Laplace’s equation in polar coordinates

Boundary value problem for disk:

ur uθθ
∆u = urr + + 2 = 0, u(a, θ) = h(θ).
r r
Laplace’s equation in polar coordinates

Boundary value problem for disk:

ur uθθ
∆u = urr + + 2 = 0, u(a, θ) = h(θ).
r r

Separating variables u = R(r )Θ(θ) gives

R 00 Θ + r −1 R 0 Θ + r −2 RΘ00 = 0 or
Θ00 −r 2 R 00 − rR 0
= = −λ.
Θ R
Laplace’s equation in polar coordinates

Boundary value problem for disk:

ur uθθ
∆u = urr + + 2 = 0, u(a, θ) = h(θ).
r r

Separating variables u = R(r )Θ(θ) gives

R 00 Θ + r −1 R 0 Θ + r −2 RΘ00 = 0 or
Θ00 −r 2 R 00 − rR 0
= = −λ.
Θ R

Eigenvalue problem
Θ00 + λΘ = 0, Θ(0) = Θ(2π), Θ0 (0) = Θ0 (2π).
Laplace’s equation in polar coordinates

Boundary value problem for disk:

ur uθθ
∆u = urr + + 2 = 0, u(a, θ) = h(θ).
r r

Separating variables u = R(r )Θ(θ) gives

R 00 Θ + r −1 R 0 Θ + r −2 RΘ00 = 0 or
Θ00 −r 2 R 00 − rR 0
= = −λ.
Θ R

Eigenvalue problem
Θ00 + λΘ = 0, Θ(0) = Θ(2π), Θ0 (0) = Θ0 (2π).

√
Can write ODE solution
√ as Θ = exp(±i λθ), which is 2π
periodic only when λ is a positive integer n. If λ = 0, the
only periodic solution is a constant. λ < 0 gives growing
exponentials.
Laplace’s equation in polar coordinates, cont.

Eigenfunctions (”circular harmonics”)


1
 λ=0
Θ = cos(nx) λ = n2

sin(nx) λ = n2 ,


where n = 1, 2, 3, . . ..
Laplace’s equation in polar coordinates, cont.

Eigenfunctions (”circular harmonics”)


1
 λ=0
Θ = cos(nx) λ = n2

sin(nx) λ = n2 ,


where n = 1, 2, 3, . . ..
Equation for radial component is Euler equation
r 2 R 00 + rR 0 − λR = 0.
Laplace’s equation in polar coordinates, cont.

Eigenfunctions (”circular harmonics”)


1
 λ=0
Θ = cos(nx) λ = n2

sin(nx) λ = n2 ,


where n = 1, 2, 3, . . ..
Equation for radial component is Euler equation
r 2 R 00 + rR 0 − λR = 0.
Solutions are just powers R = r α ; plugging
√ in,
[α(α − 1) + α − λ]r α = 0 or α = ± λ.
Laplace’s equation in polar coordinates, cont.

Eigenfunctions (”circular harmonics”)


1
 λ=0
Θ = cos(nx) λ = n2

sin(nx) λ = n2 ,


where n = 1, 2, 3, . . ..
Equation for radial component is Euler equation
r 2 R 00 + rR 0 − λR = 0.
Solutions are just powers R = r α ; plugging
√ in,
[α(α − 1) + α − λ]r α = 0 or α = ± λ.
If λ = 0, get linearly independent solutions 1 and ln r .
Laplace’s equation in polar coordinates, cont.

Eigenfunctions (”circular harmonics”)


1
 λ=0
Θ = cos(nx) λ = n2

sin(nx) λ = n2 ,


where n = 1, 2, 3, . . ..
Equation for radial component is Euler equation
r 2 R 00 + rR 0 − λR = 0.
Solutions are just powers R = r α ; plugging
√ in,
[α(α − 1) + α − λ]r α = 0 or α = ± λ.
If λ = 0, get linearly independent solutions 1 and ln r .
Reject (for now) solutions involving ln r and r −α .
Laplace’s equation in polar coordinates, cont.

Superposition of separated solutions:

∞
X
u = A0 /2 + r n [An cos(nθ) + Bn sin(nθ)].
n=1
Laplace’s equation in polar coordinates, cont.

Superposition of separated solutions:

∞
X
u = A0 /2 + r n [An cos(nθ) + Bn sin(nθ)].
n=1

Satisfy boundary condition at r = a,

∞
X
h(θ) = A0 /2 + an [An cos(nθ) + Bn sin(nθ)].
n=1
Laplace’s equation in polar coordinates, cont.

Superposition of separated solutions:

∞
X
u = A0 /2 + r n [An cos(nθ) + Bn sin(nθ)].
n=1

Satisfy boundary condition at r = a,

∞
X
h(θ) = A0 /2 + an [An cos(nθ) + Bn sin(nθ)].
n=1

This is a Fourier series with cosine coefficients an An and

sine coefficients an Bn , so that (using the known formulas)
Z 2π Z 2π
1 1
An = h(φ) cos(nφ)dφ, Bn = h(φ) sin(nφ)dφ.
πan 0 πan 0
Poisson formula

Inserting the Fourier coefficient formulas into the general

solution,
Z 2π
1
u(r , θ) = h(φ)dφ
2π 0
∞ Z 2π
X rn
+ h(φ)[cos(nφ) cos(nθ) + sin(nφ) sin(nθ)]dφ.
πan 0
n=1
Poisson formula

Inserting the Fourier coefficient formulas into the general

solution,
Z 2π
1
u(r , θ) = h(φ)dφ
2π 0
∞ Z 2π
X rn
+ h(φ)[cos(nφ) cos(nθ) + sin(nφ) sin(nθ)]dφ.
πan 0
n=1

Use the identity

cos(nφ) cos(nθ) + sin(nφ) sin(nθ) = cos(n(θ − φ)), and reverse
the order of summation and integration
Z 2π ∞  n
1 n X r o
u(r , θ) = h(φ) 1 + 2 cos(n(θ − φ)) dφ.
2π 0 a
n=1
Poisson formula, cont.

Sum is geometric series in disguise:

∞  n ∞
!n
X r X rei(θ−φ)
1+2 cos(n(θ − φ)) = 1 + 2Re
a a
n=1 n=1
rei(θ−φ)
= 1 + 2Re
a − rei(θ−φ)
a2 − r 2
= 2 .
a − 2ar cos(θ − φ) + r 2
Poisson formula, cont.

Sum is geometric series in disguise:

∞  n ∞
!n
X r X rei(θ−φ)
1+2 cos(n(θ − φ)) = 1 + 2Re
a a
n=1 n=1
rei(θ−φ)
= 1 + 2Re
a − rei(θ−φ)
a2 − r 2
= 2 .
a − 2ar cos(θ − φ) + r 2

This results in Poisson’s formula:

Z 2π
1 a2 − r 2
u(r , θ) = P(r , θ−φ)h(φ)dφ, P(r , θ) = .
0 2π a2 − 2ar cos(θ) + r 2
Consequences of the Poisson formula

At r = 0, notice the integral is easy to compute:

Z 2π Z 2π
1 1
u(r , θ) = h(φ)dφ, = u(a, φ)dφ.
2π 0 2π 0

Therefore if ∆u = 0, the value of u at any point is just the

average values of u on a circle centered on that point.
(“Mean value theorem")
The maximum and minimum values of u are therefore
always on the domain boundary (this is true for any shape
domain).