Theoretical Physics Reference, Release 0.

Now let’s see what we have got. Later we’ll show, that the Γi coefficients are just ∂iϕ in the Newtonian
theory. E.g. in our case we have: 00

Γ′100 = −x′ω2 = ∂′xϕ

Γ′2 = −y′ω2 = ∂′ ϕ
00 y
Γ′3 = 0 = ∂′ ϕ
00 z
from which:

ϕ(t, x, y, z) = − 1 (x′2 + y′2)ω2 + C(t)

2
and the force acting on a test particle is then:

F = −m∇ϕ = m (x′, y′, 0) ω2 = mr′ω2

where we have defined r′ = (x′, y′, 0). This is just the centrifugal force. Also observe, that we could have
read ϕ directly from the metrics itself — just compare it to the Lorentzian metrics (with gravitation) in the
next chapter.
The other two terms (Γ′102, Γ′201and the symmetric ones) don’t behave as a gravitational force, but rather only
act when we are differentiating (e.g. only act on moving bodies). Below we show this is just the −2ω × dr d
t
term (responsible for the Coriolis acceleration).
Let’s write the full equations of geodesics:

d2 x 0
=0
dλ2
d2 x 1 dx2 dx0
+1 dx0 + 1 =0
Γ 2Γ
2
dλ 2 00 20
dλ dλ
dλ

d2 x 2 dx1 dx0
+2 + 2 =0
Γ dx0 2Γ
2
00
dλ2 dλ 10
dλ dλ
d2 x 3
=
dλ2
0
This becomes:
d2x dy
2
+ 2ω
= xω dt
dt2
dy
2
2 dx
= yω
dt d2z — 2ω
2
dt
=0
dt2
we can define r = (x, y, 0) and ω = (0, 0, ω). Then the above equations can be rewritten as:
2
dr 2
= rω dr
2
dt — 2ω ×
dt
So we get two fictituous forces, the centrifugal force and the Coriolis
force. Now imagine a static vector in the xµ system along the x axis, i.e.
 
3.40. Differential Geometry 1 1
Theoretical Physics Reference, Release 0.5

10
Vµ=  =V
0

2 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

then

  
′ 1 1
′µ ′ ′
∂x ∂x −xω sin ωt + yω cos ωt + cos ωt y ω + cos ωt
V = −xω cos ωt − yω sin ωt − sin ωt  = −x′ω − sin ωt′  = V ′
′µ
V =∂xα V α =
∂x
0 0
In the last equality we transformed from xµ to x′µ using the relation between frames.
Differentiating any vector in the xµ coordinates is easy – it’s just a partial derivative (due to the Euclidean
metrics). Let’s differentiate any vector in the x′µ coordinates with respect to time (since t = t′, the time
is the same in both coordinate systems):
µ
∇0V ′µ = ∂0V ′µ + Γ0α V ′α
  ′0   
VV′0 ′1 ∂0V ′1 + Γ∂10VV ′0 + Γ1 V ′2
′0
∂0V ′1 − x∂′ω 0V2V ′0 − ωV ′2
∇0 V ′2 =  ∂ V ′2 + Γ20000V ′0 + Γ20102 ′1 =  ∂ V ′2 — y ω 
′ 2V ′0 + ωV ′1  =
0 0

V ′3 ∂0V ′3 V ∂0V ′3
 ′0    ′0
V ′1 −x′ω02 00 −ω 0 00 VV ′1
= ∂0  V +     (3.40.2.3)
 V   −y ω
′2 ′ 2
ω 0 0  V ′2 
V ′3 0 0 0 V ′3
0
For our particular (static) vector this yields:
   
1 0
y′ω + cos ωt′ 0
∇0  ′  = 
0
−x ω − sin ωt′
0 0

as expected, because it was at rest in the xµ system. Let’s imagine a static vector in the x′µ system along
the x′ axis, i.e.
 
11
W′µ =0 
0

µ 1   
′ ′ ′ ′ ′
1
∂x −x ω sin ωt − y ω cos ωt + cos ωt −yω + cos ωt
W =  x′ω cos ωt′ − y′ω sin ωt′ + sin ωt′   =  xω + sin ωt 
µ ′α
W =
∂x ′α
0 0
then
   0′ 2 
′µ 1 −x ω 
∇ 0W = ∇0 1  =   ′ 2
0 −y ω + ω

   0 0
1    
   0 0 0 0 0 0
−yω + cos ωt −ω sin ωt 0  0 −ω 0 cos ωt
∇ 0 W µ = ∂0   xω + sin ωt  =   ω cos ωt
= 0 ω 0 0 sin ωt = ω × W
0 0 0 0 0 0 0

3.40. Differential Geometry 3

Theoretical Physics Reference, Release 0.5

Similarly
 
0
−y ω − ω 
′ 3 2
′µ
∇0∇0W == 
—x′ω3 
0


20
∇0∇0W == −ω
µ cos ωt
−ω2 sin ωt
0
How can one prove the relation:
dA d
′
A (3.40.2.4)
=ω×A +
dt dt
that is used for example to derive the Coriolis acceleration etc.? We need to write it components to under-
stand what it really means:
    
0 0 0 0 ′1A′0  ′0 
A′0  0 0 −ω 0 A    A
A′1 A′1
∇0  ′2 =     + ∂0  
A 0 ω 0 0  A′2  A′2 
A′3 0 0 0 A ′3
A′3
0
Comparing to the covariant derivative above, it’s clear that they are equal (provided that x′ = 0 and y′ =
0, i.e. we are at the center of rotation).
Let’s show the derivation by Goldstein. The change in a time dt of a general vector G as seen by an observer
in the body system of axes will differ from the corresponding change as seen by an observer in the space
system:

(dG)space = (dG)body + (dG)rot

Now consider a vector fixed in the rigid body. Then (dG)body = 0 and

(dG)rot = (dG)space = dΩ × G

For an arbitrary vector, the change relative to the space axes is the sum of the two effects:

(dG)space = (dG)body + dΩ × G

A more rigorous derivation of the last equation follows from:

Gi = ajiG j
′

dGi = ajidG j + dajiG j

′ ′

Let’s make the space and body instantaneously coincident at time t, then aji = δji and daji = —ϵijkdΩk =
ϵikjdΩk, so we get the same equation as earlier:

dGi = dG i + ϵikjdΩkG j
′ ′

Anyhow, introducing ω by:

dΩ
ω=
dt
we get
dG dt dG dt
space
=
4 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

+ω×
body G

3.40. Differential Geometry 5

Theoretical Physics Reference, Release 0.5

Linear Elasticity Equations in Cylindrical Coordinates

Authors: Pavel Solin & Lenka Dubcova

In this paper we derive the weak formulation of linear elasticity equations suitable for the finite element
discretization of axisymmetric 3D problems.

Original equations in Cartesian coordinates

Let’s start with some notations: By u = (u1, u2, u3)T we denote the displacement vector in 3D Cartesian
coordinates, and by ϵ the tensor of small deformations,
1 ∂ui ∂uj
ϵ = + , 1 ≤ i, j ≤ 3.
ij
2 ∂xj ∂xi

The stress tensor σ has the form

σij = λδijdivu + 2µϵij, 1 ≤ i, j ≤ 3, (3.40.2.5)

where
3 ∂u 3
divu = Σ = Σ ϵ
k
= Tr(ϵ).
kk
k=1 ∂xk k=1

The symbols λ and µ are the Lam'e constants and δij is the Kronecker symbol (δij = 1 if i = j and δij = 0
otherwise). The equilibrium equations have the form
3
Σ ∂σ
∂xijj + fi = 0, 1 ≤ i ≤ 3, (3.40.2.6)
j=1

where (f1, f2, f3)T is the vector of internal forces (such as gravity).
The boundary conditions for linear elasticity are given by

ui = uˆ i on Γ1
3
Σ
σijnj = gi on Γ2,
j=1

where gi are surface forces.

Weak formulation

Multiplying by test functions and integrating over the domain Ω we obtain

∫ Σ ∫
3 ij
— ∂σ vi = fi vi, 1 ≤ i ≤ 3. (3.40.2.7)
Ω ∂x j Ω
j=1

Using Green’s theorem and the boundary conditions ∫

3 σij —∫ 3 σijnjvi = f i v i, 1 ≤ i ≤ 3.
∫ ∂vi Σ
Σ
Ω j=1 ∂xj ∂Ω j=1 Ω

6 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

Thus
g ∫
3 ∂v i ∫ f i v i, 1 ≤ i ≤ 3. (3.40.2.8)
∫ σij — i vi =
Σ
Ω j=1 ∂xj Γ2 Ω

Let us write the equations (3.40.2.8) in detail using relation (3.40.2.5)

∫ ∫ ∫
∂u1 ∂v1 ∂u1 ∂u2 ∂v1 ∂u1 ∂u3 ∂v1
λdivu + 2µ +µ + +µ + — g1 v 1 = f 1 v 1,
Ω ∂x1 ∂x1 ∂x2 ∂x1 ∂x2 ∂x3 ∂x1 ∂x3 Γ2 Ω
∫ ∫ ∫
∂u1 ∂u2 ∂v2 ∂u2 ∂v2 ∂u2 ∂u3 ∂v2
µ + + λdivu + 2µ +µ + — g2 v 2 = f 2 v 2,
∂x2 ∂x1 ∂x1 ∂x2 ∂x2 ∂x3 ∂x2 ∂x3 Γ
∫Ω ∫ 2 ∫Ω
∂u1 ∂u3 ∂v3 ∂u2 ∂u3 ∂v3 ∂u3 ∂v3
µ + +µ + + λdivu + 2µ — g3 v 3 = f 3 v 3.
Ω ∂x3 ∂x1 ∂x1 ∂x3 ∂x2 ∂x2 ∂x3 ∂x3 Γ2 Ω

Elementary transformation relations

First let us show how the partial derivatives of a scalar function g are transformed from Cartesian coordinates
x1, x2, x3 to cylindrical coordinates r, ϕ, z. Note that
x1(r, ϕ) = r cos ϕ, x2(r, ϕ) = r sin ϕ, x3(z) = z.
Since

g(x1, x2, x3) = g(x1(r, ϕ), x2(r, ϕ), x3(z)),

it is
∂g ∂g ∂g
= cos ϕ +sin ϕ,
∂r ∂x1 ∂x2
∂g ∂g ∂g
= r cos ϕ,
∂ϕ ∂x1 (—r sin ϕ) + 2
∂g ∂g ∂x
= .
∂z ∂x3
From here we obtain
∂g ∂g 1 ∂g
∂x1 = sin ϕ, cos ϕ —
∂r
∂g r ∂ϕ
∂g = sin ϕ + 1 ∂g cos ϕ,
∂x2 ∂r r ∂ϕ
∂g = ∂g .
∂x3 ∂z
The relations between displacement components in Cartesian and cylindrical coordinates are
u1 = ur cos ϕ,
u2 = ur sin ϕ,
u3 = uz.
The same relations hold for surface forces gi and volume forces fi.
Applying (3.40.2) to u1, we obtain
∂u1 ∂u1 1 ∂u1
= cos ϕ — sin ϕ,
∂x1 ∂r r ∂ϕ
∂u1 ∂u1 1 ∂u1
= sin ϕ + cos ϕ,
∂x2 ∂r r ∂ϕ
∂u1 ∂u1
∂x3 = ∂z .
3.40. Differential Geometry 7
Theoretical Physics Reference, Release 0.5

Using (3.40.2) and the fact that ur does not depend on ϕ, this yields
∂u1 ∂ur 1
= cos2 2
∂x1 ∂r ϕ + u r sin ϕ,
∂ur r 1
∂u1 = cos ϕ sin ϕ — cos ϕ sin ϕ,
u
∂x2 ∂r r r
∂u1 = ∂u r
cos ϕ.
∂x3 ∂z
Analogously, for u2 we calculate
∂u2 ∂ur 1
= cos ϕ sin ϕ — u cos ϕ sin ϕ,
∂x1 ∂r r r
∂u =
2 ∂u r 2 1 2
∂x2 sin
∂r ϕ + u r cos ϕ,
∂u r
∂u2 = r
sin ϕ.
∂x3 ∂z
For u3, using that it does not depend on ϕ, we have
∂u3 ∂uz
= cos ϕ,
∂x1 ∂r
∂u3 ∂uz
= sin ϕ,
∂x2 ∂r
∂u3 ∂uz
∂x3 = ∂z .
For further reference, transform also divu into cylindrical coordinates
∂u1 ∂u2 ∂u3
divu = + + =
∂x1 ∂x2 ∂x3
∂ur 2 1 2 ∂ur 2 1 2 ∂uz
= cos ϕ+ sin ϕ+ =
∂r ϕ + ur ∂r ϕ + ur ∂z
sin r cos r
∂ur 1 ∂uz
= + ur +
∂r r ∂z

Axisymmetric formulation

Assuming that the domain Ω is axisymmetric, we can begin to transform the integrals in (3.40.2) to cylindrical
coordinates. Recall that the Jacobian of the transformation is J(r, ϕ, z) = r. The first equation in
(3.40.2) has the form:
∫
∂ur 1 ∂uz ∂ur 1 ∂vr cos2 1
r λ( + u r + ∂z ) + 2µ( ∂r cos 2 ϕ + u r 2 ϕ) ( ∂r ϕ + v r 2 ϕ) +
Ω ∂r r sin r sin r
∂ur 1 r ∂v 1r
r2µ ∂r cos ϕ sin ϕ — r u cos ϕ sin ϕ ∂r r cos ϕ sin ϕ —
r v cos ϕ sin ϕ +
∫ ∫
∂ur ∂uz ∂vr
rµ cos ϕ + cos ϕ cos ϕ — rgr vrcos2ϕ = rf r v rcos2 ϕ,
∂z ∂r ∂z Γ2 Ω
The second equation in (3.40.2) has the form:
∫
∂ur 1 ∂vr 1
r2µ cos ϕ sin ϕ — u cosr
ϕ sin ϕ cos ϕ sin ϕ — vr cos ϕ sin ϕ +
Ω ∂r r ∂r r

∂ur 1 ∂uz ∂ur sin 2 + 1u r 2ϕ

) ( ∂vr sin2 + 1v ϕ) +
r λ( ∂r + r u r + ∂z ) + 2µ( ∂r ϕ ∂r ϕ r 2
cos r cos r

8 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

∫ ∫
∂ur ∂uz ∂vr
rµ sin ϕ + sin ϕ ( sin ϕ) — 2
rgr vr sin ϕ = rf r v r sin2 ϕ,
∂z ∂r ∂z Γ2 Ω

3.40. Differential Geometry 9

Theoretical Physics Reference, Release 0.5

Adding these two equations together we get

∫
∂ur 1 ∂uz ∂vr 1
rλ( + u r + ∂z )( + vr) +
∫Ω ∂r ∂u r∂v ∂r r
r r 4 1 ∂v r 1 ∂ur 1
rµ 2 cos sin2 ϕ cos2 ϕ + v r sin2 ϕ cos2 ϕ + ur vr sin4 ϕ +
Ω ∂r ∂r + ur r ∂r r2
ϕ r
∂r
∂ur ∂vr sin4 + 1 u r∂vr 1 ∂ur r 1 r r
2 ∂r ∂r ϕ r sin2 ϕ cos2 ϕ + r ∂r v sin2 ϕ cos2 ϕ + r2 u v cos4 ϕ +
∂r
∂ur ∂vr 1 ∂v 1 ∂u r 1r r
4 ∂r ∂rcos2 ϕ sin2 ϕ — r ur ∂rr cos2 ϕ sin2 ϕ —r ∂r r v cos2 ϕ sin2 ϕ +r2 u v cos2 ϕ sin2 ϕ +
∫ ∫
∂ur ∂vr ∂uz ∂vr
+ — gr vr r = f r v rr
∂z ∂z ∂r ∂z Γ2 Ω

This can be simplified to

∫ ∫
∂ur 1 ∂u
∂z ∂vr 1 ∂ur ∂vr 1 r r ∂u
∂z r ∂v
∂z r + ∂u
∂rz ∂v
rµ 2 ∂r ∂r + 2 u v ∂zr
z
rλ( + ur + )( + vr) + +
Ω ∂r r ∂r r ∫ Ω r
∫
—
gvr =
Γ2 r r
fr vrr
Ω

Finally, the third equation in (3.40.2) has the form

∫
∂ur ∂uz ∂vz ∂ur ∂uz ∂vz
rµ cos ϕ + cos ϕ cos ϕ +∫ rµ sin
∫ϕ+ sin ϕ sin ϕ +
Ω ∂ur∂z 1 ∂u∂rz ∂uz∂r ∂vz ∂z ∂r ∂r
r λ( ) + 2µ — g v r= f v r.
+ ur + z z z z
∂r r ∂z ∂z ∂z Γ2 Ω

This gives us
∫ ∫ ∫
∂ur ∂vz ∂uz ∂vz ∂uz ∂vz 1∂ur ∂uz ∂vz
rµ + +2 + rλ + ur + — g v r= f v r.
Γ2 z z Ω z z
Ω
∂z ∂r ∂r ∂r ∂z ∂z ∂r r ∂z ∂z
Since the integrands do not depend on ϕ, we can simplify this to integral over Ω0, where Ω0 is the
intersection of the domain Ω1 with the x+x3 half-plane. Dividing both equations by 2π we get
∫ ∫
∂ur 1 ∂u
∂z z )( ∂vr 1 ∂ur ∂vr 1 r r ∂ur ∂vr ∂uz ∂vr
rλ( + u r + + v r) +
rµ 2 ∂r ∂r + 2 u v + ∂z ∂z + ∂r ∂z
Ω0
∂r r ∂r r ∫ Ω0 r
— ∫
g v
Γ2 r r r = fr vrr
Ω0
∫ ∫ ∫
∂ur ∂vz ∂uz ∂vz ∂uz ∂vz ∂ur 1 ∂uz ∂vz
rµ + +2 + rλ + ur + — g v r= f v r.
Ω0 Γ2 z z Ω0 z z
∂z ∂r ∂r ∂r ∂z ∂z ∂r r ∂z ∂z

Coordinate Independent Way

Let’s write the elasticity equations in the cartesian coordinates again:

σij = λδij∂kuk + µ(∂jui + ∂iuj)

∂jσij + f i = 0
Those only work in the cartesian coordinates, so we first write them in a coordinate independent way:

σij = λgij∇kuk + µ(∇jui + ∇iuj)

i
∇jσ ij + f = 0
10 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

so:

∇j λgij∇kuk + µ(∇jui + ∇iuj) + f i = 0

The weak formulation is then (do not sum over i):

∫ ∫
√ √
j i i
— ij k
∇j λg ∇ku + µ(∇ u + ∇ u ) v j i 3
|g|d x = fivi |g|d3x

We apply the integration by parts:

∫ ∫
√ √
j i i j
ij k
λg ∇ku + µ(∇ u + ∇ u ) ∇jv i 3
|g|d x = fivi |g|d3x

This is the weak formulation valid in any coordinates. Using the cylindrical coordinates (see above) we get:

x = (ρ, ϕ, z)
3
d x = dρ dϕ dz
 
ij 1 01 0
g = 0 2 0
0 ρ0 1
√ q
|g| = | det gij| = ρ
k 1 √ 1
k k
∇ ku = √ ∂k( |g|u ) = ∂k(ρu ) =
|g|1 ρ ρ
= u + ∂ uρ + ∂ uϕ + ∂ uz
ρ ϕ z
ρ
(∇juz + ∇zuj)∇jvz = (gjk∇kuz + gzk∇kuj)∇jvz = (∂ρuz + ∂zuρ)∂ρvz + (∂zuz + ∂zuz)∂zvz =
= (∂ρuz + ∂zuρ)∂ρvz + 2∂zuz∂zvz
gρj ∇ vρ = gρρ∇ vρ 1
j ρ = ∂ρvρ + Γρkρ = ∂ ρv ρ + vϕ
vk ρ
+ v )=
gϕj∇ vϕ = gϕϕ∇ 1 1 ϕ 1
ρ ρ
ϕv
ϕ
j = v ϕ + Γ ϕ v k) =
(∂
ϕ
ρj2vz = gzz∇zvz = ∂zvρz2 (∂ v
kϕ
∫ gzj∇ +∫ ϕΓz vk = ∂zvz
1
λgij uρ + ∂ uρ + ∂ uϕ + ∂ uz + µ(∇jui + ∇iuj) ∇ viρ dρ dϕ dz = fkz i i
v ρ dρ dϕ dz
ρ ϕ z j
ρ
for i = 1, 2, 3 we get:
∫ ρ ϕ z ∫
1ρ ρ ρ 1 ϕ
λ u + ∂ u + ∂ u + ∂ uz
ρ ϕ
ρ + µ (2∂ uρ∂ vρ + (∂ uρ + ∂ uz)∂ vρ) ρ dρ dϕ dz = f ρ v ρ ρ dρ dϕ
∂ρv dz+ ρv ρ ρ z ρ z
∫ ∫
1 ρ 1 ϕ 1 ρ
λ u + ∂ρ uρ + ∂ϕ uϕ + ∂z uz ∂ϕv + ρv ρ + µ (2∂ρ u ∂ρ v + (∂ zu + ∂ ρu )∂ zv ) ρ dρ dϕ dz = f ϕ v ϕ ρ dρ dϕ
ρ ρ ρ z ρ
ρ ρ2
dz
∫ ∫
1 ρ
λ u + ∂ ρuρ + ∂ ϕuϕ + ∂ zuz ∂ zvzρ + µ ((∂ ρuz + ∂ zuρ)∂ ρvz + 2∂ uz
z ∂ v
z
z ) ρ dρ dϕ dz = f z v z ρ dρ dϕ
ρ
dz

3.40. Differential Geometry 11

Theoretical Physics Reference, Release 0.5

3.41 Operators

3.41.1 Introduction
The domain of the operator A is D(A), a subspace of the Hilbert space H . Linear operator is:

A(α |u⟩ + β |v⟩) = αA |u⟩ + βA |v⟩

for all |u⟩ , |v⟩ ∈ D(A). Symmetric operator is:

⟨u|Av⟩ = ⟨Au|v⟩

for all |u⟩ , |v⟩ ∈ D(A) dense in H . If D(A) is dense in H , then the adjoint operator A† is defined by

⟨u|A†v⟩ = ⟨Au|v⟩

for all |u⟩ ∈ D(A). The domain D(A†) is given by all |v⟩ for which the above relation holds. It can be
shown that D(A) ⊂ D(A†).
Operator A is self-adjoint if A = A†. Symmetric operator is self-adjoint only if D(A) = D(A†). (Bounded
symmetric operator is always self-adjoint.) Hermitean operator is a bounded symmetric operator.
Hermitian implies self-adjoint implies symmetric, but all converse implications are false. Below, we need the
operator to be self-adjoint (we assume unbounded by default).

3.41.2 Spectrum
To obtain a spectrum of the operator A, we need to solve the following problem:

A |λ⟩ = λ |λ⟩

Those values of λ for which the solution |λ⟩ ∈ H belong to the discrete part of the spectrum. λ are called
eigenvalues and |λ⟩ eigenvectors. Those values of λ for which |λ⟩ can be normalized to a delta function:

⟨λ|κ⟩ = δ(λ — κ)

belong to the continuous part of the spectrum (note that in this case |λ⟩ ∈/ H ).
Eigenvectors belonging to the continous part of the spectrum obey the completeness relation:
∫
|λ⟩ ⟨λ| dλ = 1

Eigenvectors belonging to the discrete part obey the following completeness relation:
Σ
|λ⟩ ⟨λ| dλ = 1
λ

The sum or integral runs over the whole spectrum (if the spectrum contains both discrete and continous
part, we simply combine sums and integrals).
Spectrum of a self-adjoint operator is real, because

⟨λ|A|λ⟩ = λ ⟨λ|λ⟩ = λ∗ ⟨λ|λ⟩

12 Chapter 3. Mathematics
 Theoretical Physics Reference, Release 0.5

The eigenvectors are orthogonal:

⟨λ|A|κ⟩ = κ ⟨λ|κ⟩ = λ ⟨λ|κ⟩

(κ — λ) ⟨λ|κ⟩ = 0
So for κ /= λ we get ⟨λ|κ⟩ = 0, for κ = λ the ⟨λ|λ⟩ is equal to 1 if λ belongs to the discrete spectrum and we
get:
⟨λ|κ⟩ = δλκ
or it is normalized as a delta function if it belongs to the continous part:
⟨λ|κ⟩ = δ(λ — κ)
As such, eigenvectors of a self-adjoint operator are complete and orthogonal in the above sense. Thus any
function from the space can then be expanded into the series:
Σ
f (x) = ⟨x|f ⟩ = ⟨x|λ⟩ ⟨λ|f ⟩
λ
where ⟨x|λ⟩ are the eigenvectors and the coefficients ⟨λ|f ⟩ are given by:
∫ ∫
⟨λ|f ⟩ = ⟨λ|x⟩ ⟨x|f ⟩ dx = ⟨λ|x⟩ f (x)dx

The sum over λ runs over the whole spectrum (i.e. it becomes an integral over the continuos parts). Also
the coefficients⟨ λ| f are either discrete or continous depending on the part of the spectrum. The series
converges in the⟩ norm, i.e. the following norm goes to zero as we sum over λ:
Σ
¨ (x) —
¨f
λ
⟨x|λ⟩ ⟨λ|f ⟩¨ →
¨
0
3.41.3 Derivative Operator
We have the eigenvalue problem
Au = λu

where
d
A = —i
dx
The operator A is unbounded. A is self-adjoint if:
∫ b ∫ b
u∗(x)Av(x)dx = (Au(x))∗v(x)dx
a a
So
∫ b ∫ b
u∗(x)Av(x)dx = d
a
u∗(x) i dx v(x)dx =
∫ b a —
d ∗
= i u (x) v(x)dx — i[u∗(x)v(x)]b =
dx a
a
∫ b
d
= —i ∗
u(x) v(x)dx — i[u∗(x)v(x)]b =
a dx a
∫ b
= (Au(x))∗v(x)dx —i[u∗(x)v(x)]ab
a
The operator is self-adjoint if and only if [u∗(x)v(x)]ba = 0. Few boundary conditions that satisfy this
condition:

3.41. Operators 169

Theoretical Physics Reference, Release 0.5

• Dirichlet boundary conditions

u(a) = 0, u(b) = 0
• Periodic boundary conditions

u(a) = u(b)
• Antiperiodic boundary conditions

u(a) = —u(b)
Solving the eigenproblem:

Au = λu
d
—i u = λu
dx
u(x) = eiλx

Fourier Series

We restrict our space to periodic functions. Applying the periodic boundary condition:
u(a) = eiλa = u(b) = eiλb
so

2πn eiλ(b−a) = 1
λ= for n = 0, ±1, ±2, . . .
b—a
The normalized eigenvectors are:
1 i 2πn x
un(x) = √ e b−a
b—a
These eigenvectors belong to our space and as such all λ = 2πn
b −form
a a discrete spectrum. Other solutions
do not satisfy the periodic boundary condition and so there i s n o continous part in the spectrum.
The eigenvectors must be orthogonal, as we can check:
∫ b
u∗n(x)um(x)dx =
∫ b
a
1 −i 2πn x 1 i 2πm x
= √ e b− √ e b−a dx =
a b—a a b—a
1 ∫ b
= 2π(m− n)
dx =
ei b−a x
 ∫ b—a a
b 0
 1 e dx
h ib for m = n =
= b−a
1a i
2π(m−n)
x
e for m /= n
 b−a
i2π(m−n)
a
= (1 1 for m = n
2π(m−n) 2π(m−n)
=
b i i a
i2π(m —n) b−
—e b−a for m /= n
e a
=
for m = n

for
= 1ai 2π(m−n)
−
(—
π ( m −n )
2 (b
i a) b−a
e b−a e 1 m /=
 i2π(m−n) 1 for m = n
= = mn
0 for m /= n

170 δ Chapter 3. Mathematics