Green’s functions further examples

(For the video)

Mathemaniac

July 19, 2021

This write-up is for anyone who wants a deeper understanding of how to find and use the Green’s
function. Unless I can find a relatively novel way of presenting all these in a video format, I might not
make a separate video about this on the channel. Hope that you understand. I used the oscillator example
first because it uses single-variable calculus.

1 The oscillator example

d2 x
m + ω 2 x = F (t) (1)
dt2
Finding the Green’s function in this case is not too difficult if you know how to solve basic differential
equations. The Green’s function G(t; τ ) satisfies the following:

d2 G(t; τ )
m + ω 2 G(t; τ ) = δ(t − τ ), (2)
dt2
when viewed as a function of t. The usual convention is to further assume that τ > 0 in addition to the
homogenous conditions as in the video (i.e. x(0) = x0 (0) = 0), simply because we ultimately would want
to see how the oscillator behaves after applying the force, and we originally would want the oscillator to
be at rest, only applying some force after t > 0.

Now we know that RHS of (2) is 0 almost everywhere, and the solution in that case has the following
p
form: A sin(kt) + B sin(kt), where k = ω 2 /m. However, because it is 0 except at t = τ , it is possible
that the solution on the left of τ and that on the right of τ are different, and in fact they are different.
And since we are being clever, avoiding further unnecessarily complicated algebra, we reconsider the
constants so that we have the following solution:

A(τ ) sin(k(t − τ )) + B(τ ) cos(k(t − τ )) for t < τ
G(t; τ ) = (3)
C(τ ) sin(k(t − τ )) + D(τ ) cos(k(t − τ )) for t > τ
Note in particular that all the “constants” could be functions of τ because in this equation, τ is seen as
a constant when taking derivatives with respect to t. The initial conditions and that τ > 0 imply that
A(τ ) = B(τ ) = 0, and now we only need to find C(τ ) and D(τ ).

We want continuity, i.e. at t = τ , the Green’s function should be 0, and so this implies D(τ ) = 0. (Note
that if we haven’t changed from kt to k(t − τ ), we wouldn’t be able to eliminate a “constant” that way.)

1
The next thing to note is that integrating around τ gives 1, i.e.
  Z ε
dG dG
m (ε; τ ) − (−ε; τ ) + ω 2 G(t; τ )dt = 1,
dt dt −ε

but since G is already continuous in t around τ , the integral obviously is continuous as well, and so we
get

1
lim (G0 (ε; τ ) − G0 (−ε; τ )) = (4)
ε→0 m
We know that G0 (−ε; τ ) = 0 (assuming ε > 0), so that gives C(τ ) = 1/mk. So we now rewrite (3) as
follows:

0 for t < τ
G(t; τ ) =1 (5)
 sin(k(t − τ )) for t > τ
mk
Now we want to think the general applied force in (1). Remember that in the video, the way to solve it
would be
Z ∞
x(t) = F (τ )G(t; τ )dτ. (6)
0
The lower limit being 0 is because we assumed τ > 0 to begin with. However, the Green’s function is 0
if t < τ , or equivalently, τ > t, so there is no need to integrate up to ∞, just t, and so we have
Z t
1
x(t) = F (τ ) sin(k(t − τ ))dτ (7)
mk 0
And this concludes our oscillator example.

2 Electrostatics example
This requires knowledge of multivariable calculus. The equation we need to deal with is this:
ρ
− ∇2 φ = (8)
ε0
Again, we want to find the Green’s function first, i.e.

− ∇2 G(x; ξ) = δ(x − ξ) (9)

As said in the video, we use the symmetry in the situation, and we should know that the Green’s function
should only depend on r = |x − ξ|, so let’s say G(x; ξ) = G(r), and in this case, ∇G(r) = G0 (r)er . We
now integrate (9) in a ball B surrounding x with radius R. Then

Z Z
∇2 G(r)dξ = − δ(x − ξ)dξ
B B
Z
∇G(r)dS = −1 (Divergence theorem)
∂B

4πR2 G0 (R) = −1
1
G0 (R) = ,
4πR
but this is true for all R, so that is the Green’s function, and so the solution to (8) is simply (by applying
the principle of Green’s functions in the video):
Z
ρ(ξ)
φ(x) = dξ (10)
R3 4πε0 |x − ξ|