Scribd
Upload
63 views

Handout 6 Partial Differential Equations: Separation of Variables

This document outlines the six steps of the separation of variables technique for solving linear partial differential equations (PDEs) without mixed derivatives. It provides an example of using this technique to solve the one-dimensional heat equation. The solution involves assuming the form f(x,t)=X(x)T(t), substituting into the PDE, dividing by X(x)T(t), and solving the resulting ordinary differential equations (ODEs) for specific solutions that are combined to form the general solution. Boundary and initial conditions are then applied to determine constants in the solution.

Uploaded by

Paul Benedict
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
63 views

Handout 6 Partial Differential Equations: Separation of Variables

This document outlines the six steps of the separation of variables technique for solving linear partial differential equations (PDEs) without mixed derivatives. It provides an example of using this technique to solve the one-dimensional heat equation. The solution involves assuming the form f(x,t)=X(x)T(t), substituting into the PDE, dividing by X(x)T(t), and solving the resulting ordinary differential equations (ODEs) for specific solutions that are combined to form the general solution. Boundary and initial conditions are then applied to determine constants in the solution.

Uploaded by

Paul Benedict
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Handout 6 Partial Dierential Equations: separation of variables

This is a powerful technique for solving linear PDEs that have no mixed derivatives, i.e. nothing of the
form
2
f/xy.
There are 6 essential steps:
1. Assume the solution is going to be of the form X(x)T(t) or X(x)Y (y), etc. This is called separable
form.
2. Substitute that form back into the PDE.
3. Divide by X(x)T(t) or X(x)Y (y).
4. Now each term of the equation depends on a dierent variable so they must both be constants.
5. For each possible value of the constant (positive, negative, zero), solve the two resulting ODEs and
multiply the solutions together to give one specic solution to the PDE
6. Form the general solution of the PDE by adding linear combinations of all the specic solutions.
Example: Heat equation in one dimension
This equation governs the temperature f(x, t) in a thin uniform body of conductivity (thin enough
that temperature only varies along its length (not across the width):
f
t
=

2
f
x
2
.
Our process goes:
f(x, t) = X(x)T(t) X(x)T

(t) = X

(x)T(t)
T

(t)
T(t)
=
X

(x)
X(x)
= A.
Zero constant, A = 0
T

(t) = 0 so T(t) = A
1
and X

(x) = 0 so X(x) = B
1
x +C
1
, and multiplying these and renaming
the constants gives
f(x, t) = x +.
Negative constant, A =
2
T

(t) =
2
T(t) so T(t) = A
2
exp[
2
t].
X

(x) =
2
X(x) so X(x) = B
2
cos (x) +C
2
sin(x). Multiplying and renaming again, we get:
f(x, t) = exp[
2
t] (a cos (x) +b sin(x)) .
Positive constant, A =
2
T

(t) =
2
T(t) so T(t) = A
3
exp[
2
t].
X

(x) =
2
X(x) so X(x) = B
3
exp[x] +C
3
exp[x]. Putting them together, we have:
f(x, t) = exp[
2
t] (Aexp[x] +Bexp[x]) .
Note that the temperature here grows exponentially in time: these solutions are not physical!
General solution
f(x, t) = x + +

n
exp[
2
n
t] (a
n
cos (
n
x) +b
n
sin(
n
x))
+

n
exp[
2
n
t] (A
n
exp[
n
x] +B
n
exp[
n
x]) .
Real example: heat equation in a nite length bar with cold ends
Now suppose we have a bar of length L which is initially at temperature 1 all over, and which we cool
from both ends by holding the ends at temperature 0:
f(x, 0) = 1 f(0, t) = 0 f(L, t) = 0.
Left hand end Substituting x = 0 into our general solution, we get:
f(0, t) = 0 = +

n
a
n
exp[
2
n
t] +

n
(A
n
+B
n
) exp[
2
n
t]
and forcing this for every possible t gives b = 0, a
n
= 0 and A
n
+ B
n
= 0. Putting these back in
makes the full solution become:
f(x, t) = x +

n
b
n
exp[
2
n
t] sin(
n
x) +

n
A
n
exp[
2
n
t] (exp[
n
x] exp[
n
x]) .
Right hand end Next we look at the condition f = 0 at x = L. This gives:
0 = L +

n
b
n
exp[
2
n
t] sin(
n
L) +

n
A
n
exp[
2
n
t](exp[
n
L] exp[
n
L]).
Again, this has to be true for all values of t and the t-dependence of each term is dierent: so we
end up with = 0, A
n
= 0 and
n
L = n/L.
Now almost all the terms have disappeared and the full solution becomes:
f(x, t) =

n
b
n
exp[n
2

2
t/L
2
] sin(nx/L).
Initial condition Now we put in t = 0 and get:
1 =

n
b
n
sin

nx
L

which is a Fourier sine series with period 2L that weve seen before:
b
n
=
2
L

L
0
sin

nx
L

dx =
2
L

L
n
cos

nx
L

L
0
=
2
n
(1 cos (n)) =

0 n even
4
n
n odd
We can put this back into our general solution:
f(x, t) =

n odd
4
n
exp

n
2

2
t
L
2

sin

nx
L

.
Lets plot this function for a few values of t:
0
0.2
0.4
0.6
0.8
1

You might also like