Steady Two-Dimensional Heat Problems.

We shall now extend our discussion from one to two space dimensions and
consider the two-dimensional heat equation
 2
∂ u ∂2u

∂u
= c2 ∇2 u = c2 +
∂t ∂x2 ∂y 2
for steady (that is, time-independent) problems. Then ∂u/∂t = 0 and the
heat equation reduces to Laplace’s equation

Laplace Equation

Laplace’s Equation

∂2u ∂2u
+ 2 =0 (1)
∂x2 ∂y

uxx + uyy = 0 (2)

Auxx + Buxy + Cuyy = 0 (3)

A = 1, B = 0, C = 1
The discriminant ∆ is given by:

∆ = AC − B 2 = 1 − 0 > 0 Elliptic (4)

Let U (x, y) be a function defined as the product of two functions F (x) and
G(y):
U (x, y) = F (x)G(y) (5)

1
 The second derivatives yield:

F ′′ G + F G′′ = 0 (6)

Separation of variables gives:

F ′′ G′′
=− (7)
F G
This process is known as separation of variables.
Example

∂2u ∂2u
∇2 u = + 2 =0
∂x2 ∂y
Consider the function u(x, y) such that:

u(x, y) = F (x)G(y) (8)

The partial derivatives are given by:

uxx = F ′′ (x)G(y) (9)

′′
uyy = F (x)G (y) (10)

The product of these derivatives leads to a separable differential equation:

F ′′ (x)G(y) + F (x)G′′ (y) = 0 (11)

Dividing by F (x)G(y) and rearranging, we get:

F ′′ G′′
=− = −k 2 (12)
F G
Solving these ordinary differential equations, we have:

F ′′ + k 2 F = 0, λ = ±ik (13)

The solutions to these equations are:

F = A cos(kx) + B sin(kx) (14)

Boundary conditions give us:

F (0) = A = 0 (15)
F (a) = B sin(ka) = 0, B ̸= 0 (16)
nπ
k= , n = 1, 2, . . . , ∞ (17)
a
Therefore, the function F (x) simplifies to:
 nπx 
F = B sin (18)
a

2
 G′′ − k 2 G = 0, λ1,2 = ±k (19)

G = c1 eky + c2 e−ky (20)

For G(y), applying the boundary conditions:
G(0) = c1 + c2 = 0, c1 = −c2 (21)
ky −ky


G(y) = c1 e −e (22)
= 2c1 sinh(ky) (23)
 nπy 
= 2c1 sin (for the corresponding k value) (24)
a
Consider the solution to a partial differential equation given by the series:
∞
X
u(x, y) = un (x, y) (25)
n=1

where each term in the series is defined as:

 nπx   nπy 
un (x, y) = An sin sinh (26)
a a
The function u(x, y) at y = b is matched to a boundary condition f (x):
∞  nπx   
X nπb
u(x, y) = An sin sinh = f (x) (27)
n=1
a a
To find the coefficients An , we use the orthogonality of sine functions and
integrate with respect to x:
2 a
  Z
nπb  nπx 
An sinh = f (x) sin dx (28)
a a 0 a
Solving for An , we get:
Z a
2  nπx 
An = nπb

f (x) sin dx (29)
a sinh a 0 a
Example

3
 ∂2u ∂2u
∇2 u = + 2 =0
∂x2 ∂y

T (x, 0) = 0
T (0, y) = 0
T (x, 1) = x
T (1, y) = y

The temperature distribution in a rectangular plate can be described by

T (x, y) and is given by the product of two functions F (x) and G(y):

T (x, y) = F (x)G(y) (30)

Where the second derivatives satisfy the equation:

F ′′ G + F G′′ = 0 (31)

From separation of variables we have:

F ′′ G′′
=− = −k 2 (32)
F G
This leads to two ordinary differential equations:

F ′′ + k 2 F = 0 and G′′ − k 2 G = 0 (33)

When k = 0, they could be same function

F ′′ = 0, F = Ax + B (34)
G′′ = 0 and G = Cy + D (35)
T (x, y) = (Ax + B)(Cy + D) (36)
Given the boundary conditions:

T (x, 0) = (Ax + B)D = 0

T (0, y) = B(Cy + D) = 0
T (x, 1) = (Ax + B)(C + D) = x
T (1, y) = (A + B)(Cy + D) = y

if B = D = 0, conditions are satisfied, AC = 1

The solution to the temperature distribution problem, given the boundary
conditions, is:
T (x, y) = xy (37)

4
 Example

∂2u ∂2u
+ 2 =0 (38)
∂x2 ∂y
u(x, y) = F (x)G(y) (39)
Given that:
F ′′ G′′
=− = −k 2 , (40)
F G
we have the following ordinary differential equations:

F ′′ + k 2 F = 0 and G′′ − k 2 G = 0. (41)

The boundary conditions are:

∂u
= F ′ (0)G(y) = 0, (42)
∂x x=0

which implies that F ′ (0) = 0 and F ′ (a) = 0.

The general solution for F is:

F = A cos(kx) + B sin(kx), (43)

and its derivative is:

F ′ = −Ak sin(kx) + Bk cos(kx). (44)

From the boundary conditions we obtain:

F ′ (0) = Bk = 0, B = 0 and F ′ (a) = −Ak sin(ka) = 0, (45)

with the nontrivial solution A ̸= 0 and ka = nπ, n = 0, 1, 2, . . ..

For G we have:

G′′ − k 2 G = 0 with λ = ±k, λ > 0, (46)

and the general solution is:

G = c1 eky + c2 e−ky . (47)

5
 Using the boundary conditions for G we find:

G(0) = c1 + c2 and c1 = −c2 = c, (48)

leading to the solution:

G = 2c sinh(ky). (49)
For the special case where k = 0, we have:

G′′ = 0 with G0 = Dy + E, (50)

and thus:
u0 = (Dy + E)A (if k = 0, F = A). (51)
The eigenfunctions un are:
nπ
un = Bn cos (kx) sinh (ky) with k = . (52)
a
The full solution is given by the series:
∞
X
u(x, y) = un (x, y), (53)
n=0

which expands to:

∞
X  nπx   nπy 
u(x, y) = B0 y + Bn cos sinh . (54)
n=1
a a

At y = b, we have:
∞  nπx   
X nπb
u(x, b) = B0 b + Bn cos sinh = f (x). (55)
n=1
a a

The coefficients are determined by:

1 a
Z
B0 = f (x)dx, (56)
a 0
Z a
1
B0 = f (x)dx, (57)
ab 0
Z a
2  nπx 
Bn = f (x) cos dx. (58)
a sinh(nπb/a) 0 a

6
Example

7
 The general solution for Laplace’s equation ∇2 u = 0 is assumed to be:
Given the boundary conditions:

u(1, y) = x u(x, 1) = y
u(x, 0) = 1 − x u(0, y) = 1 − y

All boundary conditions are nonhomogeneous. We are going to assume the

solution polynomial.
u = a + bx + cy + dxy (59)
Applying the boundary conditions u(0, y) = a + cy = 1 − y gives a = 1 and
c = −1.
Applying the boundary conditions u(x, 0) = a + bx = 1 − x gives b = -1.
Solving the system:

a + bx + c + dx = x b+d=1 d=2

The final solution, incorporating all boundary conditions, is:

u = 1 − x − y + 2xy (60)

Laplace’s Equation in Cylindrical Coordinates:

8
 The Laplace’s equation in Cartesian coordinates uxx +uyy = 0 is transformed
into polar coordinates where x = r cos θ and y = r sin θ. The relation between
Cartesian and cylindrical coordinates is given by:

x2 + y 2 = r 2 (61)

2x = 2rrx (62)
The partial derivatives transform as:

ux = ur rx + uθ θx (63)

The second-order partial derivatives in polar coordinates are:

x y
rx = θx = −
r r2
y2 2xy
rxx = 3 θxx = 4
r r
The Laplace’s equation in polar coordinates is given by:
1 1
∇2 u = urr + ur + 2 uθθ = 0 (64)
r r
We start with the relations in polar coordinates x = r cos(θ) and y = r sin(θ).
The partial derivative of a function u with respect to x in polar coordinates can
be found using the chain rule:
∂u ∂r ∂u ∂θ
ux = + (65)
∂r ∂x ∂θ ∂x
p For rx and θx , we differentiate r and θ with respect to x considering r =
x2 + y 2 and θ = arctan xy :

∂ p 2 x x
rx = x + y2 = p =
∂x 2
x +y 2 r
∂ y y y
θx = arctan =− 2 2
=− 2
∂x x x +y r
The second-order derivatives rxx and θxx are computed by differentiating
rx and θx with respect to x. By applying the product rule and differentiating
implicitly, we obtain:

!
∂ x ∂ x y2 y2
rxx = = = 2 =
r3
p
∂x r ∂x x2 + y 2 (x + y 2 )3/2
 
∂  y ∂ y 2xy 2xy
θxx = − 2 =− = 2 = 4
∂x r ∂x x2 + y 2 (x + y 2 )2 r

9
 Laplace’s equation in Cartesian coordinates is given by uxx + uyy = 0. To
transform this into polar coordinates, where x = r cos(θ) and y = r sin(θ), we
use the chain rule to express the second derivatives uxx and uyy in terms of r
and θ.
The first-order partial derivatives in terms of polar coordinates are:
∂u ∂r ∂u ∂θ uθ
ux = + = ur cos(θ) − sin(θ), (66)
∂r ∂x ∂θ ∂x r
∂u ∂r ∂u ∂θ uθ
uy = + = ur sin(θ) + cos(θ). (67)
∂r ∂y ∂θ ∂y r
The second-order partial derivatives become:
∂ uθ
uxx = (ur cos(θ) − sin(θ)), (68)
∂x r
∂ uθ
uyy = (ur sin(θ) + cos(θ)). (69)
∂y r
After some simplification, using the product rule and trigonometric identi-
ties, we combine these to express the Laplacian in polar coordinates:
1 1
∇2 u = urr + ur + 2 uθθ = 0. (70)
r r
This is the form of Laplace’s equation in polar coordinates.
Drum Wave Equation in Polar Coordinates:

if the problem has radial symmetry and does not depend on the angular vari-
able (), as in the case of a perfectly circular drumhead with radially symmetric
boundary conditions, there’s no need to consider variations due to the () term.
The two-dimensional wave equation utt = c2 ∇2 u can be expressed in polar
coordinates for a drumhead as:
 
2 1
utt = c urr + ur (71)
r

10
 Assuming a solution of the form u = W (r)G(t), we get:
 
1
W G̈ = c2 W ′′ G + W ′ G (72)
r

Separating variables and dividing by W G, we obtain:

G̈ W ′′ 1 W′
= + = −k 2 (73)
c2 G W r W
This leads to two ordinary differential equations:
1
W ′′ + W ′ + k 2 W = 0 (74)
r
The solution to the radial part of the wave equation can be expressed as an
extended power series in terms of r.

Transform Methods
For a function u(x, t) = f (x − at), its Fourier transform is denoted as û(ω) and
is defined as: Z ∞
1
ĝ = F{g} = √ g(x)e−iωx dx (75)
2π −∞
Example Given the wave equation:

ut + aux = 0 (76)

Applying the Fourier transform (F) in the spatial variable x:

F(ut ) + aF(ux ) = 0 (77)

This transforms to:

F(u)t + aiωF(u) = 0 (78)
We then have the ordinary differential equation in the frequency domain:

ût + aiωû = 0 (79)

Which simplifies to:

λ + aiω = 0, λ = −aiω (80)
The solution is then given by:

û = ce−aiωt , û = fˆe−aiωt (81)

The inverse Fourier transform is given by:

Z ∞
1
g(x) = F −1 {ĝ} = √ ĝ(ω)eiωx dω (82)
2π −∞

11
 The solution u in terms of the inverse Fourier transform is given by:
Z ∞
1
u= √ fˆe−iωt eiωx dω (83)
2π −∞
where û is the Fourier transform of u.
Z ∞
1
u= √ fˆeiω(x−at) dω (84)
2π −∞
Z ∞
1
u= √ fˆeiωξ dω = f (ξ) (85)
2π −∞
Here, ξ = x − at represents a change of variables or a phase shift.
Example
Simple Wave Equation The wave equation is given by:

ut + aux = 0 (86)

With the initial condition:

u(x, 0) = f (x) (87)

The definition of the Laplace transform:

Z ∞
L{g} = g(t)e−st dt = ĝ (88)
0

Applying the transform to a differential equation:

L{ut } + aL{ux } = 0 (89)

This leads to the algebraic equation:

sû − f (x) + aûx = 0 (90)

aûx + sû = f (x) (91)

Solving for û we find:
s
aλ + s = 0, λ=− (92)
a
The variation of parameters gives us:
s
f (x)e a x
Z
s s
û = ce− a x + e− a x dx (93)
a

12
 Example

Given the heat equation

∂u ∂2u
= , (94)
∂t ∂x2
with the initial condition 2
u(x, 0) = e−x , (95)
the initial temperature distribution is a Gaussian.
2
The Fourier transform of e−ax is
2
2 e−ω /4a
F (e−ax ) = √ , a > 0. (96)
2a
Applying the Fourier transform to both sides of the heat equation, we get

F (ut ) = F (uxx ), (97)

leading to the ordinary differential equation

ût = −ω 2 û. (98)

Solving for Û , we have

ût + ω 2 û = 0. (99)
2
λ + ω = 0. (100)
The solution is 2
û = ce−ω t , (101)
where c is a constant determined by the initial condition,
2
e−ω /4
F (u(x, 0)) = √ , (102)
2
so 2
e−ω /4
c= √ . (103)
2
Thus,
2 2
e−ω /4 2 e−ω (1/4+t)
û = √ · e−ω t
= √ . (104)
2 2

13
 2 1
√ e−ω (t+ a ) √ 2
û = a √ → ae−ax
2a
For the inverse Fourier transform, we have
1 1 1
= + t, a= , (105)
4a 4 4t + 1
and thus the solution to the initial heat equation is
1 x2
u= √ e− 4t+1 . (106)
4t + 1
The method of separation of variables is not applicable to irregular geome-
tries because it relies on the ability to separate the solution into functions of
individual coordinates, which is only feasible when the problem exhibits certain
symmetries.

14