in k => different k travel at different speeds => DISPERSION
Dispersion: a quantum particle spreads
A real frequency means amplitudes survive, but it does not mean a wave packet keeps its shape. Look again at the dispersion relations side by side. For the ordinary wave equation u_tt = c^2 u_xx the relation is w = c k, a straight line: every wavenumber moves at the same speed c, so a packet built from many modes travels as one rigid lump — that is why a clap reaches your ear undistorted. For Schrodinger, w = k^2 is a parabola: each mode moves at its own pace. This is dispersion, and it is the defining behavior of the equation.
The two natural speeds make this concrete. The phase velocity is w/k = k (the speed of a single ripple's crests), but the speed that matters physically is the group velocity dw/dk = 2k — the pace of the packet's envelope, where the particle's probability actually concentrates. Because the group velocity depends on k, a packet made of many wavenumbers cannot hold together: the fast components outrun the slow ones, the bump broadens, and a particle initially localized in a small region inexorably spreads out over time. An electron pinned to a point now is a smear of possibilities a moment later — not because anything pushed it, but because the PDE disperses.
Conservation, not smoothing
We promised the i protects total probability; here is the argument, and it is short. Multiply the free equation i psi_t = -psi_xx by the complex conjugate psi-bar, add the complex conjugate of that whole line, and the cross terms assemble into a perfect spatial derivative. What survives is a continuity equation d/dt |psi|^2 + d/dx J = 0, where J is a probability current. Integrate over all of space, assume psi vanishes far away, and the boundary term dies: the total probability d/dt of the integral of |psi|^2 is exactly zero. The norm is conserved for all time. This is the quantum analogue of the conservation of wave energy you met for vibrating strings — a real, exact invariant the flow can never spend.
Conservation has a sharp consequence: the Schrodinger equation does not smooth, and it is reversible. Run the heat equation forward and information is destroyed — sharp data instantly rounds off, and as guide 1 of the heat rung warned you, running it backward is hopelessly ill-posed. Schrodinger is the opposite. Because every mode only rotates by a phase, the map from psi at time 0 to psi at time t is invertible: just rotate every phase back. So the backward Schrodinger equation is perfectly well-posed — quantum evolution is time-reversible, in stark contrast to diffusion. The family of these phase rotations forms a one-parameter group of unitary operators (a group, not just a semigroup, precisely because you can go backward), the engine behind the abstract Cauchy-problem view of quantum dynamics.
Standing still: separation and energy levels
Where do the famous discrete energy levels of an atom come from? From exactly the tool you mastered on the heat and wave equations: separation of variables. Look for special solutions of the full equation i hbar psi_t = H psi that factor as psi(x,t) = phi(x) e^(-i E t / hbar) — a fixed spatial shape times a pure time-phase. Substituting, the time-phase cancels and the spatial part must satisfy the time-independent Schrodinger equation H phi = E phi. That is a spatial eigenvalue problem: phi is an eigenfunction of the Hamiltonian, and the constant E — the energy — is its eigenvalue. These special solutions are the quantum stationary states.
- Separate. Try psi(x,t) = phi(x) T(t); the time factor is forced to be T = e^(-i E t / hbar), a phase rotating at a rate set by the energy E.
- Solve the eigenvalue problem. The shape phi must satisfy H phi = E phi together with its boundary conditions — for a particle trapped in a box, phi must vanish at both walls.
- Watch quantization appear. Only special values of E admit a nonzero phi obeying the boundary conditions; those discrete allowed E are the energy levels, exactly as boundary conditions once selected discrete frequencies for a vibrating string.
- Superpose. A general state is a sum of these stationary states; each term just spins at its own phase rate e^(-i E_n t / hbar), and the interference between them is all the time-dependence there is.
Notice how exactly this mirrors the standing waves of a vibrating string. There, fixed ends allowed only certain wavelengths, giving the discrete overtones of a musical note. Here, the boundary or decay conditions on phi allow only certain energies, giving the discrete spectral lines of an atom — the colors a heated gas emits. The quantization of energy that launched quantum mechanics is, mathematically, nothing more exotic than the same eigenvalue problem that explains why a guitar string has a definite pitch. A spatial operator plus boundary conditions yields a discrete spectrum; physics merely supplies the operator.
When the equation bites back: nonlinearity
Everything so far rode on linearity — separation, superposition, the clean phase rotation. But some physics makes the medium respond to the wave's own intensity, and then the potential V depends on |psi|^2. The result is the nonlinear Schrodinger equation i psi_t = -psi_xx - |psi|^2 psi, one of the most studied nonlinear PDEs in existence. It governs intense light in an optical fiber, where the glass's refractive index shifts with the light's own brightness, and it governs cold-atom condensates. Superposition is gone; you can no longer build a general solution by adding simple modes, and the analysis turns genuinely hard.
Out of that hardness comes something beautiful. Dispersion wants to spread the packet, exactly as before — but the nonlinear term acts like a self-made potential well that focuses the wave back inward. When these two effects balance precisely, the spreading and the focusing cancel and you get a soliton: a localized lump that travels without changing shape at all, a particle-like solution of a wave equation. It is the same balance that, in the Korteweg-de Vries equation for shallow water, produces a wave that crosses a canal intact for miles. Solitons are why a pulse of light can carry data down a transoceanic fiber without dissolving.
Be honest about how special this is, though: an exact balance is a knife's edge. Tune the nonlinearity stronger, or move to two or three space dimensions, and focusing can overwhelm dispersion, driving the solution to blow up — the amplitude diverges to infinity in finite time, the wavefunction collapsing to a spike. Whether and when such blow-up happens, and what a solution even means afterward, is a live research question, a quantum cousin of the open regularity puzzle for Navier-Stokes that the final guide of this rung confronts. The same imaginary unit that made the linear equation so tame makes its nonlinear sibling one of the frontiers.