Conformal mapping turns difficult planar domains into easier ones while preserving local angle structure. Potential theory turns harmonic boundary data into interior fields. Together they form one of complex analysis’s most practical geometric engines.
This guide continues BTT’s one-variable complex-analysis route after holomorphicity, contour integration and analytic continuation. Its job is not to repeat the Cauchy–Riemann equations or residue calculus, but to show what those structures buy geometrically: local angle preservation, standard-domain normalisation, harmonic conjugacy, boundary-value reconstruction and complex-potential methods.
Prerequisites: Holomorphic Functions and Complex Differentiation, Contour Integration and Residues, and basic topology of connected and simply connected domains.
Reading route: local conformality → Möbius normalisation → disks and half-planes → Riemann mapping → harmonic functions → mean-value and maximum principles → Poisson kernel → Dirichlet problem → harmonic conjugates → complex potentials → electrostatic and flow analogies → Green functions → Schwarz reflection → boundary subtleties → verification workflow → practice and solutions.
1. Conformal means locally angle-preserving, not shape-preserving
A map is conformal at a point when it preserves oriented angles between sufficiently smooth curves through that point.
If f is holomorphic near z₀ and f'(z₀)≠0, then
f(z₀+h)=f(z₀)+f'(z₀)h+o(|h|).
Multiplication by the nonzero complex number f'(z₀) is exactly a rotation followed by a scaling. That is the local conformal model.
2. Conformality is local; global geometry can change dramatically
A conformal map may curve straight boundaries, enlarge one region and shrink another. It preserves infinitesimal angle structure, not Euclidean lengths or areas.
The exponential map sends horizontal strips to sectors or annuli. Möbius maps send lines and circles to generalized circles. The geometry can change globally even while local angles survive wherever f’≠0.
3. Möbius maps are the first normalisation tools
A Möbius transformation
T(z)=(az+b)/(cz+d), ad−bc≠0
is conformal wherever finite and nonsingular. It maps generalized circles to generalized circles and can move selected boundary points to standard positions.
The Cayley transform (z−i)/(z+i) maps the upper half-plane onto the unit disk.
4. Standard domains simplify boundary-value problems
Complex-analysis boundary problems are often easiest on the unit disk, upper half-plane, strip or annulus because their symmetries produce explicit kernels and Green functions.
The strategy is:
- map the physical or geometric domain to a standard domain;
- solve the harmonic or analytic problem there;
- compose with the inverse map;
- check boundary behaviour and singularities after transporting back.
5. The Riemann Mapping Theorem is a global normalisation theorem
A standard form states:
Every nonempty simply connected proper open subset Ω of C is conformally equivalent to the unit disk.
The exclusions matter. Ω cannot be all of C, and simple connectedness is essential. An annulus is not conformally equivalent to the disk because it has a hole.
6. Normalisation makes a Riemann map unique
The Riemann map is not unique because automorphisms of the disk can be composed with it.
If a∈Ω is fixed and one requires f(a)=0 together with f'(a)>0 real, then the conformal map from Ω to the disk is uniquely determined.
Normalisation removes the disk’s rotational freedom.
7. Multiply connected domains carry conformal moduli
An annulus r<|z|<1 cannot in general be flattened conformally to a disk without losing topology.
Two annuli are conformally equivalent only when their moduli agree, equivalently when the ratio of boundary radii matches after logarithmic normalisation.
This is an early example of a conformal invariant: topology and analytic geometry constrain the possible maps.
8. Harmonic functions are the real and imaginary parts of holomorphic functions locally
If f=u+iv is holomorphic, then
Δu=u_xx+u_yy=0, Δv=v_xx+v_yy=0.
Thus both u and v are harmonic. On simply connected domains, a harmonic function has a harmonic conjugate locally and, under standard conditions, globally up to an additive constant.
9. The mean-value property characterises harmonic averaging
If u is harmonic on a disk and continuous to the relevant circle, then its value at the centre equals the average of its values on every surrounding circle contained in the domain.
For a disk centred at a with radius r,
u(a)=1/(2π) ∫₀^{2π} u(a+re^{it})dt.
Harmonic values are therefore constrained by surrounding boundary data; they cannot create isolated interior spikes.
10. Maximum principles make boundary values decisive
A nonconstant harmonic function on a connected domain cannot attain an interior maximum or minimum.
For bounded domains with continuous boundary data, extrema occur on the boundary.
This principle gives uniqueness for many Dirichlet problems: two harmonic solutions with the same boundary values differ by a harmonic function vanishing on the boundary, hence vanish throughout the domain.
11. The Dirichlet problem asks for a harmonic interior matching prescribed boundary data
Given a domain Ω and boundary function g, the Dirichlet problem is to find u satisfying
- Δu=0 in Ω;
- u=g on the boundary, in the appropriate sense.
Existence is not automatic for every rough domain and every notion of boundary trace. But on the disk with continuous data, the Poisson integral gives an explicit solution.
12. The Poisson kernel reconstructs harmonic functions in the disk
For |z|=r<1 and boundary point e^{it}, the disk Poisson kernel is
P_r(θ−t)=(1−r²)/(1−2r cos(θ−t)+r²).
If g is continuous on the unit circle, then
u(re^{iθ})=1/(2π)∫₀^{2π} P_r(θ−t)g(e^{it})dt
is harmonic in the disk and approaches g at the boundary.
13. The Poisson kernel is a weighted boundary average
P_r is positive and its average over the circle is 1.
As r→1, the kernel concentrates near the boundary angle θ. Thus the interior value is a weighted average whose mass increasingly focuses near the nearest boundary point.
This gives a precise mathematical version of “boundary influence propagates inward”.
14. Worked Poisson example: one Fourier mode
Suppose boundary data on |z|=1 is g(e^{it})=cos(nt).
The harmonic extension is
u(re^{iθ})=r^n cos(nθ).
Similarly sin(nt) extends to r^n sin(nθ). Fourier modes are therefore eigenmodes of radial harmonic propagation in the disk.
15. Harmonic conjugates turn scalar potentials into analytic functions
On a simply connected domain, if u is harmonic, one may seek v satisfying
v_y=u_x, v_x=−u_y.
Then f=u+iv is holomorphic. Depending on the application, u may be interpreted as a potential while v acts as a stream function or conjugate coordinate.
16. Complex potentials encode two-dimensional irrotational flow
In idealised two-dimensional incompressible and irrotational flow, a complex potential
F(z)=φ(x,y)+iψ(x,y)
can combine velocity potential φ and stream function ψ.
Because φ and ψ are harmonic conjugates, streamlines and equipotential curves intersect orthogonally where the gradient is nonzero.
This is a mathematical model; physical adequacy still depends on the assumptions of incompressibility, irrotationality and the chosen boundary conditions.
17. Uniform flow has a linear complex potential
For uniform flow of speed U in the positive x-direction, take
F(z)=Uz.
Then φ=Ux and ψ=Uy. Equipotentials x=constant are vertical lines; streamlines y=constant are horizontal lines.
The derivative F'(z)=U gives the constant complex velocity representation, depending on convention.
18. Sources and sinks are logarithmic
A two-dimensional point source or sink is modelled by a logarithmic potential
F(z)=Q/(2π) Log z
on a chosen branch domain.
The multivalued argument is not an accident: the stream function winds around the source. This is a direct bridge back to branch structure and monodromy.
19. Conformal maps transport harmonic functions
If u is harmonic on a domain Ω’ and f:Ω→Ω’ is conformal, then u∘f is harmonic on Ω.
This conformal invariance of the planar Laplace equation explains why solving a boundary problem on a disk or half-plane can solve it on a more complicated simply connected domain after composition.
20. Boundary data must be transformed consistently
Mapping the domain is only half the problem. Boundary values, singularities, normal derivatives and physical interpretations must also be transported carefully.
A conformal map can magnify one part of the boundary strongly. Numerical schemes can become poorly conditioned near sharp corners or near points where the derivative of the map becomes large or small.
21. Green functions encode point-source boundary response
A Green function G(z,a) for a planar domain is harmonic in z away from a, has the logarithmic singularity appropriate to the Laplacian at a, and satisfies a specified boundary condition, commonly zero Dirichlet data.
In two dimensions the fundamental singularity is logarithmic, reflecting the same structure already seen in Log z.
22. Green functions can be built by conformal mapping
If f maps Ω conformally to the unit disk and sends a to 0, then a disk Green function can be pulled back through f.
One standard disk expression uses log|z| relative to the pole and a harmonic correction chosen to make the boundary value zero.
The exact formula depends on normalisation, but the structural idea is stable: singular local field + harmonic boundary correction.
23. Schwarz reflection extends analytic functions across symmetric boundaries
If a holomorphic function is defined on one side of a real-analytic boundary and satisfies a suitable reality condition on the boundary, it may extend analytically by reflection.
For the real axis, if f is holomorphic in the upper half-plane region, continuous to an interval, and real-valued on that interval, define the reflected extension by
F(z)=overline{f(overline{z})}
below the axis.
24. Corners affect boundary regularity and mapping derivatives
A conformal map can take a smooth domain to one with corners, but its derivative typically develops power-law behaviour near corner preimages.
Schwarz–Christoffel maps make this explicit for polygonal domains. Their derivatives contain factors whose exponents encode turning angles.
This matters both analytically and numerically: corners are where otherwise smooth boundary behaviour becomes singular or stiff.
25. Schwarz–Christoffel mapping sends the half-plane to polygons
A standard form is
f'(z)=C∏(z−x_k)^{α_k−1},
where x_k lie on the real axis and πα_k are the corresponding interior polygon angles.
Integrating f’ gives the map, up to affine constants. Determining the prevertices x_k is the parameter problem and can be numerically difficult.
26. Boundary correspondence is a separate theorem layer
The Riemann Mapping Theorem gives a conformal map between open domains, but extension to the boundary requires additional hypotheses.
For Jordan domains, Carathéodory’s theorem gives a homeomorphic extension to the boundary. Rougher domains can behave more subtly.
Never infer boundary regularity merely from interior conformality.
27. Harmonic measure formalises boundary influence
For a domain Ω and interior point z, harmonic measure assigns to a boundary set E the value at z of the harmonic function whose boundary data is 1 on E and 0 elsewhere, under suitable interpretations.
In the unit disk, harmonic measure is represented by the Poisson kernel. Under conformal maps, harmonic measure transforms naturally.
28. Potential theory links complex analysis to probability
For planar Brownian motion, harmonic measure can be interpreted as an exit probability: the probability that a path beginning at z first leaves the domain through a specified boundary set.
This probabilistic interpretation does not replace the analytic definition, but it reveals why harmonic functions are natural averages of boundary outcomes.
29. Uniqueness does not guarantee existence
Maximum principles often prove that a Dirichlet problem has at most one solution. Existence requires separate work.
On disks, Poisson integrals give existence for continuous boundary data. On general domains, Perron methods, Green functions or functional-analytic PDE methods may be needed.
Do not confuse a uniqueness theorem with a construction theorem.
30. Numerical conformal mapping needs independent validation
In practical work, conformal maps may be approximated numerically rather than expressed in closed form.
Useful checks include:
- boundary points map to the intended boundary;
- orientation is correct;
- known symmetry is preserved;
- the computed derivative does not vanish unexpectedly;
- cross-ratio or Möbius normalisation constraints are satisfied;
- forward and inverse maps approximately compose to identity;
- harmonic residuals are small when solving potential problems.
31. Common theorem-boundary errors
- Conformal does not mean length-preserving or area-preserving.
- Holomorphicity alone is not enough at a point where f’=0.
- The Riemann Mapping Theorem does not map every multiply connected domain to a disk.
- The theorem gives an interior conformal equivalence; boundary extension needs additional hypotheses.
- Maximum principles prove uniqueness or extremal behaviour, not existence by themselves.
- Branch cuts and singularities inherited by complex potentials must remain visible.
- Physical interpretations require the modelling assumptions to be stated separately from the mathematics.
32. A dependable conformal-mapping workflow
First classify the domain: simply connected or multiply connected, bounded or unbounded, smooth or cornered. Mark singularities and boundary conditions.
Choose the standard target domain whose kernel or potential solution is easiest. Normalise the conformal map so its remaining automorphism freedom is fixed.
Transport the boundary problem, solve it in the standard domain, compose back, then verify both the PDE and the original boundary data.
Finally return to the application: a mathematically valid harmonic solution may still rely on idealised physical assumptions.
33. Independent practice: twenty questions
- State a local condition making a holomorphic map conformal.
- What geometry does multiplication by a nonzero complex number perform?
- Give a Möbius map from the upper half-plane to the unit disk.
- State the Riemann Mapping Theorem.
- Why is C excluded from the theorem’s target-domain statement?
- What normalisation makes a disk-valued Riemann map unique?
- Why is an annulus not conformally equivalent to a disk?
- State the Laplace equation.
- State the mean-value property at the centre of a disk.
- State the maximum principle for nonconstant harmonic functions.
- What is the Dirichlet problem?
- Write the Poisson kernel for the unit disk.
- What is the harmonic extension of cos(nt) from the unit circle?
- How does a harmonic conjugate relate to a holomorphic function?
- Why do equipotentials and streamlines meet orthogonally in a nondegenerate complex potential?
- What does conformal invariance of the planar Laplacian mean?
- What role does a Green function play?
- State the idea of Schwarz reflection across the real axis.
- What extra theorem layer is needed to extend a Riemann map continuously to the boundary?
- List two numerical checks for an approximate conformal map.
34. Worked solutions and checks
1. f holomorphic near z₀ with f'(z₀)≠0.
2. A rotation and scaling.
3. (z−i)/(z+i).
4. Every nonempty simply connected proper open subset of C is conformally equivalent to the unit disk.
5. C is simply connected but not conformally equivalent to the bounded disk; Liouville-type arguments obstruct such a map.
6. Fix a∈Ω, require f(a)=0 and choose f'(a)>0 real.
7. It has nontrivial fundamental topology and a conformal modulus; the disk is simply connected.
8. u_xx+u_yy=0.
9. u(a)=1/(2π)∫₀^{2π}u(a+re^{it})dt.
10. A nonconstant harmonic function cannot attain an interior maximum or minimum.
11. Find a harmonic function in the domain matching prescribed boundary data.
12. (1−r²)/(1−2r cos θ+r²), up to the angular-difference notation.
13. r^n cos(nθ).
14. If u and v satisfy Cauchy–Riemann, u+iv is holomorphic; v is the harmonic conjugate of u.
15. The gradients are rotated by 90° through the Cauchy–Riemann equations where the derivative is nonzero.
16. Composing a harmonic function with a conformal map yields another harmonic function in the pulled-back domain.
17. It represents the response to a point singularity together with the chosen boundary condition.
18. Reflect by F(z)=overline{f(overline{z})} when f is real on the boundary interval and the theorem’s regularity hypotheses hold.
19. A boundary extension theorem such as Carathéodory’s for Jordan domains.
20. Examples: boundary correspondence, forward–inverse composition, derivative nonvanishing, symmetry, harmonic residual or normalisation checks.
35. Where this guide hands off
R26.06 changes scale. Instead of mapping domains, it studies functions such as Gamma, Beta and zeta together with asymptotic tools that extract dominant behaviour when exact formulas are difficult or impossible to use directly.
Sources and further study
For an undergraduate route through harmonic functions, complex potentials and conformal transformations, see MIT OpenCourseWare 18.04 Complex Variables with Applications and MIT Calculus Revisited: Conformal Mappings.