A flow that is secretly one function
Picture a thin layer of water sliding past an obstacle: the same picture in every horizontal slice, so the whole story is two-dimensional. We make the standard idealizations of an ideal fluid — the flow is steady (the picture never changes in time), incompressible (it neither piles up nor thins out anywhere), and irrotational (a tiny paddle-wheel dropped in would drift along without spinning). On the harmonic rung you learned that exactly these conditions make the velocity field the gradient of a harmonic function phi, the velocity potential: the fluid flows downhill on phi, and phi obeys Laplace's equation. So one harmonic function already governs the flow. The magic of two dimensions is that this real function is only half of something complex.
Recall the chain from the harmonic rung. A harmonic phi on a nice domain always has a harmonic conjugate psi, the partner function for which phi + i psi is holomorphic. The two are bolted together by the Cauchy-Riemann equations u_x = v_y, u_y = -v_x. The headline of this guide is that for a fluid flow these are not abstract bookkeeping — each Cauchy-Riemann equation is a physical law. One of them says the flow is incompressible; the other says it is irrotational. Holomorphy and 'nice flow' turn out to be the very same statement, dressed in different clothes.
The complex potential: one derivative gives the velocity
Bundle the velocity potential phi and its conjugate psi into a single holomorphic function. This is the complex potential Omega(z) = phi(x, y) + i psi(x, y), the object this whole guide is built around. Its real part phi is the velocity potential as before; its imaginary part psi is called the stream function, and it earns its name geometrically. The level curves phi = constant are the equipotentials, and the level curves psi = constant are the streamlines — the actual paths a speck of dust carried by the fluid would trace. Because phi and psi are harmonic conjugates, these two families cross at right angles everywhere, exactly like the grid lines of a sheet of graph paper that has been smoothly bent. The complex potential holds the flow's entire geometry in one holomorphic package.
Now the payoff that makes a complex analyst smile. The velocity field, both components at once, falls straight out of a single derivative of Omega. Write the flow velocity as (u, v); then Omega'(z) = u - i v. Notice the minus sign — the derivative gives the conjugate of the velocity, called the complex velocity. So to find the speed and direction of the flow anywhere, you differentiate one holomorphic function and read off the answer. Where Omega'(z) = 0 the velocity vanishes: that is a stagnation point, a spot where the fluid momentarily stands still, like the nose of a rock splitting a stream. One derivative of one function tells you the entire velocity field and where it pauses.
Omega(z) = phi(x,y) + i psi(x,y) holomorphic complex potential phi = const equipotential lines psi = const STREAMLINES (paths the fluid follows) Cauchy-Riemann for phi, psi <==> flow is incompressible AND irrotational Omega'(z) = u - i v (complex velocity = conjugate of velocity) Omega'(z) = 0 => stagnation point
Building flows, and the conformal superpower
The simplest complex potential is the simplest holomorphic function. Take Omega(z) = U z for a real constant U. Then phi = U x and psi = U y, so the streamlines psi = U y are horizontal lines and the velocity is Omega'(z) = U, a steady rightward flow at speed U. That is uniform flow — the empty wind tunnel. From this seed and a few cousins you assemble a whole catalogue: Omega = m log z gives a source spraying fluid outward from the origin (or a sink if m < 0); Omega = -i k log z gives a pure vortex, fluid circling the origin; and a constant over z gives a dipole. Because holomorphic functions add, you superpose these the way you add waves, building elaborate flows by stacking simple ones.
Here is the move that makes the whole subject powerful, and it is pure complex analysis. A conformal map preserves angles, so it carries the orthogonal grid of equipotentials and streamlines from one region faithfully onto another. If you know the flow on a simple shape — say uniform flow, or flow around a circular cylinder, both of which you can write down by hand — then composing the complex potential with a conformal map gives you, for free, the flow on a complicated shape. This is exactly the transplanting idea from the conformal-mapping rung: solve where it is easy, then push the answer across the map. The streamlines come along automatically, because holomorphy is preserved under composition. You never have to re-solve Laplace's equation on the hard region.
- Pick a simple region where you already know the flow — most often the exterior of a circle, where uniform flow past a cylinder has a tidy closed-form complex potential.
- Find a conformal map that sends this simple region onto the awkward region you actually care about (the outside of a wing, the channel around a corner).
- Compose: the complex potential on the hard region is just the known potential evaluated at the inverse map. Holomorphy guarantees it still describes a valid ideal flow.
- Read the physics straight off the transplanted Omega: streamlines from psi = constant, velocity from Omega'(z) = u - i v, stagnation points where that derivative vanishes.
From a circle to a wing: the Joukowski aerofoil
Now we name the conformal map that built aviation. The Joukowski map is w = (1/2)(z + 1/z), often written without the half as w = z + 1/z. Watch what it does to the unit circle: put z = e^(i theta), and the two terms combine to w = cos theta, sweeping the real segment from -1 to +1 as theta runs around. So the map flattens the unit circle into a thin line segment — a 'degenerate wing'. The interesting case is a circle that passes through the point z = 1 but is nudged slightly off-center and enlarged: its image is no longer a segment but a smooth, curved aerofoil shape, rounded at the front and tapering to a sharp point at the back.
Where does that sharp trailing edge come from? It is a conformal map failing at a single point. The derivative w' = (1/2)(1 - 1/z^2) vanishes at z = 1 and z = -1 — these are critical points, the one place a conformal map is allowed to break its angle-preserving promise. At a critical point angles get doubled, and that angle-doubling is precisely what folds the smooth circle into a corner. The cusp of the wing is manufactured by the map's critical point at z = 1, on purpose. One more honest caveat: the Joukowski map is two-to-one on the whole plane, since z and 1/z share the same image; to get a genuine conformal bijection you must restrict to the outside of the unit circle, which is exactly the region where the fluid actually flows.
Lift out of thin air: circulation and Kutta-Joukowski
Transplant the cylinder flow onto the wing and one number remains free: the circulation Gamma, the contour integral of the velocity once around the wing. It measures how much net swirl the flow carries around the aerofoil, and it is built from a vortex term added to the flow — the imaginary-logarithm potential from our catalogue. Different values of Gamma give different, all mathematically valid, flows around the same wing. Mathematics alone does not pick one out; physics must supply the missing condition.
That missing condition is the Kutta condition, and it is a piece of real physics smuggled in to fix the otherwise-arbitrary Gamma. Look back at the sharp trailing edge, born from the map's critical point. For a general Gamma the ideal-flow formula predicts the fluid whipping around that razor edge at infinite speed — physically absurd. The Kutta condition simply demands that the flow leave the trailing edge smoothly, with finite velocity, peeling off cleanly rather than wrapping around. There is exactly one value of the circulation Gamma that achieves this, and that single requirement pins it down. Nature chooses the circulation that makes the air leave the back of the wing gracefully.
With Gamma fixed, the lift drops out as a clean formula: the Kutta-Joukowski theorem says the lift force per unit span is L = rho U Gamma, where rho is the fluid density, U the free-stream speed, and Gamma the circulation just determined. Lift is literally circulation times speed times density — a contour integral made flesh. Stay honest about what this clean story leaves out: it is an idealization of an inviscid, incompressible, two-dimensional flow. Real air has viscosity and a thin boundary layer clinging to the wing; the Kutta condition is in fact a stand-in for the subtle viscous effect that genuinely sets the circulation, and the whole complex-potential picture ignores drag entirely. It predicts the lift beautifully and says nothing about the friction that resists the plane's motion.
What it grasps, and what it lets go
Step back and see the shape of the whole idea. A physical problem — steady, planar, swirl-free, source-free flow — was reformulated as a single holomorphic function, whereupon every tool from the earlier rungs became available at once: harmonic conjugates supplied the streamlines, the Cauchy-Riemann equations became conservation laws, conformal maps transplanted easy flows onto hard regions, and a critical point of a map carved the trailing edge that makes a wing a wing. The same machine works far beyond aerodynamics: the identical equations describe steady heat flow (with phi the temperature and psi the heat-flow lines) and two-dimensional electrostatics (with phi the voltage and psi the field lines). One holomorphic function, three physics courses.
And keep the limits in plain sight, because they are where students get burned. The method lives in two dimensions only; it needs the flow irrotational and source-free so the potential is harmonic; the Joukowski lift is an inviscid idealization that says nothing about drag; and the Kutta condition is physics, not mathematics, smuggled in to choose the circulation. Within those honest walls, though, the complex potential is one of the most satisfying applications in all of analysis — the place where the abstract rigidity of holomorphic functions reaches out and lifts a real aeroplane off the ground.