If is a harmonic function on a simply connected domain , a harmonic conjugate of is a harmonic function such that

is analytic on .

Equivalently, and satisfy the Cauchy–Riemann equations and .

The harmonic conjugate exists (on simply connected domains) and is unique up to an additive constant. The construction is the same recipe as finding a potential function for a conservative field.

Recipe

Given harmonic :

  1. From , integrate w.r.t. : .
  2. From , get an equation for .
  3. Integrate to find .

Worked example

.

Check harmonic: , , sum . ✓

Find : , integrate: . Differentiate: , set equal to : , . So .

Combined: .

So is the real part of .

Why it exists on simply connected domains

The C–R equations , define . This vector field has zero curl: (using harmonic). So is conservative on simply connected , and there’s a potential function with . That is the harmonic conjugate.

On non-simply-connected domains, the C–R equations may not admit a single-valued . The example: on . The “conjugate” is the polar angle , but it’s multi-valued — going around the origin adds . Multivaluedness comes from non-simply-connectedness. This is the same phenomenon as the complex logarithm being multi-valued on .

Geometric content

and its conjugate have:

  • Mutually orthogonal level curves. The level sets and meet at right angles wherever they cross. See Conformality for the proof.
  • Mean-value, max-principle properties inherited from being harmonic.

The combined is analytic, so all the rigid properties of analytic functions apply.

In context

The harmonic conjugate is what lets you recover a complex-analytic function from its real part alone. In physics applications: given a steady-state potential (say, electrostatic potential in 2D), constructing its harmonic conjugate gives a stream function — the level curves of are the field lines of . The complex potential packages both into a single analytic function.

This is the bridge between physical 2D potentials and the complex-analytic machinery (Cauchy’s theorem, conformal mapping).