PHYS 6313

METHODS OF MATHEMATICAL PHYSICS

1
 PRELIMINARIES

Stokes Theorem: Let Σ be a smooth oriented surface in R3 with boundary Γ ≡ ∂Σ. For
a smooth vector field (i.e., one with continuous first-order partial derivatives) v(x) defined
in a region containing Σ Z Z
curl v · dΣ = v · dΓ. (1)
Σ ∂Σ
Proof: Show the result for a small square, and extend to the entire domain.

Examples: If P (x, y) and Q(x, y)are real functions with continuous first-order partial
derivatives, then Z Z  
∂Q ∂P
P dx + Qdy = − dS.
∂Σ Σ ∂x ∂y

Divergence Theorem (Gauss Theorem): If V ∈ R3 is a compact set with a smooth

boundary ∂V . If v(x) is a continuously differentiable vector field defined on a neighborhood
of V , then Z Z
div v dV = v · dS. (2)
V ∂V
Proof: Show the results for a small cube, and generalize.

Examples:
(1) Consider a fluid of density ρ(x, t) that flows with a velocity v(x, t). Show that the
conservation of mass can be expressed as
∂ρ
+ ∇(ρv) = 0,
∂t
referred to as the continuity equation.

Stokes’ theorem and Gauss’ theorem are special cases of a more general theorem which is
best expressed using differential forms (ω) and exterior derivatives (dω), in the succinct
form Z Z
dω = ω.
D ∂D
Its derivation is given in “Mathematical Methods of Classical Mechanics” by V. I. Arnold.

2
 .

ELEMENTARY COMPLEX ANALYSIS

Texts:
(1) “Introduction to Complex Analysis,” H. A. Priestley, Clarendon Press, Oxford

3
 COMPLEX FUNCTIONS, CONTINUITY AND DIFFERENTIABILITY

A complex number z ∈ C can be expressed as z = x + iy = reiθ , where x, y ∈ R and

√ p
i = −1. Its magniture is x2 + y 2 = r, while its phase is arctan (y/x) = θ. The
complex conjugate of z is z̄ = x − iy = re−iθ . z z̄ and Re z ≡ 12 (z + z̄) are real while
Im z ≡ 21 (z − z̄) is imaginary.
A line segment in the complex plane is [a, b] = {(1 − t)a + tb : 0 ≤ t ≤ 1} and a circle
is |z − a| = r > 0. The upper half plane (resp. lower half plane) is {z : |z − i| < |z − i|}
(resp. {z : |z − i| > |z − i|}).

Examples:
(1) Show that (a) |Re z| ≤ |z|, (b) |Im z| ≤ |z|, (c) |z + w| ≤ |z| + |w|, and (d)
|z + w| ≥ ||z| − |w||.
(2) Give an expression that defines an annular region.

Open and Closed Sets

An open circle and a closed circle are defined as

D(a; r) ≡ {z ∈ C : |z − a| < r}
D̄(a; r) ≡ {z ∈ C : |z − a| ≤ r},

for some a ∈ C and r > 0. A punctured disc is D0 (a; r) ≡ {z ∈ C : 0 < |z − a| < r}. A set
S ⊂ C is open if ∀z ∈ S ∃r>0 s.t. D(z; r) ⊂ S. S ⊂ C is closed if C \ S is open. z ∈ C is a
limit point of S ∈ C if D0 (z; r) ∩ S 6= ∅ for every r > 0. The closure S̄ of S is the union of
S and its limit points.
A function f : S → C is continuous at a ∈ S if ∀ε > 0 ∃δ > 0 such that z ∈ D(a; δ) ∩ S
implies |f (z) − f (a)| < ε.

Examples:
(1) Show that D(a; r) is open and D̄(a : r) is closed.
(2) Prove that for any z, w ∈ C |1 − z̄w|2 − |z − w|2 = (1 − |z|2 )(1 − |w|2 ). Deduce that if
z, w ∈ D(0; 1) then |(z − w)/(1 − z̄w)| ≤ 1.

4
Holomorphic Functions and Power Series
A function f defined on an open subset G ∈ C is differentiable at z ∈ G if

f (z + h) − f (z)
f 0 (z) ≡ lim|h|→0
h
exists and is finite. The function is said to be holomorphic if it is differentiable at every
point in G. A function differentiable at z is continuous at z.

Theorem: Let f be holomorphic on an open set and write z = x + iy and f (z) =

u(x, y) + iv(x, y), where u and v are real functions defined in R2 . Then u and v satisfy the
Cauchy-Riemann equations

ux = vy ; uy = −vx . (3)

The conditions imply that the derivatives along all directions in the complex plane at z
are the same.

Examples:
(1) Suppose f and g are holomorphic. Show that f ± g and f · g are holomorphic. Under
what conditions is f /g holomorphic?
(2) Show that f : z → z̄ is not holomorphic. Deduce that Re (z) and Im (z) are not
holomorphic.
(3) Suppose f is defined on D(0; R). Then (a) if f 0 = 0 in D(0; R), it is constant. (b) If
|f | is a constant in D(0; R), then f is constant.
(4) If f and g are holomorphic on S ⊂ C, then λf , (f + g) and f · g are holomorphic on S.
f /g is holomorphic at points where g(z) 6= 0.
(5) If f is holomorphic
P∞ on S and g is holomorphic on f (S), then g ◦ f is holomorphic.
(6) Show that n=0 cn (z − a)n is holomorphic.

Complex Power Series P

Suppose han in≥0 is a complex sequence. The series sn ≡ an is saidP
to converge to s if
∀ε >P0 ∃N : n ≥ N → |sn − s| < ε, where sn is the partial sum sn = nn=0 ak . We write
s= ∞ n=0 an .

Examples:
n −1
P that the sequence hz in≥0 converges to (1 − z) when |z| < 1.
(1) Show
(2) If P an converges, then an → P0 as n → ∞. Further |an | ≤ M for all n.
(3) If |an | P
converges, so does an . P
(4) Suppose bn with bn ≥ 0 converges and if ∀n ≥ N |an | ≤ kbn , then an converges.

5
 cn z n (cn ∈ C, n ≥ 0), the radius of convergence R is
P
Given a power series
X
R ≡ sup {|z| : |cn z n | converges}. (4)

Examples:
(−1)n n2 z n , and of (b)
P P n
(1) Find the radius of convergence P ofn(a) P z /n!.
(2) Show thatPthe power series cn z and ncn z n−1 have the same radii P∞ of convergence.
(3) Suppose cn z n has a radius of convergence R > 0 and let f (z) = n
n=0 cn z . Then
f ∈ H(D(0; R)) andPf 0 (z) = ∞ n−1 (|z| < R).
P
n=1 ncn z
∞ n
(4) Suppose f (z) = n=0 cn z has a non-zero radius of convergence. Then f has derivatives
of all orders at 0, and f (n) (0) = n! cn .

Elementary Functions
The exponential function is defined by
∞
z
X zn
e ≡ . (5)
n!
n=0

Sine and cosine functions are defined as

∞ ∞
X z 2n X z 2n+1
cos z ≡ (−1)n ; sin z ≡ (−1)n . (6)
(2n)! (2n + 1)!
n=0 n=0

Hyperbolic functions are defined as

1 1
cosh z ≡ cos iz = (ez + e−z ); sinh z ≡ −i sin iz = (ez − e−z ). (7)
2 2
The argument of a complex number z is the set

[arg z] ≡ {θ : z = |z|eiθ }. (8)

The logarithm is defined as

[log z] ≡ {log |z| + iθ : θ ∈ [arg z]}. (9)

The power of a complex number z is defined through

[z α ] = {eα(log |z|+iθ) : θ ∈ [arg z]}. (10)

6
Exercises: P
(1) Evaluate nk=0 eikθ . Deduce that
n
X sin(n + 12 )θ
1+2 cos kθ = ,
k=1
sin 12 θ
n/2
X sin n+12 θ
1+2 cos kθ =
k=1
sin 12 θ

6 2mπ.
if θ =
3 2

P∞Letnω = 1, ω 6= 1. PExpress
(2) ez + eωz + eω z as a power series. Hence evaluate
∞ n
n=0 8 /(3n)!. Find also n=0 27 /(3n + 1)!.
(3) Prove that
sin 2x + i sinh 2y
tan(x + i y) = .
cos 2x + cosh 2y

7
 CAUCHY’S THEOREM & LAURENT SERIES

A curve γ on the (compact) parameter interval [α, β] is a continuous function from [α, β] →
C. γ(α) is the initial point and γ(β) the final point. The curve is simple if α ≤ s < t ≤
β −→ γ(s) 6= γ(t). γ is smooth if it has a continuous derivative on (α, β) and one-sided
derivatives at the ends. The image of [α, β] under γ is denoted γ ∗ .

Examples:
(1) For u, v ∈ C, γ(t) = (1 − t)u + tv is a line segment.
(2) For r ∈ R and a ∈ C, the path γ(t) = a + reiθ , with t ∈ [θ1 , θ2 ] is a circular arc.

We define a contour to be a simple closed curve consisting of line segments and circular
arcs, a definition that can be easily generalized.

The integral of a function f (z) along a curve γ is

Z Z β
f (z) dz = f [γ(t)]γ 0 (t) dt. (11)
γ α

The expression is easily generalized to a piece-wise continuous path.

Examples: R
(1) Show that γ(a;r) (z − a)n dz = 2πi if n = −1, and 0 otherwise.
(2) Show that γ z 2 dz = 0 on the contour containing the interval [−R, R] and the semi-
R

circular arc above the segment.

The Fundamental Theorem of Calculus: Suppose γ is a path with parameter interval

[α, β], F is defined on an open set containing γ ∗ , and F 0 (z) exists and is continuous at each
point on γ ∗ . Then
(
F [γ(β)] − F [γ(α)], in general
Z
0
F (z) dz = (12)
γ 0 if γ is closed

[Proof: F ◦ γ is differentiable on [α, β] and (F ◦ γ)0 (t) = F 0 [γ(t)]γ 0 (t). Now separate the
integral into real and imaginary parts.]

8
The Estimation Theorem: Suppose γ is a path with parameter interval [α, β] and
f : γ ∗ → C is continuous. Then
Z Z β
f (z) dz ≤ |f [γ(t)] γ 0 (t)|dt. (13)
γ α

If, in addition |f (z)| ≤ M for all z ∈ γ ∗ , then

Z Z β
f (z) dz ≤ M |γ 0 (t)| dt.
γ α
Rβ
f (z) dz = iiΦ f [γ(t)]γ 0 (t) dt.]
R
[Proof: Use the analogue for real functions and write γ α

Examples:
(1) Let f (z) = (z 4 +1)−1 and
R γ = ΓR , the upper half of the semi-circle of radius R centered
at the origin.. Show that | γ f (z) dz| ≤ Rπ/|R4 − 1|.
(2) Let f (z) = 1/z and γ(t) = eit with t ∈ [0, 2π]. Show that | γ f (z) dz| ≤ 2π.
R

R
Cauch’y Theorem: If f is holomorphic inside and on a (closed) contour γ, γ f (z) dz = 0.
[Proof: Rewrite the integral using f (z) = u(x, y) + iv(x, y), and use Green’s theorem and
Cauchy-Riemann relations.]

Indefinite-Integral
R Theorem: Let f be a continuous function on a convex
R region G
such that γ f (z) dz = 0 for any triangle γ in G, and a ∈ G. Then F = [a,z] f (w) dw is
holomorphic in G with F 0 = f .

Anti-Derivative Theorem: Let G be a convex region and f ∈ H(G). Then ∃ F ∈ H(G)

such that F 0 = f .
[Proof: Use Cauchy’s theorem in a triangular domain with vertices a, z and z + h where
z + h ∈ D(z; r) ⊂ G.]

Cauchy’s Integral Formula: If f is holomorphic inside and on a positively oriented

contour γ and a is in γ, Z
1 f (w)
f (a) = dw. (14)
2πi γ w − a
[Proof: Deform γ to γ(a; ε) with ε < r and D(a; r) ⊂ I(γ).]

Liouville’s Theorem: If f is holomorphic and bounded in C, then F is constant.

9
[Proof: Let |f (w)| ≤ M in C and a, b ∈ C. Choose R ≥ 2 max{|a|, |b|}. Using Cauchy’s
integral formula and the estimation thereom, show that |f (a) − f (b)| ≤ RM |a − b|/(R/2)2 .]

The Fundamental Theorem of Algebra: If p(z) is a non-constant polynomial with

complex coefficients, ∃ ξ ∈ C such that p(ξ) = 0.
[Proof: Suppose p(z) is non-constant and has no root, then q(z) = 1/p(z) is holomorphic.
Since |p(z)| → ∞ as |z| → ∞, q(z) is bounded in C and hence constant by Liouville’s
theorem.]

Cauchy’s Formula for Derivatives: If f is holomorphic inside and on a positively

oriented contour γ and a is inside γ, then f (n) (a) exists for all positive integers and
Z
n! f (w)
f (n) (a) = dw. (15)
2πi γ (w − a)n+1
[Proof: Differentiate Cauchy’s integral formula with respect to a.]

Poisson Integral Formula: If f is holomorphic inside and on γ(0; 1), then for r < 1
Z 2π
1 1 − r2
f (reiθ ) = f (eit ) dt.
2π 0 1 − 2r cos(θ − t) + r2

Examples: R
2
(1) Show that γ(i;1) z 2z+1 dz = −π.
(2) Show that γ(0;1) ez z −3 dz = πi.
R
Re z
dz = 12 πi.
R
(3) Show that γ(0;1) z−1/2

Taylor’s Theorem: If f ∈ H(D(a; r)), then ∃ unique constants cn such that for z ∈
D(a; r)
X∞
f (z) = cn (z − a)n , (16)
n=0
where
f (n) (a)
Z
1 f (w)
cn = dw = ,
2πi γ(a;r) (w − a)n+1 n!
where r < R.
[Proof: (w − z)−1 = ∞ n n+1 .]
P
n=0 (z − a) /(w − a)

10
Multiplication of Power Series:
∞
X ∞
X ∞
X
n n
an z · bn z = cn z n ,
n=0 n=0 n=0
Pn
with cn = r=0 ar bn−r .

Laurent Series
Most commonplace functions fail to be holomorphic at isolated points. The Taylor theorem
needs to be generalized for such cases.

Laurent’s Theorem: If A = {z : R < |z − a| < S} (0 ≤ R < S ≤ ∞) and f ∈ H(A),

∞
X
f (z) = cn (z − a)n (z ∈ A), (17)
n=−∞

where Z
1 f (w)
cn = dw,
2πi γ(a;r) (w − a)n+1
with R < r < S. Unlike the Taylor coefficients, the Laurent coefficients are not easily
evaluated. However, the Laurent expansion is unique, and this aids in computing the
coefficients.

Examples:
(1) Show that f (z) = 1/[z(1 − z)] is Pholomorphic on A1 = {z :P0 < |z| < 1} and on
A2 = {z : |z| > 1}, and that f (z) = n=−1 z on A1 and f (z) = −2
∞ n n
n=−∞ −z on A2 .
(2) 2
f (z) = 1/[z(1 − z) ] is holomorphic for 0 < |z − 1| < 1 and f (z) =
P∞Show thatn n
n=−2 (−1) (z − 1) .
(3) Find expansions for sin z, cosec z, and cot z.

Singularities
Let f be a complex-valued function. The point a is a regular point of f if f is holomorphic
at a, otherwise it is a singularity. It f has an isolated singularity at a if it has a singularity
P∞ on {z : 0n < |z − a| < r} for some r > 0. Given the Laurent
at a and is holomorphic
expansion f (z) = n=−∞ cn (z − a) , the point a is:

11
 • a removable singularity if ∀n<0 cn = 0.

• a pole of order m (m ≥ 1) if c−m = 0 and cn = 0 for all n < −m.

• an isolated essential singularity if there does not exist m such that cn = 0 for all
n < −m.

Exercises: R
cos z
(1) Evaluate γ(0;1) z dz. Deduce that
Z 2π
cos(cos θ) cosh(sin θ) dθ = 2π.
0

(2) Let γ(t) = eit (t ∈ [− 12 π, 21 π]) and let f (z) = −1 / γ∗). Compute f (n) (0)
R
γ (w − z) dw (z ∈
for n = 0, 1, 2, . . . and deduce that
∞
1 1 1X z 2n+1
f (z) = + (−1)n .
2πi 2 π (2n + 1)
n=0

(3) Evaluate when γ = γ(0; 2),

Z 3 Z
z +5 R 1 sin z
(a) dz, (b) γ z 2 +z+1 dz, (c) 2
dz
γ z−i γ z +1
2
ez
Z Z
z n (1 − z)m dz.
R cos z
(d) 3
dz, (e) γ zn dz, (f )
γ (z − 1) γ
(18)

(4) (Schwarz’s Lemma) Suppose f ∈ H(D(0; 1)) is such that |f (z)| < 1 for all z ∈ D(0; 1)
and f (0) = 0. Show that there exists g ∈ H(D(0; 1)) such that f (z) = z g(z). Prove that

|f (z)| ≤ |z| (|z| < 1) and |f 0 (0)| ≤ 1.

12
 RESIDUE THEOREM

The residue theorem is the generalization of Cauchy’s theorem for functions with poles.
It’s application is the primary technique to compute many useful integrals.

Lemma: Let f be holomorphic inside and on a positively oriented contour γ except at

a pole a in γ. Then γ f (z) dz = 2πic−1 , where c−1 is the coefficient of z −1 in the Lau-
R

rent expansion of f . The right side is referred to as the residue of f at a, and denoted
res{f (z) : a}.
[Proof: Deform γ to γ(a; r) with D̄(a; r) ∈ I(γ).]

Cauchy’s Residue Theorem: When f contains a finite number of poles at a1 , . . . , am

inside γ
Z m
X
f (z) dz = 2πi res{f (z) : ak }. (19)
γ k=1

[Proof: Construct an appropriate deformations of γ.]

Examples:
(1) If f is holomorphic inside and on a positively oriented contour γ except for P poles in
γ. Further let f 6= 0 on γ but have N zeroes within γ. Show that
Z 0
1 f (z)
dz = N − P.
2πi γ f (z)

(2) Rouche’s Theorem: If f and g are holomorphic on and inside a contour γ and
|f (z)| > |g(z)| on γ ∗ , show that f and f + g have the same number of zeros inside γ.
(3) Show that 2 + z 2 − eiz has precisely one zero in the open upper half-plane.

Classification of Poles:
Let f have a pole of order m at a. It is

• simple if m = 1; multiple otherwise

• type I if f (z) = g(z)(z − a)−m where g ∈ H(D(a; r)) and g(a) 6= 0; type II otherwise.

Examples:
(1) If f ∈ H(D0 (a; r)) and has a simple pole at a, show that res{f (z) : a} = limz→a (z −

13
a)f (z).
(2) If f (z) = g(z)/(z − a), where g ∈ H(D(a; r)) and g(a) 6= 0, show that res{f (z) : a} =
g(a). If the pole has order m, then res{f (z) : a} = g (m−1) (a)/(m − 1)!.
(3) If f (z) = h(z)/k(z) where h, k ∈ H(D(a; r)), h(a) 6= 0, k(a) = 0 and k 0 (a) =
6 0, show
0
that res{f (z) : a} = h(a)/k (a).
(4) Show that Z ∞
1 π
4
dx = √ .
0 1+x 2 2
(4) Prove Jordan’s inequality: If θ ∈ (0, π/2], then

2 sin θ
≤ ≤ 1.
π θ
(5) Let f ∈ H(D0 (a; r)) and have a simple pole of residue b at a. Let γε (θ) = a + εeiθ (θ ∈
[θ1 , θ2 ]. Show that Z
lim f (z) dz = ib(θ2 − θ1 ).
ε→0 γε

Applications of Contour Integration

Examples:
(1) Show that Z ∞
1 π
dx = .
0 (x2 2 2
+ 1) (x + 4) 18
(2) Show that the principal value integral
R
1 −√3/2
Z  
cos x 2π
lim 2
dx = √ cos e .
R→∞ −R x +x+1 3 2

(3) Show that Z ∞

sin x 1
dx = π.
0 x 2
(4) By selecting an appropriate contour, show that
Z ∞
eax
 
1
dx = π sec πa ,
−∞ cosh x 2

when a ∈ (−1, 1).

R∞ 2 √
(5) Prove that 0 e−x dx = π/2.
R∞ p
(6) Prove that 0 cos x2 dx = π/8.

14
Multifunctions
In Eqns (8), (9) and (10), we defined several multifunctions

• [arg z] ≡ {θ : z = |z|eiθ }.

• [log z] ≡ {log |z| + iθ : θ ∈ [arg z]}.

• [z α ] ≡ {eα(log |z|+iθ) : θ ∈ [arg z]}.

In each case, we selected a single (of the many) argument for analysis. Often, it is more
appropriate to consider all branches simultaneously; i.e., consider aggregate of branches as
a single function on a domain set consisting of many copies of the plane. The copies are
then glued together in order to pass continuously. This leads to a structure known as a
Riemann surface. [Think of it as a car park with ramps joining the floors.]

Excercises:
(1) Show that Z ∞
log x
dx = 0.
0 1 + x2
(2) Show that √
Z ∞
x π
3
= .
0 1+x 3
(3) Show that the Riemann zeta function ζ(2) = π 2 /6. Compute ζ(4).
(4) Let f, g ∈ H(D(a; r)) and assume that f has a zero of order m and g a zero of order
m + 1 at a. Prove that

f (m) (a)
 
f (z)
res ; a = (m + 1) (m+1) .
g(z) g (a)

(5) Evaluate
∞ ∞
x2
Z Z
cos ax
(i) dx; (ii) dx.
0 (x2 + a2 )3 0 (x2 + b2 )2
(6) Let γ be a rectangle with vertices 0, R, R + iπ and iπ. Prove that
Z ∞
sin ax 1 π πa
2x
= − + coth .
0 e −1 2a 4 2

15
 CONFORMAL MAPPINGS

Observe that the locus

z−α
=λ
z−β
represents a line when λ = 1 and a circle when λ 6= 0 (Apollonius theorem). In this
representation α and β are referred to as inverse points. When λ = 1, the locus is the
perpendicular bisector of the points given by α and β.

Examples:
(1) When λ 6= 1show that the center and the radius of the circle are a = 21 (z1 + z2 ) and
r = 21 |z1 + z2 |, where z1 and z2 are given by (α − z) = ±λ(z − β).
(2) Show that a unit circle can be expressed as
z−α
= 1.
ᾱz − 1
(3) Show that a circular arc can be expressed as
 
z−α
arg = µ (mod 2π).
z−β

A mapping f is conformal in open set G if f ∈ H(G) and f 0 (z) 6= 0 for any z ∈ G. It is

conformal at a point ξ if it is conformal in some disk D(ξ; r).

Theorem: If f is conformal in an open set G and ξ ∈ G, then f preserves angles between

paths in G meeting at ξ.

Examples:
(1) Show that the composition of conformal mappings is conformal.
(2) Show that f (z) = (z − i)/(z + i) is a conformal mapping from the open upper half-plane
to the open unit circle.
(3) The map z → z n is conformal except at 0. The map z → z π/β with 0 < β < π maps
the sector {z : 0 < arg z < β} to the upper half-plane conformally.
(4) The map z → w = ez is conformal and maps a line x = a to a circle |w| = ea , and
maps a line y = c to a half-line arg w = c. It maps a vertical strip to an annulus and the
horizontal strip to a sector.
(5) A Möbius transformation is defined as a function f : z → (az + b)/(cz + d) where

16
a, b, c, d ∈ C and ad − bc 6= 0. It is conformal in C \ {−d/c} and has an inverse f −1 : w →
(dw − b)/(a − cw). When each of {z1 , z2 , z3 } and {w1 , w2 , w3 } is a triple of distinct points
in C, there is a unique Möbius transformation that maps zk to wk . It is given by
     
w − w1 w2 − w3 z − z1 z 2 − z3
= .
w − w3 w2 − w1 z − z3 z 2 − z1

(6) An example of a Joukowski transformation is z → w = 21 (z + z −1 ) or equivalently

 2
w+1 z+1
= .
w=1 z−1

It is holomrphic except at 0 and ∞, and conformal except at ±1, where angles are doubled.
Suppose w = u + iv is the image of z = reiθ . Show that
1 1
u = (r + r−1 ) cos θ; v = (r − r−1 ) sin θ.
2 2
Find the images of the circle |z| = ρ and of the half-line arg z = µ.

Excercises:
(1) Let f : z → w = f (z) = az+b
cz+d be a Möbius transformation. A point α ∈ C is said to be
a fixed point of f if f (α) = α.
(i) Prove that f has either one or two fixed points.
(ii) Suppose f has distinct fixed points, α and β. Prove that
w−α z−α a − cα
=k , where k= .
w−β z−β a − cβ

What is the image under f of (a) the circline |z − α|/|z − β| = λ, (b) the circular arc
arg(z − α) − arg(z − β) = µ (mod 2π)?
(iii) Suppose f has a single fixed point, α. Prove that
1 1 c
= +K where K= .
w−α z−α a − cα

17
 GROUP THEORY

Texts:
(1) “Principles of Advanced Mathematical Physics, Volume II, Chapters 18, 19.” R. D.
Richtmyer, Springer-Verlag, New York

18
 ELEMENTARY GROUP THEORY

The Group Axioms

A group G is a collection of elements {a, b, c, . . . , x, y, z} together with a law of composition
◦ such that

• If a, b ∈ G then a ◦ b ∈ G.
• If a, b, c ∈ G then (a ◦ b) ◦ c = a ◦ (b ◦ c) [Associative law]
• If a, b ∈ G then there exist unique elements x, y ∈ G such that a ◦ x = b and y ◦ a = b.

Examples:
(1) The set of all integers with ◦ = +; the set of all positive real numbers under multipli-
cation; the set {0, 1, . . . , m} under addition modulo m.
(2) All rotations about the origin in a plane, where Rφ is a member which moves (x, y) to
(x0 , y 0 ) with x0 = x cos φ − y sin φ and y 0 = x sin φ + y cos φ. Further Rφ2 ◦ Rφ1 = Rφ1 +φ2 .
(3) Rotations in three dimensions.
(4) Rotations in three dimensions that leave a cube invariant.
(5) The set of all permutations of n objects in a set.
(6) Galilean and Lorentz transformations.

A subgroup of G [G 0 < G] is a subset G 0 of elements of G which itself is a group under the

same composition. An example is rotations about z-axis of the rotation group. If G 0 6= G,
it is a proper subgroup. G 0 = {e} is the trivial subgroup.

Exercises:
(1) ∃ a unique e (referred to as the identity) such that ∀a ∈ G a ◦ e = e ◦ a = a.
(2) ∀a ∈ G there is a unique a−1 (referred to as its inverse) such that a ◦ a−1 = a−1 ◦ a = e.
(3) a ◦ b = a ◦ c =⇒ b = c; b ◦ a = c ◦ a =⇒ b = c.
(4) (a ◦ b)−1 = b−1 ◦ a−1 .

Isomorphism
A one-to-one transformation φ : G → G 0 is an isomorphism if ∀a, b ∈ G : φ(a◦b) = φ(a)◦φ(b).
The groups G and G 0 are said to be isomorphic, and can be regarded as two different
realizations of the same group.

Examples:

19
(1) The set {1, i, −1, −i} under multiplication is isomorphic to rotations in the plane by
nπ/2 where n ∈ {0, 1, 2, 3}.
(2) The set of complex numbers eiθ (as is e−iθ ) under multiplication is isomorphic to the
rotation group in the plane.

Permutation Groups
A permutation is a one-to-one mapping of a set C of objects to itself, whose length is the
numbers of elements of C. Any permutation can be expressed as a product of cycles, each
of which can be expressed as a product of transpositions, i.e., cycles of length 2. Although
the cyclic expansion of a permutation is not unique, the number of transpositions is either
odd or even.

Homomorphisms
Consider a function φ : G → G 0 . It is called a homomorphism if φ(a ◦ b) = φ(a) ◦ φ(b).
The subset G0 = {x : φ(x) = e} is referred to as the kernal of φ [ker(φ)]. The set
G10 = {φ(x) : x ∈ G} is the image of G under φ [φ(G)].

Theorem: G0 and G10 are subgroups of G and G 0 respectively.

Excercises:
(1) Page 4: 1, 2, 6.

20
 CONTINUOUS GROUPS

Groups whose elements depend continuously on certain parameters are referred to as

continuous groups. The group of all non-singular (complex) n × n matrices is called
the general linear group, GL(n, C). The subgroup of real matrices is denoted GL(n, R).
Their subgroups with unit determinant are special linear groups and denoted SL(n, C) and
SL(n, R).

Orthogonal and Rotation Groups

Orthogonality of an n × n matrix R is defined to preserve properties of planar rotations,
specifically that x · y = R(x) · R(y). It follows that

RT · R = I,

and hence that det R = ±1. The orthogonal group in n-dimensions is

O(n) = {R : R = n × n real orthogonal matrix}, (20)

and the corresponding special group is

SO(n) = {R ∈ O(n) : det R = 1}. (21)

The Rotation Group SO(3): Euler’s Theorem

Suppose R is a 3 × 3 real orthogonal matrix with Det R = 1.
Euler’s Theorem: The transformation x → Rx is a rotation of the coordinate system
about a fixed direction.
Proof: Let λi and vi (i = 1, 2, 3) be the eigenvalues/eigenvectors of R. Since R is unitary,
||Rvi || = ||vi ||, and hence |λi | = 1. Further, λ1 λ2 λ3 = det R = 1. If all eigenvalues are
real one of them is unity. If there is complex conjugate pair, the other is unity. Denote the
corresponding eigenvector by v1 ; Rv1 = v1 , and v1 is invariant under R. √
Next, let iθ −iθ
√ λ2 = e and λ3 = e . Write u1 = v1 , u2 = (v2 + v3 )/ 2, and u3 =
i(v2 − v3 )/ 2, Then

Ru1 = u1
Ru2 = cos θ u2 + sin θ u3
Ru3 = − sin θ u2 + cos θ u3 .

21
Examples:
(1) Given R, show how θ and v1 can be extracted.
(2) Make the corresponding analysis for SO(n).

Intrinsic Coordinates of SO(3)

From Euler’s theorem, any rotation can be expressed as θk ≡ θ = (θx , θy , θz ), where k is
the axis of rotation and θ the rotation. We show that the rotation matrix is given by
 
0 −θz θy
R(θ) = exp  θz 0 −θx  . (22)
−θy θx 0

Toward that end, note first that Λ = log R satisfies

T
R = eΛ ; R−1 = e−Λ = RT = eΛ ,

and hence that Λ is anti-symmetric, and has the form

 
0 −c b
Λ= c 0 −a .
−b a 0

Then (1) the eigenvalues are Λ are 0, ±iθ where θ = |θ|, so that the eigenvalues of R(θ)
are 1, and e±iθ , (2) the eigenvector is proportional to θ. The assertion follows.

Unitary Groups
Unitary groups are a complex generalization of orthogonal groups, defined via U ∗ U = I.
∗
P
The Hermitian scalar product x y = j x̄j yj is invariant under a unitary transformation.
The unitary group U (n) is the group of all n × n unitary matrices.
Since det U ∗ is the complex conjugate of det U , |det U | = 1. The subset of U (n)
containing those with det U = 1 is called the special unitary group and denoted SU (n).

The Lorentz Groups

We consider homogeneous transformations; i.e., those will keep the origin fixed. Using the
standard notation x0 = ct; x1 = x; x2 = y; x3 = z, Lorentz transformation for relative

22
motion in the x-direction is given by x0µ = Lµν xν , where
 
cosh φ sinh φ 0 0
 sinh φ cosh φ 0 0
[Lµν ] = L(φ) =  0
, (23)
0 1 0
0 0 0 1
p
with sinh φ = −β/ 1 − β 2 ; β = v/c. The set {L(φ)} along with the composition defined
by the application of successive transformations [note: L(φ1 ) · L(φ2 ) = L(φ1 + φ2 )] forms
a subgroup Lx of the Lorentz group.
The rotation subgroup, R of the Lorentz group consists of matrices of the form
 
1 0
R= , (24)
0T (R0 )

where 0 = (0, 0, 0) and R0 ∈ SO(3). The group generated by elements of Lx along with
those in R [i.e., consisting of products from the two groups] is the proper Lorentz group
Lp .

Examples:
(1) Show that the fundamental quadratic form (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 is invariant
under all transformations of Lx and R, hence under all transformations of Lp .
(2) Show that the invariance in (1) is equivalent to that of x0 y 0 − x1 y 1 − x2 y 2 − x3 y 3 .

The invariant form is expressed as gµν xµ xν where [gµν ] = diag(1, −1, −1, −1).

The full (homogeneous) Lorentz group is the group of all homogeneous linear transforma-
tions under which gµν xµ xν is invariant for all 4-vectors xµ .

The Homomorphism of SU (2) onto SO(3)

Given (x, y, z) ∈ E3 , let  
z x − iy
A= , (25)
x + iy −z
which is Hermitian and traceless. For any U ∈ SU (2), define

A0 = U · A · U ∗ .

Eigenvalues of A0 and A are identical and A0 is traceless. Hence, A0 can be written as

z0 x0 − iy 0
 
0
A = ,
x0 + iy 0 −z 0

23
Further, since det A = detA0 =⇒ x2 + y 2 + z 2 = x02 + y 02 + z 02 , and a matrix R = R(U )
defined by  0  
x x
y 0  = R(U ) · y  ,
z0 z
is orthogonal.
To show that the transformation is a homomorphism, note first that since det R can only
be ±1 and R depends continuously on U , it follows that det R = 1. Further, R(U1 )·R(U2 ) =
R(U1 · U2 ) since the left side is the result of first transforming A into U2 AU2∗ , and then to
U1 U2 AU2∗ U1∗ = (U1 U2 )A(U1 U2 )∗ .
The kernal of Φ : U → R(U ) is {I, −I}; i.e., with each R ∈ SO(3) there are associates
two elements U and −U of SU (2).

In a similar manner we can establish a homomorphism of SL(2, C) onto the proper Lorentz
group Lp , which is  
t + z x − iy
A= ,
x + iy t − z
with an invariant t2 − x2 − y 2 − z 2 . The transformation is a 2-to-1 homomorphism of
SL(2, C) into Lp .

Excercises:
(1) Show that    
cos θ − sin θ 0 0 −θ 0
R(θẑ) =  sin θ cos θ 0 = exp θ 0 0 ,
0 0 1 0 0 0
and derive corresponding statements for R(θx̂) and R(θŷ).
(2) Show that the derivatives of R(θ) w.r.t. θx , θy , and θz themselves belong to the group.
(3) If P1 and P2 are two pure Lorentz transformations in the x and y directions. Do P1
and P2 commute? If not, find their difference. [Further reading: Wigner rotations and
Thomas precession.]
(4) Find the transformation R(θ, φ, ψ) resulting from successive rotations θ ẑ, φ x̂ and ψ ẑ
[Euler angles]. Show that R · RT = RT · R = I.
(5) Find an algebraic equation F (x, y, z) = 0 which determines the surface of a torus in
V 3 . Find a group of which the torus is the group manifold.

24
 .

GROUP REPRESENTATIONS

Texts:
(1) “Principles of Advanced Mathematical Physics, Volume II, Chapters 20, 21, 22.” R. D.
Richtmyer, Springer-Verlag, New York
(2) “Quantum Mechanics, Volume 2,” C. Cohen-Tannoudji, B. Liu and F. Laloë, Wiley-
VCH, Paris
(3) “Advanced Mathematical Methods for Scientists and Engineers,” C. M. Bender and S.
A. Orzag, McGraw-Hill Book Company, New York.

25
 SO(2) AND FOURIER TRANSFORMS

Group representations are closely related to various special functions. Spherical harmonics
and Legendre functions appear in problems with spherical symmetry. Trigonometric func-
tions, in the form if eimφ , appear in the representation of the two-dimensional rotation
group, and are thus associated with symmetry about an axis.

Finite Dimensional Representations of a Group

Let G be a group, and X an n-dimensional vector space. A homomorphism ρ : g → ρ(g)
of G onto a group of linear transformations in X is called an n-dimensional representation
of G. Such transformations are typically described via matrices.

Examples:
(1) The group of matrices  
cos φ sin φ
(26)
− sin φ cos φ
is a 2-dimensional representation of rotations about the z-axis. One can easily define a
corresponding 3-dimensional representation.
(2) Boosts can be represented by matrices of the Lorentz group.
(3) Give a 3-dimensional representation of rotations in space.

Representations of the identity in G, the inverse of an element g ∈ G, and the law of

composition are given by the identity matrix, inverse of the representation of g, and matrix
product respectively. If the homomorphism is an isomorphism, ρ is called faithful. If X
contains a proper subspace X1 that is invariant under all transformations ρ(g) ∀g ∈ G,
the representation is reducible and gives a sub-representation. If no such X1 exists, the
representation is irreducible.

Vector and Tensor Transformation Laws

As an example, we can consider representations of SO(2) inspired by transformations of
tensor components. Let the transformation of a point x ∈ R2 be given by x0i = 3k=1 gik xk
P
where (gik ) is given in Eqn. (26). A tensor of rank 2 is an object consisting of 22 elements
{Tij } that transform as X
Tij0 = gik gjl Tkl . (27)
(k,l)

26
 Next, express the components of Tij as (Z1 , Z2 , Z3 , Z4 )T , where the elements are ar-
ranged in row and column order.

Examples:
(1) Show that {Zk } transforms as
 0 
cos2 φ sin2 φ
  
Z1 cos φ sin φ cos φ sin φ Z1
Z20  − cos φ sin φ cos 2φ − sin 2
φ cos φ sin φ  Z2 
 = · 
Z30  − cos φ sin φ − sin2 φ cos2 φ cos φ sin φ Z3 
Z40 sin2 φ − cos φ sin φ − cos φ sin φ cos2 φ Z4

(2) Now let us consider a sub-representation. Show that the symmetry (Z2 = Z3 ) and
traceless feature (Z1 + Z4 = 0) of a vector are invariant under transformation. Thus,
consider a subgroup of R4 containing vectors of the form (Z1 , Z2 , Z2 , −Z1 )T . Show that
the resulting representation is
 0    
Z1 cos 2φ sin 2φ Z1
= · .
Z20 − sin 2φ cos 2φ Z2

This representation is irreducible.

(3) Carry out a similar analysis for SO(3).
(4) Construct representations of SO(2) inspired by tensors of rank 3 which are collections
of 23 objects Tijk that transform as
X
0
Tijk = gil gjm gkn Tlmn . (28)
(l,m,n)

The subgroup corresponding to symmetric and traceless vectors gives a 5 dimensional

irreducible representation.

Examples:
(1) Show that the trace of a second-rank tensor is preserved under rotations.

An Irreducible Representation of SO(2): Fourier Series & Fourier Transforms

Any circle centered at the origin is invariant under SO(2). Let X ∞ be the infinite-
dimensional space of infinitely differentiable functions f (φ) on the circle. The rotation
gα can be associated with a linear transformation

ρα : f (φ) → f (φ − α),

in X ∞ . Then gα → ρα is a representation of SO(2) on X∞ .

27
 A way to find sub-representations is through infinitesimal operators of a representation;
i.e., T → (ρα − ρ0 )/α as α → 0, or

(T f )(φ) = −f 0 (φ). (29)

Any subspace of X ∞ invariant under all transformations ρα will be invariant under T .

First look for invariant subspaces of smallest dimension, i.e., 1: f (φ) is invariant under T ,
leading to the eigenvalue problem

−f 0 (φ) = λf (φ).

For f (φ) to be single-valued, λ = im and thus

f (φ) = fm (φ) = e−imφ ,

and
(ρα fm )(φ) = eimα fm (φ).
For each m, the 1-(complex) dimensional subspace

Xm = {Ae−imφ : A ∈ C}

is invariant not only under T but also under all ρα . By Fourier theorem any function in
X ∞ can be expanded in these functions, hence subspaces Xm span X ∞ . Further, only
finite-dimensional representations of SO(2) on X ∞ are those constructed from Xm ’s.

For the remainder of the section, we generalize to periodic functions with period L. A
function f (x) is periodic with period L if for all x

f (x + L) = f (x).

The smallest such length is the fundamental period.

Examples:
(1) Show that cos 2π Lx , sin 2π Lx , and e±2iπx/L are periodic with period L.
(2) e±2iπnx/L has period L, but its fundamental period is L/n.

Suppose f (x) is a smooth periodic function with fundamental period L, and its derivatives
at the ends of the interval match. Then
+∞
X
f (x) = cn eikn x ,
n=−∞

with kn = n(2π/L). The set of values {cn } is called the Fourier spectrum of f (x).

28
Examples:
(1) The coefficient
Z x0 +L
1
cn = dx e−ikn x f (x).
L x0

(2) f (x) is real iff ∀n c−n =c∗n .

(3) Show the Bessel-Parseval relation
Z x0 +L +∞
1 X
dx |f (x)|2 = |cn |2 .
L x0 n=−∞

Next, consider a function f (x) that is not necessarily periodic. Define fL (x) to be a function
equal to f (x) in [−L/2, L/2] that is assumed to match at the ends. fL (x) can be expanded
as
X+∞
fL (x) = cn eikn x ,
n=−∞

where kn is defined earlier and

Z L/2
1
cn = dx e−ikn x f (x).
L −L/2

As L → ∞, fL (x) → f (x) and kn ’s approach a continuum. Real-space [f (x)] and Fourier

[typically denoted fˆ(k)] representations contain the same information.

An important application of Fourier transforms involves the following observation: if the

Fourier transform of a smooth function f (x) is fˆ(k), then that of dn f (x)/dxn is (ik)n fˆ(k).
Thus, linear differential equations in real-space convert to algebraic equations in Fourier
space.

Examples:
(1) Find the general real solution to the harmonic oscillator ẍ(t) + ω 2 x(t) = 0.
(2) Find and discuss real solutions of the damped harmonic oscillator ẍ(t)+αẋ(t)+ω 2 x(t) =
0, where α > 0. Plot the solutions.
(3) Find and discuss real solutions of the driven damped harmonic oscillator ẍ(t) + αẋ(t) +
ω 2 x(t) = γ cos(ω0 t), where α, γ > 0. Plot the solutions, in particular the amplitude of the
oscillations.

29
Excercises:
(1) Find the Fourier transform of sin x and cos x.
(2) Find the Fourier transform of δ(x) and aδ(x − b) [careful about the periodic domain]
(3) Find the Fourier transform of the standard normal
h distribution.
i
2
(4) Find the Fourier transform of f (x) = √ 2 exp − (x−a)
1
2σ 2
.
2πσ

30
 SO(3) AND SPHERICAL HARMONICS

Homogeneous Spaces
If a space Vn (V ∈ {R, C}) contains a surface S of lower dimension that is invariant under
all transformations of G, then X ∞ can be replaced by the space of functions defined on S.
Examples of S include the unit circle in the case of SO(2) or the unit sphere in the case
of SO(3). The action of the group on S also need to satisfy:
• G is effective for S: only the identity e ∈ G gives the identity in S [If g 6= e then
∃x∈G gx 6= x].
• G is transitive for S: for x, y ∈ S → ∃g∈G y = gx.
Under these conditions, S is said to be a homogeneous space for G.
A general procedure for finding irreducible representations of a continuous group of
linear transformations in Vn consists of the following steps:
• Find a homogeneous space Sand let ρ be the representation f (x) → f (g −1 x) of func-
tions on S.
• Find a complete set of infinitesimal operators.
• Find minimal invariant subspaces of X ∞ using infinitesimal operators.
Special functions associated with the symmetries of G are the elements of the invariant
subspaces of X ∞ .
We will derive special functions associated with SO(3) and SL(2, C) below.

Representations of the Rotation Group SO(3)

Let gω = gωx ,ωy ,ωz be the matrix of rotation through an angle ||ω|| about ω, as given by
Eqn. (22), and let X ∞ be the space of all infinitely differentiable functions f (x) of R3 . The
operator ρ(g) is defined according to
(ρ(g)f ) (x) = f (g −1 x), (30)
whose infinitesimal operator for changes in ωx is
1
L1 ≡ lim [ρ(gω,0,0 ) − ρ(g0,0,0 )] .
ω→0 ω
It follows that L1 (and similarly L2 and L3 ) are given by
∂ ∂
L1 = z −y ,
∂y ∂z
∂ ∂
L2 = x −z ,
∂z ∂x
∂ ∂
L3 = y −x . (31)
∂x ∂y

31
Examples:
(1) Show that [Li , Lj ] = εijk Lk .
(2) For the homogeneous space given by the unit circle, show that in polar coordinates

∂ ∂
L1 = sin φ + cot θ cos φ
∂θ ∂φ
∂ ∂
L2 = − cos φ + cot θ sin φ
∂θ ∂φ
∂
L3 = −
∂φ

(3) Show that the raising and lowering operators L± = e±iφ (∓i∂/∂θ + cot θ∂/∂φ) satisfy
[L+ , L− ] = −2iL3 = 2i∂/∂φ.

Next, we find
P the minimal invariant subspace X1 containing a non-zero function f (θ, φ). If
f (θ, φ) = gm (θ)e , and if gm∗ (θ) 6≡ 0 for some m∗ , applying L± yields functions of the
imφ
∗ ∗
type fk (θ)ei(m +k)φ and hl (θ)ei(m −l)φ , which by our assumption, must be in X1 . Write

ψm (θ, φ) = gm (θ)eimφ

for m = m∗ + 1, m∗ + 2, . . . and for m = m∗ − 1, m∗ − 2, . . ..

Next, requiring X1 to be finite-dimensional implies that there are integers l and l0 such
that L+ ψl = 0 and L− ψl0 = 0.

Examples:

l
(1) Show that ψl (θ, φ) = C eiφ sin θ .
(2) Show that L+ ψm−1 ∼ ψm and L− ψm+1 ∼ ψm .
(3) Write L+ ψm = −iαm ψm+1 and L− ψm+1 = −iβm ψm . Show that L− L+ ψm = −αm βm ψm .
Since this is the only relationship between αm and βm , we can select βm = αm . Show that
2 − α2
αm m−1 = −2m for all m < l, and hence that
p
αm = (l + m + 1)(l − m), (m = l, l − 1, . . . , −l),

which are referred to as Clebsch-Gordan coefficients. In particular α−l−1 = 0.

The functions {ψm (−l ≤ m ≤ l)} span a (2l + 1) dimensional subspace X 2l+1 of X ∞
invariant under L1 , L2 , L3 and under all transformations ρ(g) for all g ∈ SO(3). The
transformations restricted to X 2l+1 are called ρl (g), and constitute finite-dimensional rep-
resentations of SO(3).

32
Associated Legendre Polynomials
Note that ψm ’s defined earlier require an index l as well. The standard analysis of these
functions rely on the variable x = cos θ ∈ [−1, 1] rather than θ ∈ [0, π]. Let us write
(L− )(l−m) ψl (θ, φ) as Plm (x)eimφ . Then the statements of Exercise (3) above can be re-
expressed as

dPlm (x)
p 
+ m imφ i(m+1)φ 2
imx m
L Pl (x)e = e i 1−x +√ Pl (x) , (32)
dx 1 − x2
" (m+1)
#
− (m+1) i(m+1)φ imφ
p dP l (x) i(m + 1)x (m+1)(x)
L Pl (x)e = e −i 1 − x2 + √ P .
dx 1 − x2 l

Examples:
(1) Writing P ≡ Plm (x), show that

d2 P m2
  
− + imφ imφ 2 dP 2
L L Pe = −e −(1 − x ) 2 + 2x + −m − m + .
dx dx 1 − x2

The associated Legendre polynomials are rescaled to tesseral harmonics (or spherical harmonics)
R 2π Rπ ∗ (θ, φ)Y
Yl,m (θ, φ) so that 0 dφ 0 dθ sin θ Yl,m k,m (θ, φ) = δk,l ; specifically
s
(2l + 1) (l − m)! m
Yl,m (θ, φ) = (−1)m P (cos θ)eimφ , −l ≤ m ≤ l. (33)
4π (l + m)! l

The key point is that any function f (θ, φ) within the invariant subspace can be expressed
as a linear combination of the corresponding spherical harmonics. Spherical harmonics are
complete.

Excercises:
(1) (2) For the homogeneous space given by the unit circle, show that in polar coordinates

∂ ∂
L1 = sin φ + cot θ cos φ
∂θ ∂φ
∂ ∂
L2 = − cos φ + cot θ sin φ
∂θ ∂φ
∂
L3 = −
∂φ

33
(2) Using Example (3) above, show that

d2 Plm (x) dPlm (x) m2

 
2
(1 − x ) − 2x + l(l + 1) − Plm (x) = 0. (34)
dx2 dx (1 − x2 )

Plm (x) are known as the associated Legendre polynomials.

(3) Derive the orthogonality
Z 1
2(l + m)!
Pkm (x)Plm (x) dx = δk,l . (35)
−1 (2l + 1)(l − m)!

(4) Express all Plm (x) for l = 1, 2, 3 and plot them.

(5) Write all spherical harmonics for l = 1, 2, 3, 4 and plot them.

34
 RIGID MOTIONS AND BESSEL FUNCTIONS

Rigid motions (∈ Mn ) in Vn consists of a rotation x → Rx [R ∈ SO(n)] followed by a

translation x → x + ζ, ζ being a constant vector. The following inclusions are apparent:
(1) SO(2) ⊂ M2 and (2) SO(3) ⊂ M3 .
An element of M2 is a mapping g = gξ,η,θ given by
   0  
x x x cos θ − y sin θ + ξ
g: → = .
y y0 x sin θ + y cos θ + η

One can also consider an equivalent homogeneous representation

 
cos θ − sin θ ξ
g ↔  sin θ cos θ η  .
0 0 1

Examples:
(1) Verify that the composition g2 g1 two mappings is associated with the product of the
corresponding 3 × 3 matrices.

Representations of M2
Let f (x, y) ∈ X ∞ and note that the plane is a homogeneous space for M2 . A representation
is obtained by transforming each f (x, y) into

(ρ(g)f ) (x, y) = f ([x − ξ] cos θ + [y − η] sin θ, −[x − ξ] sin θ + [y − η] cos θ) .

The inifinitesimal operators are

∂
(L1 f )(x, y) = (ρ(g)f )(x, y)|ξ=η=θ=0 ,
∂ξ
etc.

Examples:
(1) Show that

∂ ∂ ∂ ∂
L1 = − ; L2 = − ; L3 = y −x ,
∂x ∂y ∂x ∂y

35
and with L± = L1 ± iL2
 
∂ ∂ i ∂
L3 = − ; L± = e±iφ ± .
∂φ ∂r r ∂φ

Show also that L+ L− = L− L+ = ∇2 .

Some Irreducible Representations

To search for an invariant subspace X1 that is as small as possible, note that X1 must
be transformed into itself under L1 , L2 and L3 . By Fourier series expansion the releavnt
functions are of the form
ψm (x, y) = i−m eimφ gm (r),
where gm is a smooth function on (0, ∞). ψm is transformed to itself by L3 , while L+ ψm
is of the form ei(m+1)φ h(r); further, L− L+ φm is of the form g̃(r)eimφ . Let us check if we
can express g̃(r) = −αm2 g (r); i.e., we wish to choose g (r) so that
m m

L+ ψm = iαm ψm+1 ; L− ψm+1 = iαm ψm .

Since L+ and L− commute, αm

2 = α2
m−1 , and all α’s can be taken to b e equal. Consequently

∇2 ψm = −α2 ψm ,

for all m. The resulting ψm ’s are proportional to Bessel functions Jm (αr), which satisfies
the Bessel differential equation

d2 m2
 
1 d
+ + 1 − Jm (z) = 0. (36)
dz 2 z dz z2

Series Solutions of Bessel Functions of the First Kind

The following approach is referred to as the Frobenius Method: in order to construct a
series solution for Jm (z) write
∞
X
Jm (z) = ak z k+α ,
k=0

with a0 6= 0. The solution will be singular at z = 0 if α < 0.

Examples:
(1) By using the two smallest powers of z, show that α2 = m2 (and hence α = ±m) and

36
that a1 = 0. 
(2) Show that (k + α + 2)2 − m2 ak+2 = −ak .


(3) When α = m, show that (with an appropriate normalization)

zm (z/2)2 (z/2)4
 
Jm (z) = 1− + + ...
2m Γ(m + 1) (m + 1) 2 × (m + 1)(m + 2)
∞
X (−1)k  z m+2k
=
k!(m + k)! 2
k=0

What is the normalization?

(4) When α = −m, show that (with an appropriate normalization and starting term)
∞
m
X (−1)k  z 2k+m
J−m (z) = (−1)
k!(k + m)! 2
k=0

[Careful here. How can you avoid the series diverging?] Note that Jm (z) and J−m (z) are
not independent. Hence we need to include another class of Bessel functions in order to
form a complete set. The additional functions are Neumann functions defined as

Jν (z) cos νπ − J−ν (z)

Nν (z) ≡ . (37)
sin νπ

Excercises:
(1) Normalize the functions using ψ0 (0, 0) = 1. The Bessel function of order m is defined
as Jm (z) = gm (z/α), so that ψm (x, y) = i−m eimφ Jm (αr). show that
 
d m
− Jm (z) = −Jm+1 (z)
dz z
 
d m+1
+ Jm+1 (z) = Jm (z),
dz z

and hence derive the Bessel differential equation

d2 m2
 
1 d
+ + 1 − Jm (z) = 0.
dz 2 z dz z2

(2) Show that Z ∞

1
Jn (z) = dt eiz sin t+int .
2π −∞

37
Deduce that a generating function for Bessel functions is
   +∞
kr 1 X
exp t− = Jm (kr)tm .
2 t m=−∞

(3) Show that

1  z m
Jm (z) ≈ ,
Γ(m + 1) 2
when z  1.
(4) Plot the first 4 Bessel functions for z ∈ [0, 20). (You can use Matlab or Python.).
Compare the Bessel functions and the approximation for z ∈ [0, 1).
(5) Starting from the expansion of Example (3), show that
 
d Jm (z) Jm+1 (z)
m
=− ,
dx z zm

and
d m
[z Jm (z)] = z m Jm−1 (z).
dx
Deduce that J1 (z) = −J00 (z), Jm+1 (z)+Jm−1 (z) = (2m/z)Jm (z), and Jm+1 (z)−Jm−1 (z) =
0 (z).
−2Jm

38
 GROUP REPRESENTATIONS AND QUANTUM MECHANICS

We showed earlier the 2-to-1 mappings of SU (2) → SO(3) and of SL(2, C) → Lp . Some
quantum mechanical objects transform under rotations as components of spinors. The more
appropriate representations for these quantities are in SU (2) rather than SO(3). Similarly,
under Lorentz transformations, spinor components give representations in SL(2, C). SU (2)
and SL(2, C) determine the ray-representation, where the (quantum) state of the system
is defined, not by a single vector ψ, but to a ray {αψ}, of all numerical multiples of ψ. If
the wave function needs to be normalized |α| = 1. There are many subtle facets of the ray
representations which will not be followed here.

A Representation of SU (2)
Let {ω1 , ω2 , ω3 } be the intrinsic coordinates in SO(3) and let gω1 ,ω2 ,ω3 be the corresponding
rotation matrix in SO(3). Let ±uω1 ,ω2 ,ω3 be the elements in SU (2) that are mapped to
gω1 ,ω2 ,ω3 by the homomorphism given in (25). One can select
 
cos ω/2 −i sin ω/2
uω,0,0 = ,
−i sin ω/2 cos ω/2
 
cos ω/2 − sin ω/2
u0,ω,0 = ,
sin ω/2 cos ω/2
 
exp[−iω/2] 0
u0,0,ω = ,
0 exp[iω/2]
(38)

Excercises:
(1) Show that the corresponding gω,0,0 , g0,ω,0 and g0,0,ω are (intrinsic) rotations about the
three axes.
(2) Show that the infinitesimal group elements of SU (2) are
 
∂ 1 0 −i
T1 = uω,0,0 |ω=0 =
∂ω 2 −i 0
 
∂ 1 0 −1
T2 = u0,ω,0 |ω=0 =
∂ω 2 1 0
 
∂ 1 −i 0
T3 = u0,0,ω |ω=0 = .
∂ω 2 0 i

39
(3) Show that the corresponding differential operators are
 
∂ i ∂ ∂
L1 = ρ(uω,0,0 )|ω=0 = x2 + x1
∂ω 2 ∂x1 ∂x2
 
∂ 1 ∂ ∂
L2 = ρ(u0,ω,0 |ω=0 ) = x2 − x1
∂ω 2 ∂x1 ∂x2
 
∂ i ∂ ∂
L3 = ρ(u0,0,ω |ω=0 ) = x1 − x2 .
∂ω 2 ∂x1 ∂x2

(4) Show that (a) [Ti , Tj ] = εijk Tk and (b) [Li , Lj ] = εijk Lk .

40
 .

STURM-LIOUVILLE THEORY

Texts:
(1) “Mathematics for Physicists,” S. M. Lea, Thomson Books/Cole

41
 STURM-LIOUVILLE THEORY

The Sturn-Liouville problem can be stated as follows: Given a weight function w(x) where
x ∈ [a, b], consider a function y(x) that satisfies the linear ordinary differential equation
 
d dy
f (x) − g(x)y + λw(x)y = 0, (39)
dx dx
along with boundary conditions

dy
α1 y(a) + β1 (a) = 0,
dx
dy
α2 y(b) + β2 (b) = 0. (40)
dx
We wish to determine the eigenvalues λ for which there are non-trivial solutions y(x).
α1 = 0 = α2 correspond to Neumann boundary conditions and β1 = 0 = β2 to Dirichlet
boundary conditions.

Examples:
(1) Verify that the Lagrange equation and the Bessel equation, with appropriate boundary
conditions, are examples of the Sturm-Liouville problem.
(2) Show that a general second-order ODE can be reduced to the form of (39) (Hint: use
an integration factor).

Theorem: The eigenfunctions ym (x) and yn (x) of distinct eigenvalues λm and λn are
orthogonal with respect to the weight w(x); i.e.,
Z b
w(x)ym (x)yn (x) dx = 0 for λm 6= λn . (41)
a

Proof: Integrate the equations by parts and use the boundary conditions.
In addition, the eigenfunctions form a complete basis. Thus, any smooth function in
[a, b] can be expanded as
∞
X
f (x) = an yn (x),
n=0
with Rb
f (x)ym (x)w(x) dx
am = aR b
2
a [ym (x)] w(x) dx

42
Examples:
(1) Show that the completeness of the basis implies that
∞
X yn (x)yn (x0 )w(x0 )
Rb = δ(x − x0 ).
2
n=0 a [ym (x)] w(x) dx

Theorem: The eigenvalues λm are real.

Proof: Once again use integration by parts and boundary conditions to show that λn = λ∗n .

Theorem: The Sturm-Liouville operator

 
d dy
Ly ≡ f (x) − g(x)y,
dx dx
is self-adjoint; i.e.,
Z b Z b
yn Lym dx = ym Lyn dx.
a a

Excercises:
(1) A hollow copper sphere of radius a is divided into two halves at the equator by a thin
insulating strip. The top and bottom halves are held at potentials V0 and 0. What is the
potential inside the sphere?
(2) A cylindrical shell of radius a and height h has its curved surface and its bottom
grounded, while the top surface is kept at a potential V0 . Find the potential inside the
cylinder.
(3) Quantum mechanical treatment of the harmonic oscillator results in the Hermite
differential equation
y 00 − 2xy 0 + λy = 0. (42)
Write the equation in the standard Sturm-Liouville form. If the boundary conditions are
y(x) → 0 as x → ±∞, show that the solutions are orthogonal in the range (−∞, ∞),
and find the weight function w(x). Solve the equation to find a series expansion for the
Hermite functions. What value of the eigenvalue λ is required for the function to remain
bounded throughout the interval, including x → ±∞? Normalize the solutions by choos-
ing the coefficient of the highest power xn to be 2n , and hence determine the first three
eigenfunctions.

43