GREEN’S FUNCTIONS

D R K M U D AYA N A N D A N , N A S C O L L E G E
Introduction

Green’s functions are named after the British mathematician George

Green, who developed the concept in the 1830s. Green’s function
methods enable the solution of a differential equation containing an
inhomogeneous term (often called a source term) to be related to
an integral operator. It can be used to solve both partial and exact
differential equations.

George Green
He (14 July 1793 to 31 May 1841) was a British mathematical physicist
who wrote Figure 1: This are the writings on
Green’s tomb
’ An Essay on the Application of Mathematical Analysis to the
Theories of Electricity and Magnetism’. The essay introduced several
important concepts, among them a theorem similar to the modern
Green’s theorem, the idea of potential functions as currently used in
physics,and the concept of what are now called Green’s functions.
Green was the first person to create a mathematical theory of
electricity and magnetism and his theory formed the foundation for
the work of other scientists such as James Clerk Maxwell, William
Thomson, and others.

Simple homogeneous differential equations

Consider the differential equation

d2 y Figure 2: George Green

=0
dx2
This can be solved very easily and we will get the solution as

y = Ax + B

which is the equation for a straight line. The constants can be found
if boundary conditions are given. Similarly consider another homoge-
neous equation
d2 y
+ k2 y = 0 Figure 3: George Green’s house and his
dx2 father’s mill
4

This can be solved to get

y = A sin k x + B cos kx

. Thus there are simple techniques available to solve homogeneous

equations. But if we replace them with source terms like

d2 y
= ln x
dx2
d2 y
+ k2 y = tan x
dx2
then the problem become difficult to solve. Before thinking of solving
such nonhomogeneous equations let us look at different types of
differential operators.

Sturm Liouville operator

Sturm Liouville operator is the most general form of second order

differential operator which can be written in the equation form as
 
d dy
Ly = p( x ) + q( x )y = 0
dx dx

For
d2 y
=0
dx2
p( x ) = 1 and q( x ) = 0 and for

d2 y
+ k2 y = 0
dx2
p( x ) = 1 and q( x ) = k2 . Any differential operator can be changed
into SL operator form.

Dirac delta function

While studying GF techniques we will encounter some properties of

Dirac delta function. They are
Z
δ( x − t)dx = 1
allspace
Z
δ( x − t) f (t)dt = f ( x )
Greens function technique

Suppose SL operator operating on a function y( x ) gives as

L y( x ) = f ( x ) (1)

which is a non homogeneous equation. To solve this NHE let us define

L G ( x, t) = δ( x − t) (2)
R
so that we can show that if we define y( x ) = G ( x, t) f (t)dt we will get equation (1). The proof of this
argument is given below
Proof:
Z
L y( x ) = L G ( x, t) f (t)dt

Interchanging integral and differential

Z
= L G ( x, t) f ( x )dt

Using the definition of Greens function

Z
= δ( x − t) f (t)dt
R
and using the property Dirac delta function that δ( x − t) f (t)dt = f ( x ) we get

L y( x ) = f ( x )

Definition

Generally speaking, a Green’s function is an integral kernel that can be used to solve differential equations
from a large number of families including simpler examples such as ordinary differential equations with
initial or boundary value conditions, as well as more difficult examples such as inhomogeneous partial
differential equations (PDE) with boundary conditions.

One dimensional Greens function and its properties

So let us start with

L G ( x, t) = δ( x − t)
6

Taking the SL operator  

d d
p( x ) G ( x, t) + q( x ) G ( x, t) = δ( x − t)
dx dx
Integrating over x for a small interval t − e to t + e
t+e   t+e t+e
d d
Z Z Z
p( x ) G ( x, t) dx + q( x ) G ( x, t)dx = δ( x − t)dx
dx dx
t−e t−e t−e

e is a small quantity. Hence RHS is 1. Taking the second part as zero

dG dG
p(t + e) − p(t − e) =1
dx t+e dx t−e
In the limit e −→ 0  
dG dG
p(t) − =1
dx t+e dx t−e
dG dG 1
− =
dx t+e dx t−e p(t)
dG2 dG1 1
− =
dx dx p(t)
This property shows that the values of GF must be different for x less tha t and x greater than t. So let label
GF before t as G1 ( x, t) and GF after t as G2 ( x, t). We had taken the second integral as zero which means
that
G2 ( x, t + e) − G1 ( x, t − e) = 0

At x = t G1 = G2 or Greens function is
1. Continuous at boundary and
2. Derivative of the Greens function is discontinuous.
These are the two properties of one dimensional Green’s function.

Form of Greens function

Next is to find G1 and G2 . Assume

G1 ( x, t) = C1 u1 ( x )

and
G2 ( x, t) = C2 u2 ( x )

where C1 and C2 which are functions of t are to be determined. The Greens functions are determined using
the two properties we got. The continuity of Greens function demands that

C2 u2 (t) − C1 u1 (t) = 0

Discontinuity of Greens function demands that

1
C2 u20 (t) − C1 u10 (t) =
p(t)
 7

Multiplying the first equation by u20 (t) and second by u10 (t)

C2 u20 (t)u2 (t) − C1 u20 (t)u1 (t) = 0

−1
C2 u20 (t)u2 (t) − C1 u10 (t)u2 (t) = u2 ( t )
p(t)
Subtracting gives
u2 ( t )
C1 u10 (t)u2 (t) − C1 u20 (t)u1 (t) =
p(t)
u2 ( t )
C1 (u2 u10 − u1 u20 ) =
p(t)
If W = u1 u20 − u2 u10 , (is also called Wronskian) Then

u2 ( t )
C1 =
W p(t)

u1 ( t )
C2 =
W p(t)
Hence
u1 ( x ) u2 ( t )
G1 ( x, t) =
W p(t)
u2 ( x ) u1 ( t )
G2 ( x, t) =
W p(t)
Then we get the solution as
Zt Zb
y( x ) = G1 ( x, t) f (t)dt + G2 ( x, t) f (t)dt
a t

Solved Problems

Case 1 Finite initial and final boundary values given

d2
1. Derive the Green’s function for the operator dx2
with the boundary conditions y(0) = 0 and y(1) = 0.
Solution Here it is given that
d2 y
= f (x)
dx2
For the homogeneous equation
d2 y
=0
dx2
d dy dy
dx dx = 0, dx = constant. Integrating
y = Ax + B

First bc implies y(0) = 0 ⇒ B = 0 u1 ( x ) = A x, u1 (t) = A t, u10 ( x ) = A. Second bc implies y(1) = 0 ⇒

0 = A + B, B = − A, u2 ( x ) = A x − A, u2 (t) = A t − A, u20 ( x ) = A Then Wronskian,

W = u1 (t)u20 (t) − u10 (t)u2 (t) = A2

8

For x < t
u1 ( x ) u2 ( t )
G1 ( x, t) =
A
G1 ( x, t) = x (t − 1)
For x > t
u2 ( x ) u1 ( t )
G2 ( x, t) =
A
G2 ( x, t) = t( x − 1)
d2
2. Derive the Green’s function for the operator dx2
with the boundary conditions y(0) = 0 and y( a) = 0.
Solution

d2 y
=0
dx2
y = Ax + B
y (0) = 0 ⇒
B=0
u1 ( x ) = A x
u1 ( t ) = A t
u10 ( x ) = A
y( a) = 0 ⇒
0 = Aa + B
B = − Aa
u2 ( x ) = A x − A a
u2 ( t ) = A t − A a
u20 ( x ) = A
Then Wronskian,
W = u1 u20 − u10 u2
= A1 xA2 − A1 ( A2 x − A2 a)
= A1 A2 x − A1 A2 x + A1 A2 a
W = A2 a
For x < t
u1 ( x ) u2 ( t )
G1 ( x, t) =
W p(t)
x (t − a)
G1 ( x, t) =
a
For x > t
u2 ( x ) u1 ( t )
G2 ( x, t) =
W p(t)
t( x − a)
G2 ( x, t) =
a
 9

d2
3. Obtain the Green’s function for the operator dx2
corresponding to the boundary conditions y(0) = 0 ;
y0 ( a) = 0.
Answer:

d2 y
=0
dx2
It’s solutions is,
y = Ax + B
The first BC gives y(0) = 0 implies 0 = A × 0 + B and hence B = 0.

u1 ( x ) = A x

u1 ( t ) = A t
u10 ( x ) = A
The second BC gives y( a) = 0 implies
A=0
u2 ( x ) = B
u2 ( t ) = B
u20 ( x ) = 0
Then Wronskian,
W = u1 u20 − u10 u2
W = − AB
For x < t
u1 ( x ) u2 ( t )
G1 ( x, t) =
W p(t)
G1 ( x, t) = − x
For x > t
u2 ( x ) u1 ( t )
G2 ( x, t) =
W p(t)
G2 ( x, t) = −t
d2
4. Obtain the Green’s function for the operator dx2
corresponding to the boundary conditions y(0) = 0 ;
y0 (1) = 0.
Answer:
Solution is same as in the above problem with same answer
d2 y
5. Deduce Green’s function of the operator ( dx2 + k2 ) with boundary condition y(0) = 0 , y( L) = 0

d2 y
+ k2 y = f ( x )
dx2
its solution is
y = A sin kx + B cos kx
10

u1 ( x ) is defined as the value of y at applying first boundary condition and u2 ( x ) is defined as the value
of y at applying second boundary condition.
y (0) = 0 ⇒

B=0
u1 ( x ) = A sin kx
u1 (t) = A sin kt
u10 ( x ) = A k cos kx
Similarly y( L) = 0 ⇒

0 = A sin kL + B cos kL
− A sin kL
B=
cos kL
sin kx cos kL − sin kL cos kx
 
a2 cos kL
u2 ( x ) = A sin kx − cos kx = A
cos kL cos kL
sin k ( x − L)
u2 ( x ) = A
cos kL
sin k(t − L)
u2 ( t ) = A
cos kL
k cos k( x − L)
u20 (t) = A
cos kL
A2 k
Then Wronskian,W = cos kL (sin kL) For x < t

sin kx sin k (t − L)
G1 ( x, t) =
k sin kL
For x > t
sin k ( x − L) sin kt
G2 ( x, t) =
k sin kL
6. Find an appropriate Green’s function for the equation y00 + 1/4y = f ( x ) with boundary condition
y(0) = y(π ) = 0.
d2 y 1
+ y=0
dx2 4
2  2
d y 1
+ y=0
dx2 2
So it’s solution is
1 1
y = A sin x + B cos x
2 2
y (0) = 0 ⇒
B=0
1
u1 ( x ) = A sin x
2
1
u1 (t) = A sin t
2
 11

1 1
u10 ( x ) = A cos x
2 2
y(π ) = 0 ⇒
A=0
1
u2 ( x ) = B cos x
2
1
u2 (t) = B cos t
2
0 1 1
u2 ( x ) = − B sin x
2 2
Then Wronskian,
W = u1 u20 − u10 u2
− AB
W=
2
For x < t
1 1
G1 ( x, t) = −2 sin x cos t
2 2
For x > t
1 1
G2 ( x, t) = −2 cos x sin t
2 2
Case 2: Initial bc given, final bc not given

Solved Problem
d2
1. Derive the Green’s function for the differential equation dx2
= 0 with the boundary conditions y(0) =
0 = y 0 (0).
Solution:

d2 y
=0
dx2
Its solution is
y = Ax + B
y (0) = 0 ⇒
B=0
Then u1 ( x ) = A x, u1 (t) = A t, u10 ( x ) = A. y0 (0) = 0 ⇒

u2 ( x ) = A

u2 ( t ) = A
u20 ( x ) = 0
Then Wronskian,W = − A2 . For x < t
G1 ( x, t) = − x
For x > t
G2 ( x, t) = −t

Case 3, Boundaries at infinity

12

Solved Problem
d2 y
1. Find Greens function for dx2
− k2 y = f ( x ) ;y(±∞) = 0
It’s solution is
y = Aekx + Be−kx
then, first boundary condition y(+∞) = 0 ⇒

0 = Ae∞ + Be∞

Ae∞ = 0, A = 0. So u1 ( x ) = B e−kx , u1 (t) = B e−kt , u10 (t) = −k B e−kt . Second boundary condition
y(−∞) = 0 ⇒
0 = Ae−∞ + Be+∞
0 = 0 + Be+∞
Be+∞ = 0
B=0
So u2 ( x ) = A ekx , u2 (t) = A ekt , u20 (t) = k A ekt . Then Wronskian, W = 2kAB For x < t

ek(t− x )
G1 ( x, t) =
2k
For x > t
− ek( x −t)
G2 ( x, t) =
2k

Green’s Function for Poisson’s Equation

We’ve by definitions
L y( x ) = f ( x )
L G ( x, t) = δ( x − t)
Z
y( x ) = G ( x, t) f (t)dt

Poisson’s equation says

−ρ
∇2 φ =
ε0
∇2 G = δ(~r1 −~r2 )
Then we get using definitions
ρ (r1 ) 3
Z
φ(~r2 ) = G (~r1 ,~r2 ) d r1
ε0
But from electrodynamics we know
ρ (r1 ) d3 r1
Z
φ(~r2 ) =
4πε 0 |~r2 −~r1 |
Comparing we get
1
G (~r1 ,~r2 ) =
4π |~r2 −~r1 |
This is the Greens function for Poisson’s equation.
 13

Green’s Function as a series of Eigen functions

Readers are requested to read the chapter on Sturm Liouville operator in Arfkan 7th Edition before read-
ing this section. Let
L φn ( x ) = λn φn ( x )
where φn ( x ) is the eigen function and λn is the eigen value. Usually we have

Ly( x ) = f ( x )

and the solution is always written in terms of Greens function. Here we assume Greens function in terms
of eigen functions.
G ( x, t) = ∑ cn (t)φn ( x )
n

where φn ( x ) are orthogonal eigen functions and cn (t) is unknown which is to be found out.

L G ( x, t) = δ( x − t)

Substituting
L ∑ cn φn ( x ) = ∑ cn (t)Lφn ( x )
n n

δ( x − t) = ∑ cn (t)λn φn (x)
n

Multiplying with ∗ (x)

φm and integrating over x
Z Z
∗
φm ( x )δ( x − t)dt = ∑ cn (t)λn ∗
φm ( x )φn ( x )dx
n

∗
φm (t) = ∑ cn (t)λn δmn
n
∗
φm (t) = cm (t)λm
Thus
∗ (t)
φm
cm (t) =
λm
or
φn∗ (t)
cn (t) =
λn
Then,
φn∗ (t)φn ( x )
G ( x, t) = ∑ λn
n
This is the eigen function expansion of Green’s function.

University questions
0
1. By the method of Green’s function G (~r,~r ), solve the Poisson’s equation in electro-statistics ∇2 u(~r ) =
−4πρ(~r ) , ρ(~r ) is the charge density.
2. Define Green’s function in one and three dimensions.
3. Explain how the method of Green’s function is useful in obtaining the solutions of Poisson’s equation.
14

4. Deduce Green’s function of the Helmholtz operator (∇2 + k2 ) for an outgoing spherical wave.
5. a) Define Green’s function b)Expand the Green’s function as a series in eigen functions of the corre-
sponding homogeneous equation c)Hence prove the symmetry of the Green’s function.
6. What are the important properties of Green’s function?
7. Develop the Green’s function for the Laplace equation ∇2 φ = 0 in an infinite region with the boundary
condition φ → 0 as r → ∞.
8. Explain how the Green’s function can be expressed as an eigen function expansion.
exp(ik|~r1 +~r2 |)
9. Show that 4π (~r1 +~r2 )
is the Green’s function of the Helmholtz operator (∇2 + k2 )
d dy
is the Green’s function for dx [ x dx ]