Lectures

Summary of Lecture #1: fundamental concepts
Goals and Requirements:
• Reflect on what you have learned about the basic objects in quantum mechanics:
wavefunctions & operators.
• Establish the basic picture about the math structure of quantum mechanics
(NOTE: these are not mathematically rigorous)
Hilbert space H linear space equipped with an inner product

quantum states (‘ket’) |ψi elements in the linear space H
‘bra’ hψ| linear functionals defined on H: H 7→ ❈
quantum mechanical operators linear mappings: H1 7→ H2 .
• Be familiarized with the general description of a quantum state: the density matrix.
Get some taste of quantum entropy and quantum entanglement if time permits.
• By the end of this lecture, you should feel comfortable about

dealing with abstract quantum states without reference to the wavefunctions, and
dealing with abstract quantum operators without reference to matrices.
• NOTE: statements with ⋆ are advanced topics/challenge questions/extra exercises

(NOT required).
References:
J.J. Sakurai, Modern Quantum Mechanics, Chapter 1.
P.A.M. Dirac, The Principle of Quantum Mechanics, Sections I.5-6, Chapter III.
J. von Neumann, Mathematical Foundations of Quantum Mechanics, Chapters I,II.
Advanced Quantum Mechanics, Fall 2018 1/15

I THE HILBERT SPACE
I. THE HILBERT SPACE
A. The Wavefunction
• ψ(x) is a complex-valued function defined on some “coordinate” space V (x ∈ V ).

R
• Strictly speaking, ψ(x) shall be normalizable: |ψ(x)|2 dV < ∞.
2
• Probability of the system being in “volume element” dV is R|ψ(x)| dV . − Max Born.
|ψ(x)|2 dV
• Normalized ψ(x) has “dimension”(unit) of “volume”−1/2 (usually not dimensionless).
• Be careful when you parametrize the coordinate space. You might need to absorb the
Jacobian into the wavefunction (depending on the definition of dV ), and/or introduce
artificial “boundary conditions”.
• We may need non-normalizable wavefunctions when the space V is not compact

R
( dV = ∞), e.g. plane waves in open space. We usually regularize this problem
by first taking a finite V , and finally taking the limit that the volume goes to infinity.
• Most wavefunctions we will deal with are continuous, and (piecewise) smooth.
• Example:
V is the unit sphere S 2 parametrized by polar and
azimuth angles x = (θ, ϕ) ∈ [0, π] × [0, 2π], dV = sin θ dθdϕ,
legitimate ψ(θ, ϕ) are normalizable complex functions
R 2π R π
( ϕ=0 θ=0 |ψ(θ, ϕ)|2 sin θ dθdϕ < ∞)
with the “boundary condition” ψ(θ, 0) = ψ(θ, 2π), ∀θ
and ψ(0, ϕ) = ψ(0, 0), ψ(π, ϕ) = ψ(π, 0), ∀ϕ.
A basis of such wavefunctions are spherical harmonics Yℓm (θ, φ).
B. The Hilbert Space
• The Hilbert space H(V ) defined on a coordinate space V is the complex linear space
formed by normalizable wavefunctions defined on V .
– Being a linear space: if ψ1 and ψ2 are elements of H (legitimate wavefunc.), then

so does λ1 ψ1 + λ2 ψ2 , for any complex numbers λ1 and λ2 . Exercise: prove this.

I THE HILBERT SPACE
• There is a natural inner product(“overlap”) of two wavefunctions φ and ψ,

R
(φ, ψ) = φ∗ (x) ψ(x) dV , satisfying the three important “axioms”
– Hermiticity: (φ, ψ) = (ψ, φ)∗ .
– Linearity: (φ, λ1 ψ1 + λ2 ψ2 ) = λ1 · (φ, ψ1 ) + λ2 · (φ, ψ2 ), for λ1 , λ2 ∈ ❈.

∗ Above two properties lead to, (λ1 φ1 + λ2 φ2 , ψ) = λ∗1 · (φ1 , ψ) + λ∗2 · (φ2 , ψ).
– Positive definiteness: (ψ, ψ) ≥ 0, and (ψ, ψ) = 0 if and only if “ψ = 0”.

P P P
• One consequence: ( i λi ψ i , j λj ψ j ) = i,j λ∗i · (ψi , ψj ) · λj ≥ 0 ⇒
the Gram matrix (ψi , ψj ) (where i is row-index, j is column-index) is Hermitian and
positive semi-definite.
– The Gram determinant det[(ψi , ψj )] ≥ 0.
– n = 2 case is the Cauchy-Schwarz inequality, (ψ1 , ψ1 )(ψ2 , ψ2 ) ≥ |(ψ1 , ψ2 )|2 .
– The states ψi are linearly dependent if and only if the matrix (ψi , ψj ) is singular,
or equivalently det[(ψi , ψj )] = 0.
C. Combining Hilbert Spaces: Direct Sum & Tensor Product
• Direct sum of two Hilbert spaces 

H1 (V1 ) ⊕ H2 (V2 ):
 ψ (x), x ∈ V ;
1 1
the wavefunction (ψ1 ⊕ ψ2 )(x) =
 ψ2 (x), x ∈ V2 .
The inner product becomes (φ1 ⊕ φ2 , ψ1 ⊕ ψ2 )V1 ⊕V2 = (φ1 , ψ1 )V1 + (φ2 , ψ2 )V2 =
R ∗ R
φ1 ψ1 dV1 + φ∗2 ψ2 dV2 .
– Note: you need consistent definition of the “volumes” of V1,2 .
• Tensor product of two Hilbert spaces H1 (V1 ) ⊗ H2 (V2 ):

the wavefunction (ψ1 ⊗ ψ2 )(x1 , x2 ) = ψ1 (x1 ) · ψ2 (x2 ), for x1 ∈ V1 and x2 ∈ V2 .
The inner product becomes (φ1 ⊗ φ2 , ψ1 ⊗ ψ2 )V1 ⊗V2 = (φ1 , ψ1 )V1 · (φ2 , ψ2 )V2 .
– Entanglement: wavefunctions in H1 ⊗ H2 may not be a direct product ψ1 ⊗ ψ2 .
– Identical particles: will be treated later.

I THE HILBERT SPACE
• Example:
direct sum: vs. tensor product:
one particle in two potential wells two inequivalent particles in two wells
need to know if it is in left or right need to know both left & right particles’ state
H1 ⊕ H2 H1 ⊗ H 2
D. The Dirac Notation
• ‘kets’ |ψi: element(quantum state) in Hilbert space corresponding to wavefunction ψ.
• ‘bras’ hψ|: linear functional defined on the Hilbert space: H 7→ ❈, φ 7→ (ψ, φ).
R
– Short-hand notation: hψ|φi ≡ (ψ, φ) = ψ ∗ φ dV .
– ‘bra’ is a linear functional: hψ|λ1 φ1 + λ2 φ2 i = λ1 hψ|φ1 i + λ2 hψ|φ2 i.
– ‘bras’ form an anti-linear space: λ∗1 hψ1 | + λ∗2 hψ2 | = hλ1 ψ1 + λ2 ψ2 |.
– Any ‘continuous’ linear functional f : H 7→ ❈, φ 7→ f (φ), corresponds to a

wavefunction ψf so that f = hψf |, f (φ) = (ψf , φ). − Riesz-Fréchet theorem.
P
With complete orthonormal basis |ei i, hψf | = i f (ei ) hei |,
P
so |ψf i = i f (ei )∗ |ei i.
• Short-hand notation: |ψ1 i + |ψ2 i with ψ1 ∈ H1 , ψ2 ∈ H2 means the direct sum state
ψ 1 ⊕ ψ 2 ∈ H1 ⊕ H 2 .
• Short-hand notation: |ψ1 i|ψ2 i with ψ1 ∈ H1 , ψ2 ∈ H2 means the tensor product state
ψ 1 ⊗ ψ 2 ∈ H1 ⊗ H 2 .
• Short-hand notation: |ψ1 ihψ2 | with ψ1 ∈ H1 , ψ2 ∈ H2 is a linear operator: H2 7→

H1 , φ 7→ (ψ2 , φ)ψ1 ≡ |ψ1 ihψ2 |φi.
• Other labels like quantum numbers or just an index, are often used in ‘bras’ & ‘kets’:
e.g. |L = 2, Lz = 0i, |0i.

I THE HILBERT SPACE
E. Complete Orthonormal Basis
• (Discrete) Orthonormal basis ei (i = 0, 1, . . . ) satisfy hei |ej i = δij .

P
• Complete orthonormal basis: for any ψ ∈ H, |ψi = i |ei ihei |ψi.
– For finite dimensional Hilbert space, this is just

‘number of orthonormal basis’=‘dimension of Hilbert space’.
– For infinite dimensional Hilbert space, completeness is usually very hard to prove.
P
• Resolution of identity: ✶ = i |ei ihei |, the sum is over a complete orthonormal basis.
– We will see the resolution of identity in terms of overcomplete basis later.

P P P
– Application: change of basis, for |ψi = j c̃j |ẽj i = i ci |ei i = i,j ci |ẽj ihẽj |ei i,
P
coefficients c̃j = i ci hẽj |ei i, where ẽj are another complete orthonormal basis.
P
∗ The matrix Uji = hẽj |ei i is a unitary matrix. (U · U † )jk = i Uji (U † )ik =
P
i hẽj |ei ihei |ẽk i = hẽj |ẽk i (by resolution of identity) = δjk = (✶)jk . Above
P
relation is c̃j = i Uji ci or c̃ = U · c in short form.
∗ Conversely, given a unitary matrix U and complete orthonormal basis |ei i,
P
then |ẽj i ≡ i Uji |ei i form a new set of complete orthonormal basis.
Exercise: is the “U ” here same as the “U ” in previous item?
• Example: Fourier series.

Particle moving on a ring parametrized by angle θ, legitimate
wavefunctions are normalizable ψ(θ) with period 2π, ψ(θ + 2π) = ψ(θ).
A complete orthonormal basis is en (θ) = √1
2π
exp(✐nθ) for n ∈ ❩.
• Basis of composite Hilbert space:

if |ei i (i = 1, . . . , n) are the basis of H1 , and |e′j i (j = 1, . . . , m) are the basis of H2 ,
– the (n + m) basis of H1 ⊕ H2 can be chosen as

(|e1 i, |e2 i, . . . , |en i, |e′1 i, |e′2 i, . . . , |e′m i).
– the (n × m) basis of H1 ⊗ H2 can be chosen as |ei i ⊗ |e′j i, namely

(|e1 i|e′1 i, |e1 i|e′2 i, . . . , |e1 i|e′m i, |e2 i|e′1 i, . . . , |e2 i|e′m i, · · · , |en i|e′1 i, . . . , |en i|e′m i).

II QUANTUM MECHANICAL OPERATORS
II. QUANTUM MECHANICAL OPERATORS
A. Quantum Mechanical Operators
• Linear operators: linear mappings between two (often the same) Hilbert spaces:
Ô|ψi ∈ H2 for |ψi ∈ H1 , and Ô|λ1 ψ1 + λ2 ψ2 i = λ1 Ô|ψ1 i + λ2 Ô|ψ2 i.
• Anti-linear operators: replace the last condition of linear operators by

Ô|λ1 ψ1 + λ2 ψ2 i = λ∗1 Ô|ψ1 i + λ∗2 Ô|ψ2 i.
– Example: the operator of “taking complex conjugate” K: φ(x) 7→ φ(x)∗ .
• Hermitian conjugate (adjoint) of linear operators: Ô† is a linear operator satisfying

(Ô† ψ, φ) = (ψ, Ôφ) for any ψ & φ, or hÔ† ψ| = hψ|Ô for any ψ.
– (Ô† )† = Ô.
“Proof”: for any ψ & φ, by definitions of inner product and hermitian conjugate,
(ψ, ((Ô† )† )φ) = (((Ô† )† )φ, ψ)∗ = (φ, (Ô† )ψ)∗ = ((Ô† )ψ, φ) = (ψ, Ôφ). Then
((Ô† )† ) and Ô must be the same.
– (λÔ)† = λ∗ Ô† , (ÂB̂)† = B̂ † Â† .

Exercise: try to “prove” these as the “proof” for the previous relation.
– Hermitian operators: those satisfy Ô† = Ô.

Anti-Hermitian operators: Ô† = −Ô.
– Any operator is the sum of its Hermitian & anti-Hermitian part:

Ô+Ô† Ô−Ô†
Ô = 2
+ 2
, the 1st term is Hermitian, 2nd term is anti-Hermitian.
• Matrix representation: under a complete orthonormal basis |ni, the operator Ô has
‘matrix elements’ Omn ≡ hm|Ô|ni. (O† )mn = (Onm )∗ .
– Matrix represention under non-orthogonal/overcomplete basis can also be useful.
– Expectation value of Ô in state ψ: hψ|Ô|ψi/hψ|ψi.
P
• Trace: TrÔ = n hn|Ô|ni, summing over a complete orthonormal basis.
The result of trace is independent of the choice of basis.
Cyclic property: Tr(ÂB̂) = Tr(B̂ Â), for ‘finite’ operators Â, B̂ (e.g. finite dimensional)

• Eigenvalue λ and eigenstate |Ô = λi of operator Ô: defined by Ô|Ô = λi = λ|Ô = λi.
– Eigenvalues of Hermitian operators are real.
• (Not required) Singular value decomposition (SVD):
P
– Any operator can be written as Ô = n |ñi ρn hn|, where n labels the singular
value ρn ≥ 0, and the two sets of orthonormal basis |ni & |ñi are eigenstates of
Ô† Ô & ÔÔ† respectively.
– In complete orthonormal basis |ei i, the above relations becomes

P
Oij = hei |Ô|ej i = n hei |ñi ρn hn|ej i = (U · ρ · V † )ij ,
where the unitary matrices Uin = hei |ñi and Vjn = hej |ni,
and the diagonal matrix ρ has diagonal elements ρn . And
P
(O† O)ij = hei |Ô† Ô|ej i = n hei |ni ρ2n hn|ej i = (V · ρ2 · V † )ij , and
P
(OO† )ij = hei |ÔÔ† |ej i = n hei |ñi ρ2n hñ|ej i = (U · ρ2 · U † )ij .
• Projection operators: operators P̂ : H 7→ H, satisfying P̂ P̂ = P̂ .

P
– Hermitian projection operators P̂ = i |ei ihei |, ei are a set of orthonormal basis,
and P̂ have eigenvalues 1 and 0 only.
– ✶ − P̂ is also a projection operator. Exercise: check that (✶ − P̂ )(✶ − P̂ ) = (✶ − P̂ ).
• Inverse of an operator: Â−1 must satisfy Â−1 · Â = ✶ and Â · Â−1 = ✶
– ⋆ In infinite dimensional Hilbert space, there can be cases with B̂ Â = ✶ while

ÂB̂ 6= ✶, B̂ shall not be called Â−1 .
• Unitary operators: linear operators with

(Û ψ, Û φ) = (ψ, φ), ∀ψ, φ. Or equivalently Û † Û = ✶.
P
– Unitary operators are of the form i |ẽi ihei |, where ei is a set of complete or-
thonormal basis, ẽi is another set of orthonormal basis.
– If Ĥ is Hermitian, then exp(✐Ĥ) is unitary. Exercise: is the converse true?
• Anti-unitary operators: anti-linear operators with

(Û ψ, Û φ) = (ψ, φ)∗ , ∀ψ, φ.

B. Abstract Calculations with Operators
• Commutator & anti-commutator of Â & B̂: [Â, B̂] ≡ ÂB̂ − B̂ Â, {Â, B̂} ≡ ÂB̂ + B̂ Â.
– For notation simplicity, define ‘Lie derivative’ LÂ B̂ ≡ [Â, B̂].
• Elementary functions of operators may be defined by their power series expansion, e.g.
P
exp(Â) = ∞ n
n=0 (Â) /n! (let’s not worry about convergence).
– Note: Â · f (B̂) · Â−1 = f (Â · B̂ · Â−1 ) for such functions f that can be defined as
power series, because Â · (B̂)n · Â−1 = (Â · B̂ · Â−1 )n .
• Jacobi identity: [Â, [B̂, Ĉ]] + [B̂, [Ĉ, Â]] + [Ĉ, [Â, B̂]] = 0. Or [LÂ , LB̂ ]Ĉ = L[Â,B̂] Ĉ.
• ‘Leibniz’s rule’:
[Â, B̂1 B̂2 · · · B̂n ] = [Â, B̂1 ]B̂2 · · · B̂n + B̂1 [Â, B̂2 ] · · · B̂n + · · · + B̂1 B̂2 · · · [Â, B̂n ]. Or
LÂ (B̂1 B̂2 · · · B̂n ) = (LÂ B̂1 )B̂2 · · · B̂n + B̂1 (LÂ B̂2 ) · · · B̂n + · · · + B̂1 B̂2 · · · (LÂ B̂n ).
• Baker-Hausdorff formula: eÂ B̂e−Â = B̂ + [Â, B̂]/1! + [Â, [Â, B̂]]/2! + . . .

Or formally eÂ B̂e−Â = exp(LÂ )B̂.
– A heuristic “proof”:
define fˆ(t) = etÂ B̂e−tÂ , then fˆ(0) = B̂.
Take derivative with respect to t, note that ddt etÂ = ÂetÂ = etÂ Â,
then d fˆ(t) = Â · fˆ(t) − fˆ(t) · Â = [Â, fˆ(t)] = L fˆ(t).
dt Â
The formal solution of this ordinary differential equation is then fˆ(t) = etLÂ fˆ(0),
so eÂ B̂e−Â = fˆ(1) = exp(L )B̂. Â
• Direct sum & tensor product of operators are defined similarly to wavefunctions:
for operator Â defined on H1 , and B̂ defined on H2 , Â ⊗ B̂ is an operator defined on
H1 ⊗ H2 , such that (Â ⊗ B̂)|ψ ⊗ φi = (Â|ψi) ⊗ (B̂|φi), for states ψ ∈ H1 and φ ∈ H2 .
– (Â ⊗ B̂) · (Ĉ ⊗ D̂) = ÂĈ ⊗ B̂ D̂.
– When Â is referred to within Ĥ1 ⊗ Ĥ2 , it usually means Â ⊗ ✶. With this

convention, Â⊗ B̂ is usually written as ÂB̂, which means (Â⊗ ✶)·(✶ ⊗ B̂) = Â⊗ B̂.
– Tr1⊗2 (Â ⊗ B̂) = Tr1 (Â) · Tr2 (B̂), where the three different traces are taken in
Hilbert spaces H1 ⊗ H2 , H1 , H2 , respectively.

– With complete orthonormal basis, |ei i ∈ H1 (i = 1, . . . , n) and |e′j i ∈ H2 (j =

1, . . . , m), the matrix representation of Â ⊗ B̂ is a (n × m)-row (n × m)-column
matrix, (Â ⊗ B̂)(i,j)(i′ ,j ′ ) ≡ hei |he′j |Â ⊗ B̂|ei′ i|e′j ′ i = hei |Â|ei′ i · he′j |B̂|e′j ′ i = Aii′ Bjj ′ .
The combination (i, j)[(i′ , j ′ )] is the row[column] index (i, i′ = 1, . . . , n and j, j ′ =
1, . . . , m).
C. Back to Wavefunction
• The coordinate operator x̂: φ(x) 7→ x · φ(x). It is obviously Hermitian.
– Worry #1: x · φ(x) may not be normalizable!
– Worry #2: What are the eigenstate wavefunctions of x̂? Are they normalizable?
• Despite the above worries, denote the eigenstates of x̂ by |xi, i.e. x̂|xi = x|xi.
– ‘Normalization’: hx′ |xi = δ(x′ − x), where δ is the Dirac-δ ‘function’.

R
– Resolution of identity: ✶ = |xihx| dx.
– The wavefunction ψ(x): expansion coefficients of state ψ in the basis |xi.

R
ψ(x) = hx|ψi. And |ψi = ψ(x)|xi dx.
• The momentum operator p̂: φ(x) 7→ −✐h̄ ∂∂x φ(x). It is not-so-obviously Hermitian.
– Canonical commutation relation [x̂, p̂] = ✐h̄.
– Similar worries as for the coordinate operator.
– Nonetheless, denote the eigenstate of p̂ as |pi, p̂|pi = p|pi.
– ‘Normalization’: hp′ |pi = δ(p′ − p).

R
– Resolution of identity: ✶ = |pihp| dp.
✐px/h̄
e√
– hx|pi = 2πh̄
for 1D infinite space.
• ⋆ Construct examples of normalizable φ(x) so that x̂ φ(x) or p̂ φ(x) is not normalizable.
• IMPORTANT: h̄ will be frequently omitted hereafter.

III DENSITY MATRIX & ENTANGLEMENT
III. DENSITY MATRIX & ENTANGLEMENT
A. Density Matrix
• Density matrix of a normalized ‘pure state’ ψ: ρ̂ψ = |ψihψ| is a projection operator.
– Expectation value of Ô in ψ is hψ|Ô|ψi = Tr(ρ̂ψ Ô) = Tr(Ô ρ̂ψ ).
– ρ̂ is independent of the complex phase of |ψi, is a ‘better’ description of the state.
• Generic density matrix ρ: linear Hermitian non-negative operator of trace unity.

ρ̂† = ρ̂; hφ|ρ̂|φi ≥ 0, ∀φ; and Tr(ρ̂) = 1.
P P
– ρ= i λi |ei ihei |, with some orthonormal basis ei , and λi > 0, i λi = 1.
– Expectation value of Ô in generic ‘mixed state’ is Tr(ρ̂ Ô).

P
– If ρ̂j are density matrices, and cj > 0, and j cj = 1,
P
then j cj ρ̂j is also a density matrix.
b at finite temperature T :
• The density matrix of Hamiltonian H
b B T )/Z = P exp(−Ei /kB T ) |Ei ihEi |,
ρ = exp(−H/k Ei Z
where Ei are eigenvalues, |Ei i are corresponding eigenstates,

b B T )] = P exp(−Ei /kB T ).
Z = Tr[exp(−H/k Ei
B. ⋆ Some Quantum Information Basics (not required)
P
• von Neumann entropy of a density matrix ρ̂: S ≡ −Tr(ρ̂ ln ρ̂) = − i λi ln λi .
– Pure states have zero entropy & ρ̂2 = ρ̂.

Mixed states have positive entropy & ρ̂2 < ρ̂.
– In n(finite)-dimensional Hilbert space, 0 ≤ S[ρ̂] ≤ ln(n).

ln[Tr(ρ̂n )]
– Rényi entropy: Sn ≡ 1−n
. Note: formally limn→1 Sn = S.
• ⋆ ⋆ ⋆ Concavity of von Neumann entropy: mixing two systems increases entropy.

S[λ ρ̂1 + (1 − λ) ρ̂2 ] ≥ λ S[ρ̂1 ] + (1 − λ) S[ρ̂2 ],
for two density matrices ρ̂1,2 and 0 < λ < 1.

III DENSITY MATRIX & ENTANGLEMENT
• Reduced density matrix: given a density matrix ρ̂ on Ha ⊗ Hb , reduced density matrix

ρa on Ha is ρ̂a = Trb (ρ̂), obtained by taking partial trace over Hb .
P
– Meaning of partial trace: for any ψ1,2 ∈ Ha , hψ1 |ρ̂a |ψ2 i = i hψ1 ⊗ φi |ρ̂|ψ2 ⊗ φi i,
and the sum is over a complete orthonormal basis φi of Hb . The matrix elements
of ρ̂a under a orthonormal basis Ha can be computed by this relation.
– ⋆ ⋆ ⋆ Subadditivity of entropy: information is ‘lost’ by separating two subsystems.

Sa⊗b [ρ̂] = Tra⊗b (−ρ̂ ln ρ̂) ≤ Sa [ρ̂a ] + Sb [ρ̂b ] = Tra (−ρ̂a ln ρ̂a ) + Trb (−ρ̂b ln ρ̂b ) =
Sa⊗b [ρ̂a ⊗ ρ̂b ].
• Special case: ρ̂ = |ψihψ| is for a normalized pure state ψ ∈ Ha ⊗ Hb .
– “Entanglement entropy”: von Neumann entropy of reduced density matrix:

Sa = −Tra (ρ̂a ln ρ̂a ), where ρ̂a = Trb (ρ̂).
– The degrees of freedom of a and b are ‘entangled’ in this state ψ if Sa > 0.
– The state ψ is a disentangled product state ψa ⊗ φb if and only if Sa = 0.

P
– Schmidt decomposition (just SVD): |ψi = i λi |φi i ⊗ |ϕi i,
where (λi )2 are eigenvalues of ρ̂a , φi (ϕi ) are orthonormal eigenstates of ρ̂a (ρ̂b ).
– ⋆ Reciprocity: define reduced density matrix ρ̂b = Tra (ρ̂) on Hb .

Sb = Trb (−ρ̂b ln ρ̂b ) equals to Sa above for a pure state in Ha ⊗ Hb .
• Example: Bell state.

Ha (Hb ) is 2-dimensional, with orthonormal basis |0i, |1i (|0̃i, |1̃i).
One of the Bell states is √1 (|0i|1̃i − |1i|0̃i).
2
Exercise: write down the reduced density matrices ρ̂a and ρ̂b . Compute the entangle-
ment entropy.

IV MEASUREMENT & THE UNCERTAINTY PRINCIPLE
IV. MEASUREMENT & THE UNCERTAINTY PRINCIPLE
A. Measurement
• Measurement can be done for a Hermitian operator Â on pure or mixed states ρ̂.
• The outcome of the measurement will be eigenvalues of Â.
• The probability of outcome λ is Pλ = Tr(P̂λ ρ̂), P̂λ is the projection to eigenvalue-λ

P
subspace. P̂λ = |Â = λihÂ = λ|, summing over orthonormal eigenstates of Â with
eigenvalue λ.
– If all eigenvalues λ′ of Â are known, then P̂λ can be formally obtained by the
Q
Â−λ′ ✶
“Lagrange interpolating polynomial”, P̂λ = λ′ , λ′ 6=λ λ−λ′ .
– The statistical average of outcome is the expectation value of Â in state ρ̂,

P P P
Tr(Â ρ̂) = Tr[( λ λ P̂λ ) ρ̂] = λ λ Tr(P̂λ ρ̂) = λ λ Pλ .
For pure state ρ̂ = |ψihψ|, this is hψ|Â|ψi.
• The “collapse” postulate:

P̂λ ρ̂ P̂λ
if the measurement outcome is λ, the quantum state will “collapse” to Tr(P̂λ ρ̂)
.
– If eigenvalue-λ eigenstate is unique, this is the familiar statement

that the system collapses to eigenstate |Â = λi.
• ⋆ ⋆ ⋆ Information is gained by measurement: ‘entropy’ decreases.

P
S[ρ̂] ≥ λ Pλ S[ρ̂λ ], Pλ is the probability of outcome λ, ρ̂λ is the collapsed state.
• An example:
– DefinePauli
 matrices
     
1 0 0 1 0 −✐ 1 0
σ 0 =  , σ 1 =  , σ 2 =  , σ 3 =  .
0 1 1 0 ✐ 0 0 −1
– Consider a state described by the density matrix ρ̂, represented in some basis as
 
1 0 0 ✐
 
0 1 − ✐ 0
  1
ρ̂ = 14   = 4 [✶ − σ1 ⊗ σ2 ]. Exercise: is this a pure state?
0
 ✐ 1 0 
−✐ 0 0 1

IV MEASUREMENT & THE UNCERTAINTY PRINCIPLE
– Measure a Hermitian operator Â, represented in the same basis as Â = σ1 ⊗ σ3 .
– Eigenvalues of Â are ±1.

(−✶)−Â ✶−Â .
Corresponding projection operators are P̂+1 = (−1)−1
and P̂−1 = 1−(−1)
– Outcome +1: probability Tr(ρ̂ P̂+1 ) = 1/2,

collapsed state is ρ̂+1 = (1/4)[✶ + σ1 ⊗ σ3 ].
Outcome −1: probability Tr(ρ̂ P̂−1 ) = 1/2,
collapsed state is ρ̂−1 = (1/4)[✶ − σ1 ⊗ σ3 ].
– ⋆ Exercise: compute entropies S[ρ̂], S[ρ̂+1 ] and S[ρ̂−1 ], check if any informa-
tion can be gained by this measurement, namely whether S[ρ̂] > (1/2)S[ρ̂+1 ] +
(1/2)S[ρ̂−1 ] ?
B. The Uncertainty Principle
• For Hermitian Â & B̂, (hÂ2 i − hÂi2 )(hB̂ 2 i − hB̂i2 ) ≥ 14 |h[Â, B̂]i|2 . − W. Heisenberg
h·i is the expectation value under a quantum state ρ̂.
– Rough description: product of variances of measurement outcomes for Â & B̂ is

bounded below by the square of their commutator’s expectation value.
P P
– Variances of measurement outcome: λ Pλ (λ − λ̄)2 = ( λ Pλ λ2 ) − λ̄2 , where λ
P
are eigenvalues, Pλ is the probability of outcome λ, λ̄ = λ Pλ λ is the ‘average’.
– Proof:
Define the inner product of two operators Â, B̂ as (Â, B̂) = hÂ† B̂i = Tr(Â† B̂ ρ̂).
Exercise: check that this indeed satisfies the “axioms” of inner product.
Define two new operators Â′ = Â − hÂi, B̂ ′ = B̂ − hB̂i. For Hermitian Â, B̂,
1
4
|h[Â, B̂]i|2 = 14 |h[Â′ , B̂ ′ ]i|2 = 21 (Â′ , B̂ ′ )(B̂ ′ , Â′ ) − 14 (Â′ , B̂ ′ )2 − 41 (B̂ ′ , Â′ )2
= [Im(Â′ , B̂ ′ )]2 ≤ |(Â′ , B̂ ′ )|2 ≤ (Â′ , Â′ )(B̂ ′ , B̂ ′ ) = (hÂ2 i − |hÂi|2 )(hB̂ 2 i − |hB̂i|2 ).
The last inequality used here is Cauchy-Schwarz.
– Exercise: what is the condition for the equality to be true?
• Familiar case: Â = x̂, B̂ = p̂. hx̂2 − x̄2 ihp̂2 − p̄2 i ≥ h̄2 /4.

A ⋆ ABOUT THE STATEMENTS WITH ⋆ (NOT REQUIRED)
Appendix A: ⋆ About the Statements with ⋆ (not required)
• ⋆ In infinite dimensional Hilbert space, there can be cases with B̂ Â = ✶ while ÂB̂ 6= ✶,
B̂ shall not be called Â−1 .
– Example:
Assume |0i, |1i, . . . are complete orthonormal basis.
P
Define Â = ∞ n=0 |n + 1ihn| = |1ih0| + |2ih1| + . . . .
P P∞
Consider B̂ = ∞ n=0 |nihn + 1|, then B̂ · Â = n=0 |nihn| = ✶.
P∞
However Â · B̂ = n=0 |n + 1ihn + 1| = ✶ − |n = 0ihn = 0| 6= ✶.
• ⋆ Construct examples of normalizable φ(x) so that x̂ φ(x) or p̂ φ(x) is not normalizable.
– Example:
sin(x3 )
φ(x) = x
, defined on real axis of x.
Check that φ is normalizable while both x̂ φ and p̂ φ are not.
• ⋆ ⋆ ⋆ Concavity of von Neumann entropy: S[λ ρ̂1 + (1 − λ) ρ̂2 ] ≥ λ S[ρ̂1 ] + (1 − λ) S[ρ̂2 ],

for two density matrices ρ̂1,2 and 0 < λ < 1.
– Proof: see e.g., M.A. Nielsen, I.L. Chuang, Quantum Computation and Quantum
Information, section 11.3.5.
• ⋆ ⋆ ⋆ Subadditivity of entropy: Sa⊕b [ρ̂] = Tra⊕b (−ρ̂ ln ρ̂) ≤ Sa [ρ̂a ] + Sb [ρ̂b ] =

Tra (−ρ̂a ln ρ̂a ) + Trb (−ρ̂b ln ρ̂b ) = Sa⊕b [ρ̂a ⊗ ρ̂b ].
– Proof: see H. Araki, E. H. Lieb, Commun. Math. Phys. 18, 160 (1970).
• ⋆ Reciprocity: define reduced density matrices ρ̂a = Trb (ρ̂) & ρ̂b = Tra (ρ̂) on subspace
Ha & Hb respectively, where ρ = |ψihψ| is a pure state on Ha ⊗ Hb . Then Sb =
Trb (−ρ̂b ln ρ̂b ) = Sa = Tra (−ρ̂a ln ρ̂a ).
– Proof:
This a simple consequence of the Schmidt decomposition of a pure state.
P
|ψi = i λi |ei i⊗|ẽi i, with orthonormal basis ei for Ha and ẽi for Hb , and real pos-
P
itive singular values λi . Then the reduced density matrix on Ha is i λ2i |ei ihei |,
P P
and on Hb is i λ2i |ẽi ihẽi |. So Sa = − i λ2i ln(λ2i ) = Sb .

A ⋆ ABOUT THE STATEMENTS WITH ⋆ (NOT REQUIRED)
• ⋆ ⋆ ⋆ Information is gained by measurement: ‘entropy’ decreases.

P
S[ρ̂] ≥ λ Pλ S[ρ̂λ ], Pλ is the probability of outcome λ, ρ̂λ is the collapsed state.
– Proof: see G. Lindblad, Commun. Math. Phys. 28, 245 (1972).

Summary of Lecture #2: identical particles
• Get a clear picture of the Fock space:

direct sum of identical particle Hilbert spaces for all possible particle numbers.
F = H0 ⊕ H1 ⊕ H2 ⊕ . . . .
• Get a clear picture of the many-body Hilbert space of fermions and bosons:
(Anti-)Symmetrized tensor product space.
– n-body Hilbert space for identical particles Hn is a subspace of (H1 )⊗n .
– Hn is the image of the multi-linear (anti-)symmetrization mapping:

S: (H1 )⊗n 7→ Hn .
– The (anti-)symmetrization mapping is defined on a tensor product basis as

P
S : |ψ1 i . . . |ψn i 7→ √1n! σ∈Sn |ψσ(1) i . . . |ψσ(n) i ≡ |ψ1 , . . . , ψn i, for boson;
P
S : |ψ1 i . . . |ψn i 7→ √1n! σ∈Sn (−1)σ |ψσ(1) i . . . |ψσ(n) i ≡ |ψ1 , . . . , ψn i, for fermion.
– NOTE: the above picture is rather inconvenient, and will not be used in practice.
• Be familiarized with the second quantization language:

creation & annihilation operators, and their (anti-)commutation relations.
Be able to use it to understand/formulate many-body Hamiltonians.
• Be familiarized with several simple(‘free particle’) many-body wavefunctions:

e.g. boson coherent states, fermion product states, BCS states.
– They are the “vacuum” of certain single particle “annihilation” operators.
References:
P.A.M. Dirac, The Principle of Quantum Mechanics, Chapter IX.
L.D. Landau, E.M. Lifschitz, Quantum Mechanics: Non-relativistic Theory, Chapter IX.
A. Altland, B.D. Simons, Condensed Matter Field Theory, Chapter 2.

I THE FOCK SPACE
I. THE FOCK SPACE
A. Trivia about the Permutation Group Sn

 
1 2 ··· n
• A permutation σ =   means replacing 1 by σ(1), 2 by σ(2), · · · ,
σ(1) σ(2) · · · σ(n)
n by σ(n), where σ(1), σ(2), · · · , σ(n) is a rearrangement of 1, 2, · · · , n.
• Product of permutations σ and µ (in one convention): (σ · µ)(i) = σ(µ(i)).
• Transposition (i, j): exchange i and j while keeping the others fixed.
• Any permutation can be represented as a product of transpositions.

In fact only transpositions of neighbors (i, i + 1) are needed.
 
1 2 3 4
– Example:   = (2, 3)(3, 4)(1, 2) = (1, 3)(2, 3)(3, 4) = . . .
3 1 4 2
• The parity of a permutation: parity of the number of transpositions.

Even(Odd) permutation: product of even(odd) number of transpositions.
• The sign (signature) of a permutation (−1)σ [ also denoted by sgn(σ) ]:

+1 for even permutations; −1 for odd permutations.
Q

 +1, x > 0;
σ
– An explicit formula: (−1) = i<j sgn[σ(j) − σ(i)]. Here sgn[x] =
−1, x < 0.
– (−1)σ·µ = (−1)σ · (−1)µ . Namely, (even perm.)·(even perm.)=(even perm.), . . . .
• The permutation group has only two 1-dimensional “representations”:

trivial representation: R(σ) = 1; and “alternating representation”: R(σ) = (−1)σ .
B. Identical Particle: Traditional Treatment using Wavefunctions
• The configuration of n identical(indistinguishable) particles

x = (x1 , x2 , . . . , xn ) should be equivalent to (all permutations of x1 , x2 , . . . , xn ).
• n-body state ψ(x) = ψ(x1 , x2 , . . . , xn ) should be ‘invariant’ under permutations of xi .
– The wavefunction may get complex phase, the density matrix should be the same.

I THE FOCK SPACE
• Assume: the n-body wavefunction is a one-dimensional representation of the permu-

tation group Sn , then there are only two possibilities: bosons and fermions.
– Being a 1D representation means, for a permutation σ ∈ Sn ,

ψ(xσ(1) , xσ(2) , . . . , xσ(n) ) = R(σ) · ψ(x1 , x2 , . . . , xn ), where R(σ) is a complex
number of unit modulus, and R(σ · µ) = R(σ) · R(µ).
– Note that pairwise exchange(transposition) σi,j (xi ↔ xj ) is its own inverse,

(σi,j )2 = ✶, then [R(σi,j )]2 = 1, namely R(σi,j ) = ±1.
∗ σi′ ,j ′ = σi,i′ σj,j ′ σi,j σi,i′ σj,j ′ , therefore R(σi′ ,j ′ ) = R(σi,j ).
– Bosons: a pairwise exchange (any permutation) has no effect on the wavefunction.

Trivial representation of permutation group.
– Fermions: a pairwise exchange changes the sign of the wavefunction.

Odd permutations(odd # of pair exchanges) changes the sign of the wavefunction.
Alternating representation of permutation group.
– In two-dimensional space, braiding group instead of permutation group should

be considered. There are particles(anyons) beyond bosons and fermions.
See e.g. C. Nayak et al., Rev. Mod. Phys. 80, 1083 (2008).
C. The Structure of Many-body Hilbert Space
• Fock space: the Hilbert space of identical particles with indefinite particle number, is
the direct sum of 0-particle & 1-particle & . . . Hilbert spaces. F = H0 ⊕ H1 ⊕ . . . .
– generic states in Fock space: ‘linear superpositions’ of 0-particle state(‘vacuum’),

and 1-particle state ψN =1 (r 1 ), and 2-particle state ψN =2 (r 1 , r 2 ), and . . . .
• 0-particle Hilbert space H0 : Hilbert space containing only the “vacuum” state.
– The “vacuum” state is usually denoted by |vaci or |0i.
– About the “vacuum”: roughly speaking, it means that no particle (that we

are considering) is in the coordinate space of 1-particle wavefunctions in H1 .
Example: for this particle confined in the left well,
the state is |ψileft ⊗ |vaciright

I THE FOCK SPACE
• 1-particle Hilbert space H1 : linear space of 1-body wavefunctions,

can be of finite or infinite dimension. Denote the 1-body states by e.g. |ψi.
• n(≥ 2)-particle Hilbert space Hn : a subspace of the tensor product (H1 )⊗n
= H1 ⊗ H1 ⊗ · · · ⊗ H1 , with (anti-)symmetrization between the factor H1 s.
(Anti-)Symmetrization is for identical particles so that they are indistinguishable.
• Consider the (anti-)symmetrization operation, which maps H1⊗n to Hn ,

S: H1⊗n 7→ Hn , |ψ1 i ⊗ |ψ2 i ⊗ · · · ⊗ |ψn i 7→ |ψ1 , ψ2 , · · · , ψn i.
– S is multi-linear, namely linear with respect to each factor |ψi i.
– Permutations of ψi produce the same n-body state, up to an overall phase.
– If ψi are orthonormal, |ψ1 , . . . , ψn i is normalized (for both bosons and fermions).
• Bosons: permutations of ψi are trivial.

P
– S : |ψ1 i ⊗ · · · ⊗ |ψn i 7→ |ψ1 , . . . , ψn i ≡ √1 |ψσ(1) i ⊗ · · · ⊗ |ψσ(n) i.
n! σ∈Sn
– |ψσ(1) , · · · , ψσ(n) i = |ψ1 , · · · , ψn i, ∀σ ∈ Sn .
• Fermions: permutations of ψi produces the sign of permutation.
– Exclusion principle: if ψi = ψj (i 6= j), fermion state |ψ1 , · · · , ψn i = 0.

(by W. Pauli ).
– “Fermion exchange sign”: | · · · ψi · · · ψj · · · i = −| · · · ψj · · · ψi · · · i.
∗ “Proof”: 0 = | · · · (ψi + ψj ) · · · (ψi + ψj ) · · · i

= | · · · ψi · · · ψi · · · i + | · · · ψj · · · ψj · · · i + | · · · ψi · · · ψj · · · i + | · · · ψj · · · ψi · · · i
= 0 + 0 + | · · · ψi · · · ψj · · · i + | · · · ψj · · · ψi · · · i.
P
– S : |ψ1 i ⊗ · · · ⊗ |ψn i 7→ |ψ1 , . . . , ψn i ≡ √1n! σ∈Sn (−1)σ |ψσ(1) i ⊗ · · · ⊗ |ψσ(n) i
– |ψσ(1) , · · · , ψσ(n) i = (−1)σ |ψ1 , · · · ψn i, ∀σ ∈ Sn .
D. Basis of Many-body Hilbert Space
• The wavefunction is the expansion coefficient in the coordinate eigenstate basis.

This time the coordinate basis are the tensor product states |x1 i ⊗ · · · ⊗ |xn i.
The (un-normalized) wavefunction is ψ(x1 , · · · , xn ) = (|x1 i ⊗ · · · ⊗ |xn i, |ψ1 , · · · , ψn i).

I THE FOCK SPACE
P
• Bosons: ψ(x1 , · · · , xn ) = √1n! σ∈Sn hx1 |ψσ(1) i · · · hxn |ψσ(n) i
P
= √1n! σ∈Sn ψσ(1) (x1 ) · · · ψσ(n) (xn ) = √1n! perm[ψj (xi )].
P Q
– Permanent of a square matrix Aij : perm[A] ≡ σ∈Sn i Ai,σ(i) .
P Q
– hφ1 , . . . , φn |ψ1 , . . . , ψn i = perm[hφi |ψj i] = σ∈Sn i hφi |ψσ(i) i. Exercise.
P
• Fermions: ψ(x1 , · · · , xn ) = √1n! σ∈Sn (−1)σ hx1 |ψσ(1) i · · · hxn |ψσ(n) i
P
= √1n! σ∈Sn (−1)σ ψσ(1) (x1 ) · · · ψσ(n) (xn ) = √1n! det[ψj (xi )].
This is the Slater determinant.
P Q
– Determinant of a square matrix Aij : det[A] ≡ σ∈Sn (−1)σ i Ai,σ(i) .
P Q
– hφ1 , . . . , φn |ψ1 , . . . , ψn i = det[hφi |ψj i] = σ∈Sn (−1)σ i hφi |ψσ(i) i. Exercise.
• Suppose H1 has complete orthonormal basis |ei i. For simplicity, assume a m(finite)-
dimensional H1 (i = 1, . . . , m). The goal is to construct a basis for Hn .
• Bosons: basis are |ei1 , ei2 , · · · , ein i, for all 1 ≤ i1 ≤i2 ≤· · · ≤in ≤ m.
– These basis are orthogonal, but not normalized.

m+n−1
(m+n−1)!
– The number of basis (dimension of Hn ) is n
= n!(m−1)!
.
• Fermions: basis are |ei1 , ei2 , · · · , ein i, for all 1 ≤ i1 <i2 <· · · <in ≤ m.
– Obviously, if n > m, there is no legitimate n-body state (exclusion principle).
– These basis are orthonormal.

m
m!
– The number of basis (dimension of Hn ) is n
= n!(m−n)!
.
• Occupation number representation:

denote the above basis |ei1 , ei2 , · · · , ein i as |en1 1 , en2 2 , · · · , enmm i,
where nj is the number of appearance of ej . Note that n1 + n2 + · · · nm = n.
The occupation basis |n1 , n2 , · · · , nm i is the state |en1 1 , en2 2 , · · · , enmm i normalized:
|n1 , n2 , · · · , nm i = (n1 !)−1/2 (n2 !)−1/2 · · · (nm !)−1/2 |en1 1 , en2 2 , · · · , enmm i.
– For fermions, nj = 0 or 1, the normalization factor is trivial.

II SECOND QUANTIZATION
II. SECOND QUANTIZATION
A. Creation and Annihilation Operators
• The goal is to define linear operators which creates(destroys) a particle of the 1-body
state ψ, in the Fock space.
– The creation operator ψ̂ † maps Hn to Hn+1 .

The annihilation operator ψ̂ maps Hn to Hn−1 (vanishes on “vacuum” H0 ).
• Creation operator: (“add a 1-body state as a new factor”)

ψ̂ † : Hn → Hn+1 , |ψ1 , · · · , ψn i 7→ |ψ, ψ1 , · · · , ψn i.
– In occupation basis (exercise: check these by definition):

√
for bosons, ê†i | · · · , ni , · · · i = ni + 1| · · · , (ni + 1), · · · i;
Pi−1
for fermions, ê†i | · · · , ni = 0, · · · i = (−1) j=1 nj
| · · · , ni = 1, · · · i.
• Annihilation operator: (“try to remove a 1-body state from each factor respectively”)
P
ψ̂ : Hn → Hn−1 , |ψ1 , · · · , ψn i 7→ ni=1 (±1)i−1 hψ|ψi i |ψ1 , · · · , ψi−1 , ψi+1 , · · · , ψn i,
where +1 is for bosons and −1 is for fermions.
– In occupation basis (exercise: check these by definition):

√
for bosons, êi | · · · , ni , · · · i = ni | · · · , (ni − 1), · · · i;
Pi−1
nj
for fermions, êi | · · · , ni = 1, · · · i = (−1) j=1 | · · · , ni = 0, · · · i.
• Occupation basis |n1 , · · · , nm i = (n1 !)−1/2 · · · (nm !)−1/2 (ê†1 )n1 · · · (ê†m )nm |0i.
• (Anti-)Commutation relations: Bosons: [ψ̂, ψ̂ † ] = 1. Fermions: {ψ̂, ψ̂ † } = 1.

And mode occupation number operator: n̂ψ = ψ̂ † ψ̂, for normalized 1-particle state ψ,
– [n̂ψ , ψ̂ † ] = ψ̂ † , namely ψ̂ † increases eigenvalue of n̂ψ by 1.

[n̂ψ , ψ̂] = −ψ̂, namely ψ̂ decreases eigenvalue of n̂ψ by 1.
Exercise: check these statements for both bosons and fermions.
– Eigenvalues of n̂ψ are non-negative integers: n̂ψ is hermition, positive semi-

definite (because ψ̂ † is indeed the hermitian conjugate of ψ̂, see later)
– Occupation basis |n1 , · · · , nm i are eigenstates of n̂ei with eigenvalue ni .

• (Anti-)Commutation relations between annihilation/creation operators of different

modes (exercise: check these by definition):
Bosons: [ψ̂, ψ̂ ′† ] = hψ|ψ ′ i, [ψ̂, ψ̂ ′ ] = 0 = [ψ̂ † , ψ̂ ′† ].
Fermions: {ψ̂, ψ̂ ′† } = hψ|ψ ′ i, {ψ̂, ψ̂ ′ } = 0 = {ψ̂ † , ψ̂ ′† }.
– For the 1-body orthonormal basis ei , corresponding operators satisfy:

Bosons: [êi , ê†j ] = δij . Fermions: {êi , ê†j } = δij .
– For coordinate eigenstates |xi, denote corresponding operators by ψ(x)d and ψd † (x):
d ψd
Bosons: [ψ(x), d ψd
† (x′ )] = δ(x − x′ ). Fermions: {ψ(x), † (x′ )} = δ(x − x′ ).
NOTE: these are not related to some state ψ, symbols like φ may also be used.
d & ψd
– (Anti-)Commutation relations of momentum eigenstate operators ψ(p) † (p)
are similar.
P
• Basis change: ψ̂ † = i hei |ψi ê†i , sum is over a complete orthonormal basis.
R
In particular ψ̂ † = ψ(x)ψd † (x) dx, where ψ(x) = hx|ψi is the wavefunction.
– If ei and e′i are two sets of complete orthonormal 1-body basis, then
P
ê′ i = j he′i |ej iêj , or column vector eˆ′ = U · ê,
where Uij = he′i |ej i is a unitary matrix.
– Exercise: check the converse of the above statement.

For orthonormal boson(fermion) basis êi satisfying [êi , ê†j ] = δij ({êi , ê†j } =
P
δij ), the transformed operators ê′ i = j Uij · êj satisfy the same form of
commutation(anti-commutation) relations, [ê′i , ê′† ′ ′†

j ] = δij ({êi , êj } = δij ), namely
that ê′ i also form orthonormal basis, if U is a unitary matrix.

P
– (USEFUL) This can be used to “diagonalize” bilinear operators M̂ = i,j ê†i Mij êj
= ê† · M · ê = ê′† · U · M · U † · ê′ . By choosing the unitary matrix Uij =
he′i |ej i, the matrix U · M · U † may become diagonal with eigenvalues λi on the
P P
major diagonal. Then M̂ = i λi ê′† ′
i êi =
′ ′
i λi n̂i , and ei occupation basis are
normalized eigenstates of M̂ in the entire Fock space.
• “vacuum”: ψ̂|0i = 0 for any “annihilation” operator ψ̂,

and h0|ψ̂ † = 0 for any “creation” operator ψ̂ † .

1. Creation and Annihilation Operators: Consistency Check
• Check that ψ̂ † is indeed (ψ̂)† : show that

(ψ̂ † |ψ1 , . . . , ψn−1 i, |φ1 , . . . , φn i) = (|ψ1 , . . . , ψn−1 i, ψ̂|φ1 , . . . , φn i),
for any n-body state |φ1 , . . . , φn i and (n − 1)-body state |ψ1 , . . . , ψn−1 i.
– Bosons: (check Laplace expansion of permanent)

(ψ̂ † |ψ1 , . . . , ψn−1 i, |φ1 , . . . , φn i) = hψ, ψ1 , . . . , ψn−1 |φ1 , . . . , φn i
P
= σ∈Sn hψ|φσ(1) ihψ1 |φσ(2) i . . . hψn−1 |φσ(n) i.
P
(|ψ1 , . . . , ψn−1 i, ψ̂|φ1 , . . . , φn i) = ni=1 hψ|φi ihψ1 , . . . , ψn−1 |φ1 , . . . , φn without φi i
P P
= ni=1 hψ|φi i permutation σ′ of (1,. . . ,n, without i) hψ1 |φσ′ (1) i . . . hψn−1 |φσ′ (n−1) i,
σ ′ (1), . . . , σ ′ (n − 1) is a rearrangement of n − 1 numbers (1, . . . , n, without i),
therefore the sequence i, σ ′ (1), . . . , σ ′ (n − 1) is a permutation σ of 1, . . . , n,
the two final results are summations over the same n! terms.
– Fermions: (check Laplace expansion of determinant). Exercise.
• Note that ψ̂1† · · · ψ̂n† |0i = |ψ1 , · · · , ψn i, by definition of creation operators.
– Action of creation operator ψ̂ † : ψ̂1† · · · ψ̂n† |0i 7→ ψ̂ † ψ̂1† · · · ψ̂n† |0i,

consistent with ψ̂ † : |ψ1 , · · · , ψn i 7→ |ψ, ψ1 , · · · , ψn i.
– Action of annihilation operator ψ̂:
∗ Bosons: ψ̂1† · · · ψ̂n† |0i 7→ ψ̂ ψ̂1† · · · ψ̂n† |0i

P
= [ψ̂, ψ̂1† · · · ψ̂n† ]|0i + ψ̂1† · · · ψ̂n† ψ̂|0i = i ψ1† · · · [ψ̂, ψi† ] · · · ψn† |0i + 0
P
= i hψ|ψi iψ1† · · · ψi−1 †
ψi+1†
· · · ψn† |0i,
consistent with the definition of ψ̂ and commutation relation.
∗ Fermions: ψ̂1† · · · ψ̂n† |0i 7→ ψ̂ ψ̂1† · · · ψ̂n† |0i
P
= i (−1)i−1 ψ1† · · · {ψ̂, ψi† } · · · ψn† |0i + (−1)n ψ̂1† · · · ψ̂n† ψ̂|0i
P
= i (−1)i−1 hψ|ψi iψ1† · · · ψi−1 † †
ψi+1 · · · ψn† |0i,
consistent with the definition of ψ̂ and anti-commutation relation.
R R ∗
– Under coodinate basis: ψ̂ † = ψ(x)ψd † (x) dx, ψ̂ = d dx.
ψ (x)ψ(x)
† RR ∗ d ψd
∗ Bosons: [ψ̂, ψ̂ ′ ] = ψ (x)ψ ′ (x′ ) [ψ(x), † (x′ )] dxdx′
RR ∗ R
= ψ (x)ψ ′ (x′ ) δ(x − x′ ) dxdx′ = ψ ∗ (x)ψ ′ (x) dx = hψ|ψ ′ i.
Anti-commutation relation of fermions is similar.

2. Creation and Annihilation Operators: Some Calculation Tricks
• For orthonormal basis of creation(annihilation) operators ê†i (êi ), the commutator

[ê†i êj , ê†k ] = δjk ê†i . This is true for both bosons and fermions. Exercise: check this.

• By the above fact, ê†i êj · ê†i1 . . . ê†in |vaci = sum of ê†i′ . . . ê†i′n |vaci, where the sequence
1

(i′1 , . . . , i′n ) is (i1 , . . . , in with one appearance of j replaced by i) .
For example: ê†1 ê2 · ê†1 ê†2 ê†2 ê†3 |vaci = ê†1 ê†1 ê†2 ê†3 |vaci + ê†1 ê†2 ê†1 ê†3 |vaci.
B. The Second Quantization
• The goal: use the creation & annihilation operators to simplify the presentations of
operators for identical particles in Fock space.
• The rule of thumb: to get a many-body term (defined on the Fock space),
replace the 1-body wavefunctions ψ(x) [ ψ ∗ (x) ] in the expectation value formula for a
product states by operator ψ̂(x) [ ψd
† (x) ], remove the summations over particle indices.
Some ‘normal ordering’ may be needed.
• Generic 1-body term Ô(x):
– Ô(x) can be ‘taking derivatives with respect to x’ and ‘multiplication by a func-

tion of x’ and so on. Here x is the particle’s coordinate.
– For n identical particles with coordinates x1 , . . . , xn , the corresponding many-

P
body term in the ‘first quantized’ language is ni=1 Ô(xi ).
– In a (anti-)symmetrized tensor product state |ψ1 , . . . ψn i, the expectation value

P R
would be ni=1 ψi∗ (xi ) Ô(xi ) ψ(xi ) dxi .
R
– The corresponding second quantized form is ψd † (x) Ô(x) ψ(x)d dx.
P R 2 2
• Example: 1-body kinetic energy term: i ψi∗ (x)(− h̄2m∂x )ψi (x) dx.
R d dx = R ψd
Corresponding many-body term is ψd d dp
2 2 2
† (x)(− h̄ ∂x )ψ(x) † (p)( p )ψ(p)
2m 2m
P R
• Example: 1-body potential term: V (x) ψi∗ (x)ψi (x) dx. i
R
Corresponding many-body term is V (x) ψd d dx.
† (x)ψ(x)
Exercise: convert this into momentum eigenstate representation.

R d d
• Example: total particle number operator: N̂ = ψ † (x)ψ(x) dx.
It is difficult to write down the corresponding ‘first quantized’ form.
• Generic 2-body term Ô(x, x′ ): x and x′ are the two particles’ coordinates.
– For n identical particles with coordinates x1 , . . . , xn , the corresponding many-

P
body term in the ‘first quantized’ language is 21 i,j,i6=j Ô(xi , xj ). Here the factor
1
2
is to remove the double-counting of the same pair (xi , xj ).
– In a (anti-)symmetrized tensor product state |ψ1 , . . . ψn i, the expectation value

P RR ∗
would be 21 i,j,i6=j ψi (xi )ψj∗ (xj ) Ô(xi , xj ) ψj (xj )ψi (xi ) dxi dxj .
– The corresponding second quantized form is

1
RR d d d′ )ψ(x)
d dxdx′ .
2
ψ † (x)ψ † (x′ ) Ô(x, x′ ) ψ(x
P R
• Example: 2-body potential term: (1/2) V (x, x′ ) ψi∗ (x)ψi (x) ψj∗ (x′ )ψj (x′ ) dxdx′ .
i6=j
R
Corresponding many-body term is V̂ = (1/2) V (x, x′ ) ψd † (x)ψd d′ )ψ(x)
† (x′ )ψ(x d dxdx
– Note the “normal ordering”: put all creation operators in front of annihilation
operators, be careful about the exchange sign in case of fermions.
– Exercise: convert this into momentum eigenstate representation, in case that

V (x, x′ ) = V (x − x′ ) depends only on the distance of two particles.
– In the ‘first quantized’ language, the 2-body potential term should be

P
V̂ : ψ(x1 , · · · , xn ) 7→ (1/2) i6=j V (xi , xj )ψ(x1 , · · · , xn ).
– Check: Assume the case of bosons.

Many-body states can generically be expanded in terms of |x1 , · · · , xn i as
R
ψ(x1 , · · · , xn )|x1 , · · · , xn i dx1 · · · dxn . Apply the many-body term to coordi-
nate basis state |x1 , · · · , xn i. The result is
R P
(1/2) V (x, x′ ) ψd† (x)ψ̂ † (x′ ) ′
i6=j δ(x − xi ) δ(x − xj ) |xi &xj removedi dxdx
′
P
= (1/2) i6=j V (xi , xj ) ψ̂ † (xi )ψ̂ † (xj )|xi &xj removedi
P
= (1/2) i6=j V (xi , xj ) |x1 , · · · , xn i.
The above result shows that the action of V̂ on many-body wavefunction
ψ(x1 , · · · , xn ) is the same as the ‘first quantized’ description.
Exercise: check the case of fermions, be careful about fermion exchange signs.

III SPECIAL MANY-BODY STATES
• Example: the Bose-Hubbard model.

X UX
Ĥ = −t (b̂†j b̂i + h.c.) + n̂i (n̂i − 1),
<ij>
2 i
where n̂i = b̂†i b̂i is the occupation number operator, h.c. means the Hermitian conjugate
of the previous term. [b̂i , b̂†j ] = δij and all other commutators between them vanish.
– On each site i of a lattice, there is one single-boson mode φi (x), and all φi form a
complete orthonormal basis of 1-body Hilbert space. b̂†i & b̂i are the corresponding
creation/annihilation operators.
– The Hamiltonian consists of a kinetic energy term and an interaction term. The
kinetic energy term makes a particle to ‘hop’ from site i to one of its neighbors
j, with a matrix element −t. The interaction term creates repulsion energy U
between each pair of particles on the same site.
– Example: consider only two sites i & j, use the occupation number basis, the
action of Ĥ on |ni = 3, nj = 1i is Ĥ |ni = 3, nj = 1i =
√ √ √
−t 3 2 |ni = 2, nj = 2i − t 4 |ni = 4, nj = 0i + 3 U |ni = 3, nj = 1i.
III. SPECIAL MANY-BODY STATES
A. Fermion “Product” State (Fermi Sea)
Qn
• |ψ1 , · · · , ψn i = i=1 ψ̂i† |0i.
p
Norm of this state is given by the Gram determinant det[hψi |ψj i].
• If ψi are linearly dependent, this state vanishes.
• Linearly independent ψi span a n-dimensional 1-body Hilbert space. Given a complete

orthonormal basis of this space ci , then |ψ1 , · · · , ψn i = det[hci |ψj i] · |c1 , · · · , cn i,
– Gram-Schmidt orthogonalization: the orthonormal basis can be constructed as

|c1 i ∝ |ψ1 i, e.g. |c1 i = √ |ψ1 i ,
hψ1 |ψ1 i
|c2 i ∝ |ψ2 i − |c1 ihc1 |ψ2 i = |ψ2 i − |ψ1 i hψ 1 |ψ2 i

hψ1 |ψ1 i
,
|c3 i ∝ |ψ3 i − |c1 ihc1 |ψ3 i − |c2 ihc2 |ψ3 i, . . . .

• If ci (i = 1, · · · , m) form a complete orthonormal basis of 1-body Hilbert space,

P
the total particle number n̂ = i ĉ†i ĉi is invariant under basis change.
The state |ψ1 , · · · , ψn i is an eigenstate of n̂ with eigenvalue n.
Pn † Pm †
• The ‘parent’ Hamiltonian of this state: Ĥ = − i=1 ci ci + i=n+1 ci ci .
Namely the unique ground state of this Hamiltonian is this fermion product state.
1. Particle-hole Transformation of Fermions
• Particle-hole transformation of a single fermion mode: formally ĉi ↔ ĉ†i ,

note that this preserves the anti-commutation relations.
– This corresponds to a unitary transformation on the Fock space:

†
Û = (ĉi + ĉ†i ) · (−1)
P
j6=i ĉj ĉj .
Exercise: check the following, Û † Û = ✶, Û ĉi Û † = ĉ†i , Û ĉj Û † = ĉj for j 6= i.
– The unitary transformation on occupation number basis is

P
nj
| · · · , ni = 0, · · · i ↔ (−1) j>i | · · · , ni = 1, · · · i.
is to preserve the matrix elements of ψ̂j &ψ̂j† for j > i.
P
nj
Note: the factor (−1) j>i
– In particular, the new ‘vacuum’ is originally |0, · · · , ni = 1, · · · , 0i.
• Particle-hole transformation of all fermion modes: formally ĉi ↔ ĉ†i for all i.
Exercise: what is the corresponding unitary transformation?
what is the new ‘vacuum’ ?
• By particle-hole transformation,
 the fermion product state ĉ†1 · · · ĉ†n |0i can be viewed
 c† , 1 ≤ i ≤ n, ‘hole’ annihilation operators
′ i
as the ‘vacuum’ of ĉ i =
 c , n < i, ‘particle’ annihilation operators.
i
B. Fermion Pairing State (BCS State)
• Consider two orthonormal fermion modes c1 & c2 , the pairing state is

|λi = (1 + |λ|2 )−1/2 exp(λ ĉ†1 ĉ†2 ) |0i, where λ ∈ ❈ is a complex number.
– This state is not an eigenstate of fermion number operator ĉ†1 ĉ1 + ĉ†2 ĉ2 .

– Bogoliubov transformation: define ‘Bogoliubov quasiparticles’,

γ̂1 = uĉ1 + vĉ†2 , γ̂2 = −vĉ†1 + uĉ2 ,
where u = (1 + |λ|2 )−1/2 and v = −λ (1 + |λ|2 )−1/2 .
The pairing state is vanished by γ̂1,2 , namely γ̂1,2 |λi = 0.
Exercise: check this statement, and check that {γ̂i , γ̂j† } = δij .
– A ‘parent’ Hamiltonian is Ĥ = γ̂1† γ̂1 + γ̂2† γ̂2 . Exercise: rewrite this in terms of ĉs.
P
• (Not required) Generic fermion pairing state |{fij }i ∝ exp( 12 i,j fij ĉ†i ĉ†j )|0i,
where ĉi are some orthonormal basis, fij = −fji are complex numbers.
†
– By a orthogonal transformation ĉ†i → Oij ĉ′ j , where O is an orthogonal matrix,
the antisymmetric

f matrix can

be brought into a standard form
0 λ1 0 0 ···
 
−λ 0 0 0 · · ·
 1 
OT · f · O
 
 0
=

0 0 λ2 · · ·

,
 0 0 −λ 2 0 · · ·
 
 . . . . ..

. . . . .
. . . .
† † † † ˆ′ † ˆ′ † ˆ′ † ˆ′ †
the state becomes ∝ exp(λ1 ĉ′ 1 ĉ′ 2 + λ2 ĉ′ 3 ĉ′ 4 + . . . )|0i = eλ1 c 1 c 2 eλ2 c 3 c 4 . . . |0i.
Bogoliubov transformations can then be defined on ĉ′ 2i−1 & ĉ′ 2i .
C. Boson Coherent State
2 /2 †
• The coherent state from a single boson mode b̂ is |zi = e−|z| ezb̂ |0i,
where z ∈ ❈ is a complex number.
Exercise: check the normalization of |zi.
– This state is not an eigenstate of boson number b̂† b̂.
– (USEFUL) This state is an eigenstate of b̂, b̂|zi = z|zi.

Therefore the coherent state is vanished by b̂′ = b̂ − z.
†
Exercise: check this statement, and that [b̂′ , b̂′ ] = 1.
†
– The ‘parent’ Hamiltonian is thus Ĥ = b̂′ b̂′ .
– Expectation value of ‘normal ordered’ polynomials of b̂† and b̂ (all b̂† s appear in
front of b̂s) in state |zi can be obtained by simply replacing b̂† by z ∗ and b̂ by z.
Example: hz|(b̂† b̂)2 |zi = hz|(b̂† b̂† b̂ b̂ + b̂† b̂)|zi = z ∗ z ∗ zz + z ∗ z = |z|4 + |z|2 .

D. Boson Pairing State
• Consider two orthonormal boson modes b̂1 & b̂2 , the boson pairing state is
|λi = (1 − |λ|2 )1/2 exp(λ b̂†1 b̂†2 ) |0i, where λ ∈ ❈ is a complex number, and |λ| < 1.
Exercise: check the normalization of |λi.
– This state is not an eigenstate of boson number b̂†1 b̂1 + b̂†2 b̂2 .
– Bogoliubov transformation: define γ̂1 = ub̂1 + v b̂†2 , γ̂2 = ub̂2 + v b̂†1 ,

where u = (1 − |λ|2 )−1/2 and v = −λ (1 − |λ|2 )−1/2 .
Then |λi is vanished by γ̂1,2 , namely γ̂1,2 |λi = 0.
Exercise: check this statement, and check that [γ̂i , γ̂j† ] = δij .
– A ‘parent’ Hamiltonian is Ĥ = γ̂1† γ̂1 + γ̂2† γ̂2 . Exercise: rewrite this in terms of b̂s.
E. Summary of These Special Many-body States
• All these special states are “free particle” states, they can be defined as the ‘vacuum’
of a complete set of single-particle “annihilation” operators.
Free particle state complete set of “annihilation” operators
Fermion product state (particle-hole transformed) fermion annihilation operators
Boson coherent state boson annihilation operators shifted by constants
Boson(Fermion)pairing state Bogoliubov quasi-particles (superposition of particles and ‘holes’ )
• The Wick expansion: rough statement.

Expectation value of a product of single-particle creation/annihilation operators in
these states (except boson coherent states), can be expanded into a sum of products of
pair expectation values, over all pair combinations with appropriate sign for fermions.
• The Wick expansion:

Let |0i be the single-particle ‘vacuum’. Let Âi (i = 1, . . . , 2n) be a set of single-particle
operators, namely linear combinations of annihilation and creation operators. Then
h0|Â1 Â2 · · · Â2n |0i is the Hafnian(Pfaffian) of matrix h0|Âi Âj |0i for bosons(fermions).
– Hafnian of 2n × 2n symmetric matrix Mij is

P
Hf(M ) = n!1 σ∈S2n ,σ(1)<σ(2),σ(3)<σ(4),... Mσ(1)σ(2) Mσ(3)σ(4) · · · Mσ(2n−1)σ(2n) .

– Pfaffian of 2n × 2n anti-symmetric matrix Mij is

P
Pf(M ) = n!1 σ
σ∈S2n ,σ(1)<σ(2),σ(3)<σ(4),... (−1) Mσ(1)σ(2) Mσ(3)σ(4) · · · Mσ(2n−1)σ(2n) .
– NOTE: this is true only for such ‘free particle’ states |0i.
– NOTE: the matrix h0|Âi Âj |0i may not be symmetric or anti-symmetric. But the
above definition for Hafnian/Pfaffian still works for the Wick expansion.
– Example: hÂ1 Â2 Â3 Â4 i = hÂ1 Â2 ihÂ3 Â4 i + hÂ1 Â4 ihÂ2 Â3 i ± hÂ1 Â3 ihÂ2 Â4 i,
the ± sign is for boson or fermion cases respectively.
• The Wick expansion: sketch of a proof.

Consider boson case first. Use mathematical induction.
For the case of two operators, the Wick expansion is trivially true.
Suppose the expansion is correct for product of 2n and less operators.
Add two more operators, we just need to prove that
hÂ1 · · · Â2n · Â2n+1 Â2n+2 i = hÂ1 · · · Â2n ihÂ2n+1 Â2n+2 i
P
+ i,j,i6=j h(Â1 · · · Â2n without Âi , Âj )ihÂi Â2n+1 ihÂj Â2n+2 i.
Because of linearity, only need to consider four possible cases, with (Â2n+1 , Â2n+2 ) =
(I) both annihilation operators (ψ̂, φ̂), this is trivially 0 = 0;
(II) both creation operators (ψ̂ † , φ̂† ), try to move A2n+1 and A2n+2 to the left side by
commutation relations, hÂ1 · · · Â2n Â2n+1 Â2n+2 i
= hÂ2n+1 · Â1 · · · Â2n · Â2n+2 i + h[Â1 · · · Â2n , Â2n+1 ] · Â2n+2 i
P
= 0 + i h(Â1 · · · Â2n without Âi )[Âi , Â2n+1 ] · Â2n+2 i
(NOTE: [Âi , Â2n+1 ] is a c-number)
P
= i hÂ2n+2 · (Â1 · · · Â2n without Âi )i · [Âi , Â2n+1 ]
P
+ i h[(Â1 · · · Â2n without Âi ), Â2n+2 ]i · [Âi , Â2n+1 ]
P P
= 0 + i j,j6=i h(Â1 · · · Â2n without Âi , Âj )i · [Âi , Â2n+1 ][Âj , Â2n+2 ].
In this case, [Âi , Â2n+1 ] = hÂi Â2n+1 i, [Âj , Â2n+2 ] = hÂj Â2n+2 i, and hÂ2n+1 Â2n+2 i = 0.
So this extends the Wick expansion to 2n + 2 operator case.
(III) (ψ̂ † , φ̂), and (IV) (ψ̂, φ̂† ) are left for exercise.
Exercise: repeat the above reasoning for fermions.

Summary of Lecture #3: quantum dynamics
d
• Get a clear understanding about the Schrödinger { ✐h̄ dt |ψ, ti = ĤS (t)|ψ, ti }
and the Heisenberg { h̄ ddt ÔH (t) = ✐ [ĤH (t), ÔH (t)] }
pictures about time evolution.
• Get some basic understanding about propagators (Green’s functions):

matrix element of time-evolution operator.
• Get some basic understanding about path integrals.
• Get some taste about geometric phase: Berry’s phase.
• Get some basic understanding about gauge invariance in quantum mechanics.
• Optional references:
P.A.M. Dirac, The Principle of Quantum Mechanics, Chapter V.
R.P. Feynman, A.R. Hibbs, Quantum Mechanics and Path Integral, Chapter 2.
A. Altland, B.D. Simons, Condensed Matter Field Theory, Chapter 3.
M.E. Peskin, D.V. Schroeder, An Introduction to Quantum Field Theory, Section 9.1
• Note: for simplicity I will frequently assume the space is one-dimensional,

generalization to higher spatial dimensions should be obvious in most cases.

I TIME EVOLUTION
I. TIME EVOLUTION
A. Unitary Time Evolution
• The basic assumption of quantum dynamics:

time evolution of a closed system is unitary.
– State at t is related to state at t0 by a unitary operator Û (t, t0 ):

|ψ, ti = Û (t, t0 )|ψ, t0 i, Û † (t, t0 )Û (t, t0 ) = ✶.
– Û (t, t1 ) · Û (t1 , t0 ) = Û (t, t0 ), and Û (t0 , t0 ) = ✶.
– The time evolution is usually ‘continuous’, and reversible, Û † (t, t0 ) = [Û (t, t0 )]−1 .
• The infinitesimal time evolution: Û (t0 + dt, t0 ) = ✶ − h̄✐ Ĥ(t0 ) dt + O(dt2 ).

The Hamiltonian Ĥ(t0 ) = ✐h̄ d
dt
Û (t, t0 ) = ✐h̄ d
dt
Û (t, t0 ) · Û † (t, t0 ) .
t=t0 t=t0
– Ĥ(t) = ✐h̄ dt
d
Û (t, t0 ) · Û † (t, t0 ), independent of the choice of t0 :
✐ dt
d
Û (t, t′0 ) · Û † (t, t′0 ) = ✐ ddt Û (t, t0 )Û (t0 , t′0 ) · [Û (t, t0 )Û (t0 , t′0 )]†
= ✐ dtd
Û (t, t0 )Û (t0 , t′0 ) · Û † (t0 , t′0 )Û † (t, t0 ) = ✐ ddt Û (t, t0 ) · Û † (t, t0 ).
– Ĥ is Hermitian:
Ĥ † = −✐h̄Û (t, t0 ) dt
d
Û † (t, t0 ) = −✐h̄{ ddt [Û (t, t0 ) · Û † (t, t0 )] − d
dt
Û (t, t0 ) · Û † (t, t0 )}
= −✐h̄{ ddt (✶) − d
dt
Û (t, t0 ) · Û † (t, t0 )} = ✐h̄ ddt Û (t, t0 ) · Û † (t, t0 ) = Ĥ
B. The Schrödinger Equation and The Schrödinger Picture
• From the definition of Hamiltonian, we have the

Schrödinger equation for the time evolution operator: ✐h̄ ddt Û (t, t0 ) = Ĥ(t) · Û (t, t0 ).
• This translates into the

Schrödinger equation for the quantum state: ✐h̄ ddt |ψ, ti = Ĥ(t)|ψ, ti,
and the Schrödinger equation for the ‘bra’: −✐h̄ ddt hψ, t| = hψ, t|Ĥ(t).
• Explicit form of the time evolution operator in terms of Ĥ:
– If Ĥ is independent of t, Û (t, t0 ) = exp[−✐(t − t0 )Ĥ/h̄]. (IMPORTANT)

Conversely, if Û (t, t0 ) depends only on (t − t0 ), then Ĥ is independent of time.

I TIME EVOLUTION
Rt
– In general, Û (t, t0 ) = T (exp[− h̄✐ Ĥ(t′ )dt′ ]), where T means time-ordering.
t′ =t0

 Â(t)B̂(t′ ), t > t′ ,
′
∗ Time-ordering for bosonic Â & B̂, T [Â(t)B̂(t )] =
 B̂(t′ )Â(t), t′ > t.
– This is equivalent to the Dyson series,

Û (t, t0 ) = ✶ + −h̄✐ t1 =t0 Ĥ(t1 )dt1 + ( −h̄✐ )2 t2 =t0 t12=t0 Ĥ(t2 )Ĥ(t1 )dt2 dt1 + · · · .
Rt Rt Rt
• Stationary states: eigenstates of Ĥ, |Ĥ = E, ti = e−✐E(t−t0 ) |Ĥ = E, t0 i.

Expectation value of any operator does not change over time (stationary).
Density matrix does not change over time.
• This is the Schrödinger picture (subscript S hereafter):

time evolution is implemented on the states.
C. The Heisenberg picture
• The Heisenberg picture (subscript H hereafter):

time evolution is encoded in operators, while the states have no evolution.
• Consider the time evolution of matrix elements of an operator ÔS :

hφ, t|ÔS |ψ, ti = hφ, t0 = 0|Û † (t)ÔS Û (t)|ψ, t0 = 0i.
• Define the time-dependent operator ÔH (t) in the Heisenberg picture:

(IMPORTANT) ÔH (t) = Û † (t)ÔS Û (t).
The time-dependent matrix element is simply hφ|ÔH (t)|ψi,
where the states φ & ψ do not evolve over time in Heisenberg picture.
d
• The Heisenberg equation of motion: h̄ dt ÔH (t) = ✐ [ĤH (t), ÔH (t)].
NOTE: the Hamiltonian ĤH (t) here is also in the Heisenberg picture, ĤH (t) =
✐h̄ Û † (t) dt
d
Û (t) = Û † (t)ĤS (t)Û (t).
– Proof: use h̄ ddt Û (t) = −✐Û (t) · ✐h̄Û † (t) ddt Û (t) = −✐Û (t) · ĤH (t), and
d
h̄ dt Û † (t) = h̄ ddt Û † (t) · Û (t)Û † (t) = −h̄Û † (t) ddt Û (t) · Û † (t) = ✐ĤH (t) · Û † (t).
d d d
Then h̄ dt ÔH (t) = h̄ dt Û † (t) · ÔS · Û (t) + Û † (t) · ÔS · h̄ dt Û (t)
= ✐ĤH (t) · Û † (t) · ÔS · Û (t) − ✐Û † (t) · ÔS · Û (t) · ĤH (t) = ✐ [ĤH (t), ÔH (t)].
– If ĤS (t) is independent of time, then Ĥ commutes with Û (t), and ĤH = ĤS .

I TIME EVOLUTION
D. Some Applications
• The Schrödinger picture: quantum Liouville equation,

d
time evolution of density matrix: ✐h̄ dt ρ̂(t) = [ĤS (t), ρ̂(t)].
p̂2
• Consider time-independent Ĥ = 2m
+ V (x̂).
d
– The Schrödinger picture: continuity equation, dt
[ρ(x, t)] + ∇ · J (x, t) = 0,
where the probability density ρ(x, t) = ψ ∗ (x, t)ψ(x, t) = hx|ρ̂(t)|xi,
probability current J (x, t) = Re[ψ ∗ (x, t) −✐m
h̄∂x
ψ(x, t)].
– The Heisenberg picture: Ehrenfest theorem:

the equation of motion of position x̂ and momentum p̂ are (subscript H omitted)
d ✐
dt
x̂(t) = h̄
[Ĥ, x̂(t)] = p̂/m,
d ✐ ∂
dt
p̂(t) = h̄
[Ĥ, p̂(t)] = ✐[V (x̂), p̂] = − ∂x V (x).
2 2
∂
Combine them, m ddt2 x̂(t) = − ∂x V (x), or equivalently, m ddt2 hx̂i = −h ∂x
∂
V (x)i,
which looks like the classical equation of motion.
p̂ 2 2
• 1D harmonic oscillator: time-independent Ĥ = 2m + mω 2
x̂2 = h̄ω(â† â + 1/2),
q q
✐p̂
p mω † h̄ † mωh̄
where â = 2h̄
(x̂ + mω
), [â, â ] = 1. x̂ = 2mω
(â + â ), p̂ = − ✐ 2
(â − â† ).
Normalized ground state |0i ( â|0i = 0 ), and excited states |ni ≡ √1 (â† )n |0i, energy
n!
eigenvalues En = h̄ω · (n + 21 ).
– Equation of motion: d
dt
â(t) = h̄✐ [Ĥ, â(t)] = −✐ω â(t). Then
â(t) = e−✐ωt â(0), â† (t) = e✐ωt â† (0).
– Solution to the equation of motion of x̂ and p̂: ellipse on hxi-hpi plane,

1
x̂(t) = cos(ωt) x̂(0) + mω
sin(ωt) p̂(0),
p̂(t) = −mω sin(ωt) x̂(0) + cos(ωt) p̂(0).
2 /2 †
– Example: t = 0 state is coherent state, |ψ(0)i = e−|z| ezâ |0i, with hâ(0)i = z.
q
2h̄
√
Then hx̂(0)i = mω Re(z), hp̂(0)i = 2h̄mω Im(z).
The expectation value of x̂ at time t is hx̂(t)i = cos(ωt)hx̂(0)i + sin(ωt)
mω
hp̂(0)i
q
✐
2h̄
= mω Re(ze−✐ωt ). You do not need to solve |ψ(t)i = e− h̄ Ĥ·t |ψ(0)i.
– The coherent state satisfies the minimal uncertainty relation for x̂ and p̂:
h̄ h̄mω h̄2
hx̂2 i − (hx̂i)2 = 2mω
, hp̂2 i − (hp̂i)2 = 2
, so h(∆x̂)2 i h(∆p̂)2 i = 4
= 41 |h[x̂, p̂]i|2 .
Exercise: check the proof of uncertainty relation to see why.

I TIME EVOLUTION
1 2 1
• Landau level: time-independent Ĥ = 2m
P̂ = 2m
[p̂ − qA(r)]2 ,
charge-q particle in xy-plane under uniform magnetic field B = Bez along z (B > 0),
r = (x, y), p̂ = −✐( ∂∂x , ∂∂y ), A is vector potential with ∇×A = B or ∂ ∂
A − ∂y
∂x y
Ax = B.
– Note: [r̂a , r̂b ] = 0, [r̂a , P̂b ] = ✐h̄δab ,

P
[P̂a , P̂b ] = [p̂a − qAa , p̂b − qAb ] = ✐h̄q (∂a Ab − ∂b Aa ) = ✐h̄q c ǫabc Bc .
– Equation of motion: d
dt
r̂ = h̄✐ [Ĥ, r̂] = 1
m
P̂ ,
q q
d
dt
P̂ = h̄✐ [Ĥ, P̂ ] = 2m
(P̂ × B − B × P̂ ), for uniform B, d
dt
P̂ = m
P̂ × B.
d 2d
Combine these, m dt 2 r̂(t) = dt r̂(t) × qB, Lorentz force & cyclotron motion.
q
1
– Define b̂ = 2h̄qB (P̂x + ✐P̂y ), then [b̂, b̂† ] = 1, and Ĥ = h̄ωc (b̂† b̂ + 1/2).
ωc = qB/m is the cyclotron frequency, energy levels are En = h̄ωc · (n + 1/2) for
non-negative integer n. Exercise: check these statements.
ez P̂y P̂x
– Guiding center coordinates: R̂ = (X̂, Ŷ ) = r̂ − qB
× P̂ = (x̂ + qB
, ŷ − qB
).
R̂ is conserved: h̄ ddt R̂(t) = ✐[Ĥ, R̂] = 0.
✐
NOTE: [X̂, Ŷ ] = − qB 6= 0, indicates degeneracy of Landau level.
Exercise: e✐X̂ commutes with Ĥ, but changes eigenvalue of Ŷ by −1/qB.
• The adiabatic theorem: roughly speaking, if a system starts at (one of) the instanta-
neous ground state(s), and the Hamiltonian changes slowly with time, then the system
will remain to be (one of) the instantaneous ground state(s) at later times.
– Sketch of a proof (for non-degenerate case):

Denote the instantaneous eigenstates of ĤS (t) by |ψn (t)i, and corresponding
eigenvalues by En (t), where n = 0, 1, 2, . . . , and E0 (t) < E1 (t) < E2 (t) < . . . .
Suppose |ψ, ti satisfies ✐h̄ ∂∂t |ψ, ti = ĤS (t)|ψ, ti, and |ψ, t = 0i = |ψ0 (t = 0)i.
Rt
Expand |ψ, ti, |ψ, ti = n cn (t)e−✐θn (t) |ψn (t)i, where θn (t) ≡ h̄1 0 En (t)dt. The
P
Schrödinger equation becomes differential equations for coefficients cn (t),

∂ ∂ ✐θn (t)−✐θm (t)
P
∂t
c n (t) = − m hψn (t)|( ∂t |ψm (t)i)·e ·cm (t), and c0 (t = 0) = 1, cn6=0 (t =
0) = 0. Take t-derivative on hψn (t)|ψm (t)i = δmn and hψn (t)|ĤS (t)|ψm (t)i =
δmn En (t), we have hψn (t)|( ∂∂t |ψm (t)i) = Em (t)−E 1
n (t)
· hψn (t)| ∂ Ĥ∂tS (t) |ψm (t)i for
h i
1 ∂ ĤS (t)
m 6= n. So if the magnitude of E0 (t)−E n (t)
· hψ n (t)| ∂t
|ψ 0 (t)i · (time duration)
is small, the final cn6=0 (t) will be small, the system will approximately remain to
be the instantaneous ground state (up to an overall phase).

II PROPAGATOR AND PATH INTEGRAL
II. PROPAGATOR AND PATH INTEGRAL
A. Brief Review of Gaussian Integrals
R∞ 2 /2a √
• One dimensional case: −∞
e−x dx = 2π a.
R 2
x2 e−x /2a dx
– hx2 i = R 2
e−x /2a dx
= a.
R 2n −x2 /2a
x e dx
– hx2n i = R 2
e−x /2a dx
= an · (2n − 1)!!, satisfies the ‘Wick expansion’ [(2n − 1)!! =
(2n − 1)(2n − 3) · · · (1) ways of pairing up xs].
R∞ 2 √ 2
– −∞ e−x /2a+yx dx = 2π a · eay /2 .
−1
·x
√
exp(− x·A2 ) dm x = (2π)m/2 det A, where
R
• Higher dimensional Gaussian integral:
x = (xi ) is m-component real vector, A = (Aij ) is m × m real symmetric positive-
definite matrix, the integral is over all components of x from −∞ to +∞.
R
xi xj exp(−x·A−1 ·x/2) dm x
– hxi xj i = R = Aij .
exp(−x·A−1 ·x/2) dm x
– ‘Wick expansion’: all possible ways of pairing up xs,

example, hxi xj xk xℓ i = Aij Akℓ + Aik Ajℓ + Aiℓ Ajk .
√
– exp(−x · A−1 · x/2 + y · x) dm x = (2π)m/2 det A · exp(y · A · y/2).
R
∗ z/a
e−z d2 z = πa,
R
• Complex Gaussian integral:
where d2 z = dRez dImz, and the integral is over Rez and Imz from −∞ to +∞.
R ∗
z z ∗ e−z z/a d2 z
– hz z ∗ i = R ∗
e−z z/a d2 z
= a.
– hz n (z ∗ )m i = 0 if n 6= m. Consider z → e✐θ z.
– hz n (z ∗ )n i = an · n! (‘Wick expansion’: n! ways of pairing up z ∗ and z).

∗ ∗ ∗ ∗
– e−z z/a+y z+z y d2 z = πa eay y .
R
exp(−z ∗ ·A−1 ·z) d2m z = π m det(A),

R
• Higher dimensional complex Gaussian integral:
where z = (zi ) is m-component complex vector, A = (Aij ) is a m × m Hermitian
positive-definite matrix.
– hzi zj∗ i = Aij .
– Non-vanishing ‘correlators’ must contain the same number of z ∗ and z.

– ‘Wick expansion’: all possible ways of pairing up z ∗ and z,

example, hzi zj zk∗ zℓ∗ i = Aik Ajℓ + Aiℓ Ajk .
– exp(−z ∗ · A−1 · z + y ∗ · z + z ∗ · y) d2m z = π m det(A) exp(y ∗ · A · y).

R
B. Propagator
• The propagator is the time-evolution operator represented in coordinate basis:

K(x′ , t; x, t0 ) = hx′ |Û (t, t0 )|xi, then K(x′ , t; x, t) = δ(x′ − x).
– By definition, ψ(x′ , t) = K(x′ , t; x, t0 )ψ(x, t0 ) dx.

R
– It is the transition probability amplitude for the particle (the system)

to start at x at time t0 and end up at x′ at time t.
• Customarily, when t < t0 , K(x′ , t; x, t0 ) ≡ 0, then K is the Green’s function satisfying

[H ′ − ✐h̄ ddt ]K(x′ , t; x, t0 ) = −✐h̄δ(x′ − x)δ(t − t0 ), where H ′ is the Hamiltonian acting
on x′ .
• For time-independent Ĥ with energy eigenstates |Ei,

K(x′ , t; x, t0 ) = E e−✐E (t−t0 )/h̄ hx′ |EihE|xi, for t > t0 .
P
p̂2
– For Ĥ = 2m
,
dp − h̄✐ p2
1 h̄✐ p (x′ −x) ′ 2
q
K(x′ , t; x, t0 ) = (t−t0 ) m
exp[ ✐m (x −x)
R
h̄
e 2m
2π
e = 2πh̄(t−t0 )✐ 2h̄(t−t0 )
].
Exercise: draw qualitatively the shape of real/imaginary part of K.
R
• Trace of time-evolution operator G(t, t0 ) = Tr[Û (t, t0 )] = K(x, t; x, t0 ) dx.
P
For time-independent Ĥ with eigenvalues Ei , G(t) = i exp[−✐E t/h̄], for t > 0,
similar to finite temperature partition function (with β replaced by ✐t/h̄).
1. Brief Notes on Causal Functions
• The propagator and G(t) are “causal functions”: nonzero only for later (t > 0) times.
• You especially experimentalists will frequently encounter/measure such functions (re-

sponse functions): perturb the system and measure the response at later times.

G(t) e✐ωt dt.

R
• Consider the Fourier transform G̃(ω) = −✐
Because G(t < 0) = 0, G̃(ω ′ + ✐ω ′′ ) is non-singular(analytic) for ω ′′ > 0 for all real ω ′ ,
and tends to zero fast enough at infinity with ω ′′ > 0.
• Kramer-Kronig relation (Hilbert transform): for such functions G̃(ω),

R +∞ ′) R +∞ ′)
ReG̃(ω) = π1 P −∞ ImωG̃(ω
′ −ω dω ′ , ImG̃(ω) = − π1 P −∞ ReωG̃(ω
′ −ω dω ′ .
R +∞ ′)
Or equivalently, ✐π G̃(ω) = P −∞ G̃(ω ω ′ −ω
dω ′ .
Here P means Cauchy principal value.
G̃(ω ′ )
For a proof, consider the integral of ω ′ −ω
over ω ′ contour on the right.
• Measurements usually observe the imaginary part (dissipation, absorption, etc.) of

the response functions. Kramer-Kronig can be used to get the real part.
1
• Example: G(t > 0) = e−✐Et , then G̃(ω) = ω−E
− ✐ π δ(ω − E).
The imaginary part (poles) can be used to identify the energy spectrum.
C. Reminder about Classical Mechanics
• The Lagrangian for dynamical system is L(q, q̇) = T − V (kinetic − potential energy),
d
which is a function of generalized coordinate q and generalized velocity q̇ = dt
q.
• Dynamics follows the principle of least action: classical trajectory ‘minimizes’ the
R
action, S = L(q, q̇) dt, among all trajectories with the same boundary condition.
d ∂L ∂L
• Euler-Lagrange equation: dt ∂ q̇
− ∂q
= 0, from δS = 0.
• The Hamiltonian is H(p, q) = pq̇ −L(q, q̇), the Legendre transformation of Lagrangian,
∂L
where the generalized momentum p = ∂ q̇
, and q̇ should be solved in terms of p and q.
• The Hamilton’s equation: ṗ = − ∂H

∂q
and q̇ = + ∂H
∂p
.
∂A ∂B ∂A ∂B
• Poisson bracket: {A, B} = ∂q ∂p
− ∂p ∂q
.
Corresponds to quantum commutator − h̄✐ [Â, B̂] of corresponding observables.
d
• Equation of motion: dt
A(p, q) = −{H, A}.
Corresponds to the Heisenberg equation of motion, d
dt
Â(t) = h̄✐ [Ĥ, Â].

R tf
• Hamilton’s principal function: S(qf , tf ; qi , ti ) = ti
L(q, q̇) dt, with q(ti ) = qi & q(tf ) =
∂S ∂S
qf , integrated over a classical trajectory. ∂qf
= p(t = tf ), ∂tf
= −H(t = tf ).
∂S ∂S
• Hamilton-Jacobi equation: ∂tf
+ H( ∂q f
, qf ) = 0.
D. Path Integral in Quantum Mechanics
• The goal: try to describe the quantum dynamics from a ‘classical’ point of view,
as particle moving in coordinate space (or coordinate-momentum phase space).
Then quantum interference between paths must be considered.
• Path integral version #1:

Rt
K(x′ , x, t) = D[x(τ )] exp[ h̄✐ 0 L(x(τ ), ẋ(τ )) dτ ] = D[x] exp[ h̄✐ S].
R R
D[x(τ )]: functional integral over all path x(τ ) with x(0) = x and x(t) = x′ .
R
–
The measure of paths is very difficult to define.
– L(x, ẋ): the Lagrangian.

d
ẋ = dτ
x:
the ‘velocity’ on the path x(τ ).
Rt
S: the action of path x(τ ), S[x(τ )] ≡ 0 L(x(τ ), ẋ(τ )) dτ .
• Path integral version #2:

Rt
K(x′ , x, t) = D[x(τ )]D[p(τ )] exp[ h̄✐ 0 [pẋ − H(p, x)] dτ .
R
R
– D[p(τ )]: integral over all path in momentum space, with proper measure.
Path of p has no boundary condition.
– H(p, x): classical Hamiltonian.
p̂2 ✐
• An example: time-independent Ĥ = 2m
+V (x). Propagator K(x′ , x, t) = hx′ |e− h̄ tĤ |xi.
– Divide this propagation over time t into N steps, each of time ǫ = t/N ,
✐
K(x′ , x, t) = hx′ |(e− h̄ ǫĤ )N |xi, insert N − 1 resolution of identities. K(x′ , x, t) =
✐ ✐ ✐
dxN −1 · · · dx2 dx1 hx′ |e− h̄ ǫĤ |xN −1 i · · · hx2 |e− h̄ ǫĤ |x1 ihx1 |e− h̄ ǫĤ |xi.
R R R
✐ ✐ p̂2 ✐
– Approximation (Trotter-Suzuki): e− h̄ ǫĤ = e− h̄ ǫ 2m e− h̄ ǫV (x) + O(ǫ2 ).
✐ ✐ p̂
2 ✐ p m ✐ ǫ [ m ( xi+1 −xi )2 −V (x )]
hxi+1 |e− h̄ ǫĤ |xi i ≈ hxi+1 |e− h̄ ǫ 2m |xi i e− h̄ ǫV (xi ) = 2πh̄ǫ ✐
e h̄ 2 ǫ i
.

N PN −1
– K(x′ , x, t) ≈ m
) 2 exp( h̄✐ ǫ [ m2 ( xi+1ǫ−xi )2 − V (xi )]),
R
dxN −1 · · · dx1 ( 2πh̄ǫ ✐ i=0
where x0 = x & xN = x′ . The integral is over the discretized path

x0 (τ = 0) = x, x1 (τ = ǫ), · · · , xN (τ = N ǫ = t) = x′ .
PN −1
– Take N → ∞ (ǫ → 0) limit, sum in the exponent ( i=0 ǫ) becomes integral
Rt t 2
dτ , then K(x′ , x, t) = D[x(τ )] exp( h̄✐ 0 [ m2ẋ − V (x)]dτ )
R R
0
Rt
= D[x(τ )] exp( h̄✐ 0 L(x, ẋ) dτ ).
R
m N
∗ NOTE: there is an ugly normalization factor ( 2πh̄ǫ ✐
) 2 hidden in D[x(τ )].
p̂2
• The example again: Ĥ = 2m
+ V (x) = T (p̂) + V (x).
✐ ✐
K(x′ , x, t) = hx′ |e− h̄ tĤ |xi = hx′ |(e− h̄ ǫĤ )N |xi.
✐ ✐ ✐
– Trotter-Suzuki: e− h̄ ǫĤ = e− h̄ ǫT (p̂) e− h̄ ǫV (x) + O(ǫ2 ).
– Insert N − 1 resolution of identity in terms of x eigenstates,

and N resolution of identity in terms of p̂ eigenstates,
K(x′ , x, t) = h̄−N dpN −1 · · · dp0 dxN −1 · · · dx1
R
✐ ✐ ✐ ✐
× hxN |e− h̄ ǫT (p̂) |pN −1 ihpN −1 |e− h̄ ǫV (x) |xN −1 i · · · hx1 |e− h̄ ǫT (p̂) |p0 ihp0 |e− h̄ ǫV (x) |x0 i
= h̄−N dpN −1 · · · dp0 dxN −1 · · · dx1 exp{− h̄✐ N
R P −1
i=0 ǫ [T (pi ) + V (xi )]}
× hxN |pN −1 ihpN −1 |xN −1 i · · · hx1 |p0 ihp0 |x0 i

h̄−N dpN −1 · · · dp0 dxN −1 · · · dx1 (2π)−N
R
=
× exp{ h̄✐ N
P −1 xi+1 −xi
i=0 ǫ [pi ( ǫ
) − T (pi ) − V (xi )]}
Rt
– In the N → ∞ limit, K(x′ , x, t) = D[x]D[p] exp( h̄✐ 0 [pẋ − H(p, x)] dτ ).
R
∗ The measure of paths D[x]D[p] contains the (2πh̄)−N normalization factor.
– If the Hamiltonian contains terms like p̂x̂ or x̂p̂, special care is needed.
See e.g. Peskin&Schroeder, Section 9.1.
E. Equivalence of Path Integral to the Schrödinger Equation
p̂2
• The example: Ĥ = 2m
+ V (x).
• Consider K(x′ , x, t + ǫ), add one last step (xN +1 = x′ ) to the path integral.
✐
K(x′ , x, t + ǫ) = dxN hx′ |e− h̄ ǫĤ |xN i K(xN , x, t)
R
R p m ✐ ǫ [ m ( x′ −xN )2 −V (x )]
≈ dxN 2πh̄ǫ ✐
e h̄ 2 ǫ N
K(xN , x, t).

III GEOMETRIC PHASE
• Expand K(xN , x, t) around xN ∼ x′ , K(x′ , x, t + ǫ)

p m ✐ ǫ [ m ( x′ −xN )2 −V (x )] h (xN −x′ )2 ∂ 2
i
1 + (xN − x′ ) ∂∂x′ + + · · · K(x′ , x, t).
R
≈ dxN 2πh̄ǫ✐ e h̄ 2 ǫ N
2 ∂x′2
• Do the Gaussian integral, keep terms up to O(ǫ),

✐ ǫ ∂2
K(x′ , x, t + ǫ) ≈ [1 + h̄ 2m ∂x′2
− h̄✐ ǫV (x′ ) + O(ǫ2 )]K(x′ , x, t).
• Finally, taking limit of ǫ → 0,

2
∂
∂t
K(x′ , x, t) = − h̄✐ [− 2m
1 ∂
∂x′2
+ V (x′ )]K(x′ , x, t) = − h̄✐ Ĥ ′ K(x′ , x, t),
which is the Schrödinger equation for propagator.
F. Stationary Phase Approximation
e✐k f (x) dx with large k, most contribution comes from xs where f is ‘stationary’
R
• For
[f ′ (xs ) = 0], where the integrand has no rapid oscillation. Expand f around xs ,
f (x) ≈ f (xs ) + 21 f ′′ (xs ) (x − xs )2 , do the Gaussian integral, sum over all stationary xs ,
P q ✐
dx ≈ xs k f2π
R ✐k f (x) ✐k f (xs )
e ′′ (x ) e
s
.
e✐k f (x) dn x, where x = (xi ) is n-component vector,

R
• For n-dimensional integral,
1 ∂ 2 f (xs )
expand f around stationary point xs , f (x) ≈ f (xs ) + 2 ∂xi ∂xj
(x − xs )i (x − xs )j ,
do the Gaussian integral, sum over all stationary points xs ,
P 2π✐ n/2 −1/2
R ✐k f (x) n ∂ 2 f (xs )
e d x ≈ xs ( k ) det ∂xi ∂xj e✐k f (xs ) .
• In path integral formulation of quantum mechanics, the large number k is 1/h̄.

The stationary phase approximation is the semi-classical approximation (h̄ → 0).
✐
D[x] e h̄ S is
R
– The stationary phase condition for
δS
δx
= 0, or the classical equation of motion (Euler-Lagrange equation).
– (Not required) van Vleck formula: (see e.g. Prof. Littlejohn’s lecture notes #9)
✐ 2
( ∂ S )1/2 exp[ h̄✐ S(x′ , x, t)],
P 1
K(x′ , x, t) = D[x] e h̄ S ≈
R
√
2πh̄✐ ∂x′ ∂x
the sum is over all classical trajectories from x to x′ in time t.
III. GEOMETRIC PHASE
• Consider an adiabatic periodic evolution of a Hamiltonian Ĥ(t) with Ĥ(T ) = Ĥ(0).

Suppose the Hamiltonian always has a unique ground state |E0 (t)i of energy E0 (t).

III GEOMETRIC PHASE
After the periodic evolution, what is the phase acquired by the ground state?
RT
• The phase factor is hE0 (0)|Û (T )|E0 (0)i, where U (T ) = T (exp[− h̄✐ t′ =0
Ĥ(t′ )dt′ ]).
– Note that |E0 (t)i is not a “trajectory” of time evolution, t here is just a parameter
of these states. Û (t)|E0 (0)i is not exactly |E0 (t)i, but by the adiabatic theorem
they will only differ by a complex phase.
• Divide T into N intervals of ǫ = T /N , define tn = nǫ. Then up to O(ǫ2 )

✐ ✐
error, hE0 (0)|Û (T )|E0 (0)i ≈ hE0 (0)|e− h̄ ǫĤ(tN −1 ) |E0 (tN −1 )i · · · hE0 (t1 )|e− h̄ ǫĤ(t0 ) |E0 (0)i
✐ PN −1
≈ e− h̄ ǫ i=0 E(ti )
hE0 (0)|E0 (tN −1 )i · · · hE0 (t1 )|E0 (0)i.
✐
RT
• With ǫ → 0 (N → ∞) limit, the first factor becomes e− h̄ 0 E(t)dt
, which is the expected
dynamic phase acquired from time-evolution.
• The second factor is hE0 (0)|E0 (tN −1 )i · · · hE0 (tn )|E0 (tn−1 i · · · hE0 (t1 )|E0 (0)i.
IF |E0 (0)i = |E0 (tN = T )i, this is N
Q
n=1 hE0 (tn )|E0 (tn−1 )i
RT
≈ n=1 [1 − ǫhE0 (tn )| ( ∂∂tn |E0 (tn )i)] ≈ exp[ N
QN P
n=1 ǫ ✐ At (t n )] ≈ exp[ ✐ 0
At (τ )dτ ],
where At (t) = ✐hE0 (t)| ( ∂∂t |E0 (t)i).
R
• The Berry’s phase: At (t)dt, where t parametrizes a periodic evolution |ψ(t)i,
∂
the Berry connection (with respect to t) is At = ✐hψ(t)|( ∂t |ψ(t)i).
– Periodicity requirement: ψ(tfinal ) = ψ(tinitial ).

R
Otherwise At (t)dt is not the total Berry’s phase accumulated.
– Here t is just a parameter describing the path in the Hilbert space.

|ψ(t)i is not a “trajectory” of time-evolution, |ψ(t)i is not exactly
Û (t, tinitial )|ψ(tinitial )i.
• NOTE: The Berry’s phase does not depend on the speed of evolution,
it only depends on the closed path (geometry) in Hilbert space.
– Consider another evolution parametrized by u = f (t), then

Au = ✐hψ|( ∂∂u |ψi) = (f ′ )−1 · ✐hψ|( ∂∂t |ψi) = (f ′ )−1 At ,
the Berry’s phase Au (u)du = (f ′ )−1 At · f ′ dt = At dt.
R R R
• The Berry’s phase does not depend on h̄.

IV GAUGE INVARIANCE AND ELECTROMAGNETIC FIELD
• ‘Gauge transformation’ of the Berry connection:

add (t-dependent) complex phases to the wavefunctions |ψ(t)i → e✐θ(t) |ψ(t)i,
note that θ(T ) − θ(0) ≡ 0 mod 2π for periodicity.
– The Berry connection becomes,

∂
At → ✐ hψ(t)|e−✐θ(t) ∂t e✐θ(t) |ψ(t)i = At − dθ

dt
.
dθ
R R R R
– The Berry’s phase becomes At dt → At dt − dt
dt = At dt − [θ(T ) − θ(0)] =
R
At dt mod 2π.
• Example: spin-1/2 under a Zeeman field rotating in xy-plane.

 
−✐φ
0 e
– Ĥ(φ) = −(gµB B/2)(cos φ · σx + sin φ · σy ) = −(gµB B/2)  .
✐φ
e 0
p p
– The ground state is ( 1/2, 1/2e✐φ )T , periodic with φ from 0 to 2π.
p p p
– The Berry connection Aφ = ✐( 1/2, 1/2e−✐φ ) · (0, ✐ 1/2e✐φ )T = −(1/2).
R 2π
– The Berry’s phase is 0 Aφ dφ = π.
The state acquires an additional minus sign after this periodic evolution.
p p
– Another choice of eigenvector ( 1/2e−✐φ/2 , 1/2e✐φ/2 )T is not good for comput-
ing Berry’s phase (not explicitly periodic for φ from 0 to 2π).
IV. GAUGE INVARIANCE AND ELECTROMAGNETIC FIELD
• Consider non-relativistic particle described by normalized wavefunction ψ(r, t).
2
– The Schrödinger equation is [ −h̄
2m r
∂ 2 + V (r)]ψ = ✐h̄∂t ψ.
– The probability density is ρ(r, t) = |ψ|2 .

p̂ h̄
– The probability current density is J (r, t) = Re[ψ ∗ m ψ] = −✐ 2m (ψ ∗ ∂r ψ − ψ∂r ψ ∗ ).
d
– The continuity equation for probability is dt
[ρ(r, t)] + ∇r · J (r, t) = 0
• Adding a global phase factor ψ → e✐θ ψ with real θ independent of r and t will not
change the above results.

IV GAUGE INVARIANCE AND ELECTROMAGNETIC FIELD
• If θ depends on r,t, then ∂t ψ → e✐θ ∂t ψ + (✐∂t θ)e✐θ ψ, ∂r ψ → e✐θ ∂r ψ + (✐∂r θ)e✐θ ψ,

h̄
ρ → ρ, J → J + m r
(∂ θ)ρ. It seems that the Schrödinger equation is not preserved,
the probability current density changes, and the continuity equation is violated.
• To make the theory formally “gauge invariant” under arbitrary ψ → e✐θ ψ, we need to
absorb the ∂t θ and ∂r θ terms into the transformation of a “gauge field”.
– Define a 4-component space-time-dependent real-valued “gauge field” (a0 , a). De-

fine the canonical momentum P̂ = p̂ − h̄a.
– Demand the gauge transform to be: ψ → e✐θ ψ, a → a + ∂r θ, a0 → a0 + ∂t θ.

Then it is easy to see that P̂ ψ → e✐θ P̂ ψ.
2
P̂
– Modify the Schrödinger equation as [ 2m + V (r)]ψ = h̄(✐∂t + a0 )ψ.
This will be invariant under the above gauge transformation.
P̂
– Modify the definition of probability current density as J (r, t) = Re[ψ ∗ m ψ] =
h̄ h̄
−✐ 2m (ψ ∗ ∂r ψ − ψ∂r ψ ∗ ) − m aρ. This is invariant under the gauge transformation.
So continuity equation is preserved.
• For particle with electric charge q, the above “gauge field” is the electromagnetic 4-
potential, (a0 , a) = h̄q (−φ, A), (under SI units), where φ is the electrostatic potential
(electric field E = −∇φ − ∂t A), A is the vector potential (magnetic field B = ∇ × A).
The gauge transformation of wavefunction is related to the gauge transformation of
electromagnetic 4-potential.

Summary of Lecture #4: symmetry in quantum
mechanics
• Get a clear picture of the role of symmetry in quantum mechanics.
– Generators of continuous symmetry group are conserved observables.
– For a Hamiltonian with certain symmetry, the degenerate energy levels(states)

form a linear representation of the symmetry group.
– Symmetry group elements are represented by (anti-)unitary operators.
• Be familiarized with the analysis of symmetry of a system, and its application in

conservation laws & selection rules.
– Be familiarized with certain discrete symmetries of condensed matter systems:

discrete translations, point groups.
• NOTE: natural unit ~ = 1, and the Einstein convention of implicit summation of

repeated indices, are sometimes used.
L.D. Landau, E.M. Lifschitz, Quantum Mechanics: Non-relativistic Theory, Chapter
XII.
Reference for space group:
International Tables for Crystallography, Volume A, Springer, 2005.

I BASICS OF GROUP THEORY
I. BASICS OF GROUP THEORY
A. Defining a Group
• Group: a set G with a binary multiplication ◦ : G × G 7→ G defined, satisfying,
– Associativity: f ◦ (g ◦ h) = (f ◦ g) ◦ h. The ◦ will be omitted hereafter.
– Existence of identity: there is a (unique) 1 ∈ G so that(s.t.) 1 g = g 1 = g.
– Existence of inverse: for any g ∈ G, there is a g −1 ∈ G s.t. g −1 g = g g −1 = 1.
• Group multiplication table: all results of the binary multiplication ◦.

NOTE: need to check if this table satisfies the above defining properties. · h ·
· · · ·
g · gh ·
· · · ·
• Free presentation of group: {generators | defining relations}, roughly
it is all sequences of generators (free group) “modulo” the defining relations.
– Free group: ∼ all sequences of generators (& their inverses), group multiplication
is just concatenation of sequences. It is usually non-Abelian.
∗ Example: the free group generated by x, y contains {1, x, y, x−1 , y −1 , x2 ≡

xx, xy, xy −1 , x−2 ≡ x−1 x−1 , x−1 y, x−1 y−1, yx, y 2 , yx−1 , . . . }, and x ◦ y = xy,
y ◦ x = yx, . . . , yx ◦ x−1 y = yxx−1 y = yy ≡ y 2 , . . . .
– Example: the cyclic group ❩3 : {x | x3 = 1} means all x(n mod 3)

, or {1, x, x2 }.
– Example: 2D dihedral group Dn , {Cn , σ | Cnn = σ 2 = (Cn σ)2 = 1},

it is the symmetry of the regular n-sided polygon in 2D space,
the last relation is Cn−1 σ = σCn , by this all sequence Cnp1 σ q1 Cnp2 · · ·
can be converted to Cnp σ q , then by the first two relations,
(p mod n)
the 2n elements are Cn σ (q mod 2) :
{1, Cn , Cn2 , · · · , Cnn−1 , σ, Cn σ, · · · , Cnn−1 σ},
multiplication rule is
′ ′ ′ ′
Cnm · Cnm = Cnm+m , Cnm σ · Cnm = Cnm−m σ,
′ ′ ′ ′
Cnm · Cnm σ = Cnm+m σ, Cnm σ · Cnm σ = Cnm−m ,
where m ± m′ should be understood with implicit modulo n.

B. Concepts and Terminology
• order of group |G|: ‘number’ of group elements.
• order of an element |g|: minimal integer n s.t. g n = 1 (or ∞ if n does not exist).
• Abelian group: gh = hg for all g, h ∈ G. (non-Abelian: not Abelian)
• subgroup H ≤ G: a subset H ⊆ G which is also a group under multiplication ◦.
• left(right) coset gH(Hg): set of elements of the form gh (hg) for all h in subgroup H.
• normal subgroup H E G: a subgroup H satisfying gH = Hg for all g.
• quotient group G/H: the group of cosets for normal subgroup H of G.
• conjugacy class: f and h are conjugate if there is g s.t. gf g −1 = h.

All elements conjugate to f form the conjugacy class of f .
– Elements in one conjugacy class have the same order: (gf g −1 )n = gf n g −1 .
• direct product of groups G × H: the set of (g, h) with g ∈ G & h ∈ H, and (g, h) ◦
(g ′ , h′ ) = (gg ′ , hh′ ), usually the element (g, h) is denoted just as gh.
C. Examples of Groups
• Useful examples of abstract groups:
– group of integer, real, complex numbers, ❩, ❘, ❈: Abelian,

group multiplication is just number addition, group identity is number 0.
– cyclic group ❩n: Abelian, {1, g, g 2 , · · · , g n−1 }, with g n = 1; or the group of

integers modulo n.
– symmetric group SymX: generically non-Abelian, all permutation actions on

elements of the set X. If X = {1, · · · , n}, this is the permutation group Sn .
• G is isomorphic to a subgroup of the permutation group SymG. - Cayley’s Theorem.

Left(Right) multiplication of a group element just permutes all group elements.
Each row(column) of the multiplication table is a permutation of group elements.
P P P
g∈G F (g) = g∈G F (gh) = g∈G F (hg), for any ‘function’ F , and fixed h ∈ G.

• Groups of n×n real or complex non-singular matrices: generically non-Abelian. Group

multiplication is the matrix multiplication. Group identity is the identity matrix.
– general non-singular matrices GL(n, ❘), GL(n, ❈),
– real orthogonal matrices O(n, ❘) [usually just O(n)], OT O = ✶ for O ∈ O(n),
– unitary matrices U(n, ❈) [usually just U(n)], U † U = ✶ for U ∈ U(n),
– “special” verions: SL(n, ❘), SL(n, ❈), SO(n), SU(n). Determinants are unity.
All of these are Lie groups.
D. Lie Groups and Lie Algebra
• Lie group: the group elements form a differentiable manifold,
– group U(1) :
{e✐θ } with real θ mod 2π has the geometry of a circle.
– group U(1) × U(1) : {(e✐θ , e✐φ )} with real θ, φ mod 2π, is a torus.
 
u v
– group SU(2) : { }, u, v ∈ ❈, is 3-sphere S 3 ,
∗ ∗
−v u
(Reu)2 + (Imu)2 + (Rev)2 + (Imv)2 = 1.
• Lie algebra LG: roughly speaking, the linear space tangent to Lie group manifold G at
identity 1, spanned by ‘derivatives’ of Lie group elements, −✐ ∂(real
∂
g| .
parameter) g=1
– Example: SU(2) group {exp(✐θn · σ)} with real θ and unit vector n, the Lie alge-
bra su(2) is the linear space spanned by σ1,2,3 : suppose θn depends on parameter
t, and θn|t=0 = 0 [so exp(✐θn · σ)|t=0 = ✶2×2 ], then “−✐ ∂∂(real parameter) g|g=1 ”
d
= [ dt (θn)]t=0 · σ, which can be any real-coefficient linear combinations of σ1,2,3 .
– For connected Lie group, any element g is formally an exponential g = exp(✐tX),

where X is a basis vector in Lie algebra, t is a real number (may not be unique).
∗ X = −✐ g −1 dt
d
g.
√
∗ E.g. U(1) × U(1) element (e✐θ , e2✐θ ) = exp[✐ 5θ ( √15 , √25 )], where ( √15 , √25 ) is
a basis vector in the Abelian Lie algebra u(1) × u(1) ∼ ❘2 .

II BASICS OF GROUP REPRESENTATION THEORY
∗ There are disconnected Lie groups, e.g., orthogonal group O(n). The two
kinds of orthogonal matrices with det O = ±1 are disconnected.
– Commutator of two elements ✐[X, Y ] is also an element of Lie algebra:

∂ ✐tX ✐sY −✐tX −✐sY h i
✐[X, Y ] ≡ ∂s
∂
−✐ ∂t (e e e e ) t=0 s=0 . The . . . term is an element
t=0
of Lie algebra, by the above definition, and it depends on real parameter s.
II. BASICS OF GROUP REPRESENTATION THEORY
A. Basic Concepts of Group Representation
• Linear representation of group G on linear space V : homomorphism R from G to

GL(V ), R(g) is a linear transformation on V satisfying R(g)R(h) = R(gh).
– usually consider (n-dimensional) complex linear space [(n-dim’l) Hilbert space],

the linear representation is a matrix representation, R(g) ∈ GL(n, ❈).
– R(1) = ✶, R(g −1 ) = R(g)−1
– A · R(g) · A−1 is also a representation, for constant nonsingular A.
– there is always the trivial representation R(g) = ✶ for all g.
– if all R(g) are unitary, this is a unitary representation. NOTE: we will only deal
with unitary representations here.
• Adjoint representation of Lie groups: the representation space is the Lie algebra LG,
the representation R(g) satisfies g · x · g −1 = x ◦ R(g), for g ∈ G and x ∈ LG.
– g · x · g −1 is defined as another Lie algebra element −✐ dt

d
(g · e✐tx · g −1 ) t=0
.
– x ◦ R(g) means the vector x transformed by linear transformation R(g). If ei is

P P
a basis of Lie algebra, x = i ei xi , then x ◦ R(g) = i,j ej [R(g)]ji xi .
θ θ
– Example: e✐ 2 n·σ ∈ SU(2) is represented by SO(3) matrix [R(e✐ 2 n·σ )]ij = cos θ δij +
θ θ P θ
(1 − cos θ)ni nj − sin θ ǫijk nk , namely, e✐ 2 n·σ · σi · e−✐ 2 n·σ = j σj · [R(e✐ 2 n·σ )]ji .
• Direct sum ⊕ & tensor product ⊗ of representations:

similar to quantum operators and matrices.

• Reducible representations:
a representation that all R(g) can be simultaneously block-diagonalized,
namely a subspace of representation space V is invariant under action of G.
• Irreducible representations (irrep.): representations that are not reducible.

reducible representation can be decomposed into a direct sum of irreps.
B. Orthogonality Theorem
• Orthogonality theorem:
P
for two “inequivalent” unitary irreps. g∈G R(g)∗ij R′ (g)i′ j ′ = 0;
P |G|
for the same unitary irrep. g∈G R(g)∗ij R(g)i′ j ′ = dimR δii′ δjj ′ .
– If R′ (g) = A·R(g)·A−1 , they are “equivalent”; otherwise they are “inequivalent”.

P |G|
– If R′ (g) = A · R(g) · A−1 then g∈G R(g)∗ij R′ (g)i′ j ′ = dimR (A)ii′ (A−1 )jj ′ .
P R
– For compact Lie group, g F (g) should be replace by the integral F (g)dµ(g),
R R
where dµ(g) is the “Haar measure”, satisfying F (g)dµ(g) = F (gh)dµ(g) =
R
F (hg)dµ(g), for any function F and fixed h ∈ G. |G| should be replaced by
R
the ‘volume’ of the group, dµ(g).
p
– Example: U(1) group g = e✐θ = x + ✐y, dµ(g) ∝ dθ = δ( x2 + y 2 − 1)dxdy.
R
1D irreps. are Rn (g) = e✐nθ , then Rn (g)∗ Rm (g)dµ(g) = 2π δ .
1 n,m
P pP Q
– Example: SU(2) group g = a0 σ0 − ✐ 3i=1 ai σi , dµ(g) ∝ δ( a2i − 1) 4i=0 dai .
• Character: χR (g) = TrR(g). Here Tr is the matrix trace.
– χR (g) is invariant under similarity transformation of representation.

R′ (g) = A · R(g) · A−1 for all g with constant A, then χR′ (g) = χR (g).
– Elements of the same conjugacy class have the same character,

χ(hgh−1 ) = Tr[R(hgh−1 ] = Tr[R(h)R(g)R(h−1 )] = Tr[R(h)−1 R(h)R(g)] = χ(g).
– χR⊕R′ (g) = χR (g) + χR′ (g), χR⊗R′ (g) = χR (g) · χR′ (g).
– Example: SU(2),
fundamental representation R(g) = exp(✐ 2θ n · σ), χS= 1 (g) = 2 cos 2θ ;
2
adjoint representation (previous page), χS=1 (g) = 1 + 2 cos θ.

• Orthogonality of characters:
P
for two inequivalent unitary irreps. g∈G χR (g)∗ χR′ (g) = 0;
P
for two equivalent unitary irreps. g∈G χR (g)∗ χR (g) = |G|.
P
– Example: SU(2) group, g = a0 σ0 − ✐ 3i=1 ai σi (fundamental rep.),
pP Q R
dµ(g) ∝ δ( a2i − 1) 4i=0 dai , volume of group is dµ(g) = 2π 2 .
For the examples on previous page, χS= 1 (g) = 2a0 , χS=1 (g) = 4a20 − 1.
2
R R
then |χS= 1 (g)| d µ(g) = |χS=1 (g)| dµ(g) = 2π 2 ,
2 2
2
R
and χS= 1 (g)∗ χS=1 (g) dµ(g) = 0.
2
Exercise: check these results, use 4D polar coordinates.
– This relation generalizes Fourier series [U(1) group].
• For finite group, (number of inequivalent irreps) = (number of conjugacy classes);

and (sum of squares of irrep. dimensions) = (the order of the group);
and (dimension of irrep.) divides (the order of the group).
– Abelian group of order n: n conjugacy classes, n inequivalent 1-dim’l irreps.
– S3 group: three classes, identity {()}, transpositions {(12), (13), (23)}, cyclic per-
mutations {(123), (132)}, three irreps. |S3 | = 6 = 12 + 12 + 22 :
trivial irrep.: {1, 1, 1, 1, 1, 1}; alternating irrep.: {1, −1, −1, −1, 1, 1};
√ √ √ √
2dim’l irrep.: {σ0 , σ3 , −σ3 +2 3σ1
, −σ3 −2 3σ1
, −σ0 +2✐ 3σ2
, −σ0 −2✐ 3σ2
}
size & representative of conjugacy classes

name of irrep. #1 characters
• Character table: name of irrep. #2 characters
. ..
. .
.
Example: D4 group: conjugacy classes {1}, {C4 , C43 }, {C42 }, {σ, C42 σ}, {C4 σ, C43 σ},
generators C4 : (x, y) → (−y, x), and σ : (x, y) → (x, −y)
D4 (C4v ) 1 2C4 C42 2σ 2C4 σ irrep. basis in space of homogeneous functions

Γ1 (A1 ) 1 1 1 1 1 1, x2 + y 2 , . . .
Γ2 (A2 ) 1 1 1 −1 −1 xy(x2 − y 2 )
Γ3 (B1 ) 1 −1 1 1 −1 x2 − y 2
Γ4 (B2 ) 1 −1 1 −1 1 xy
Γ5 (E) 2 0 −2 0 0 (x, y)

III CONSERVATION LAW AND DEGENERACY
• Projection operator:
given a possibly reducible representation R of group G, and the characters of one irrep.
χR′ , it is possible to build an irrep. R′ within the representation space of R.
– Denote the orthonormal basis of representation R by |ei i, i = 1, . . . , dim(R). The

P
action of group element g on the basis is ĝ|ei i = j |ej iR(g)ji .
P P P
– Build new basis |ẽi i = g∈G ĝ|ei i · χ∗R′ (g) = g∈G j |ej iR(g)ji χ∗R′ (g).
These are usually not linearly independent and not orthonormal.
– If R contains irrep. R′ , then |ẽi i will span a finite dimensional space (not all |ẽi i
are zero), then the group G is represented on this subspace as (several copies of)
the irrep. R′ .
III. CONSERVATION LAW AND DEGENERACY
A. Symmetry as Unitary Operator: 1-Particle Hilbert Space
• Think of a symmetry group G acting on the coordinate space,

e.g. spatial translations/rotations/reflections. Such symmetries x → gx induce
unitary transformations (normalization depends on convention) |xi → |gxi.
R R R
– |ψi = ψ(x)|xi dx → |gψi = ψ(x)|gxi dx = ψ(g −1 x)|xi J −1 dx, so the
∂(gx)j
wavefunction (gψ)(x) = J −1 ψ(g −1 x), where J = | det ∂xi
| is the Jacobian.
– Assume J = 1 hereafter, no need to worry about normalization.
– Associativity: we write |hgψi with no ambiguity, because

(h(gψ))(x) = (gψ)(h−1 x) = ψ(g −1 h−1 x) = ψ((hg)−1 x) = ((hg)ψ)(x).
– Given a complete orthonormal basis |ei i, |g ei i are also orthonormal basis.

P
g induces a unitary transformation ĝ = i |g ei ihei |, |g ei i = ĝ|ei i.
– ĝ has a unitary matrix representation, (ĝ)ij = hei |g ej i.

P c ij = P (ĥ)ik (ĝ)kj ,
Note: |g ej i = i |ei i (ĝ)ij in this convention, and (hg) k
g P h P P
from |hg ej i == k |h ek i(ĝ)kj == i k |ei i(ĥ)ik (ĝ)kj .
– Basis change induces similarity transformation of the representation:

P ′ ′ ′ ′
P ′ ′ ′ ′
hei |g ej i = i′ ,j ′ hei |ei′ ihei′ |g ej ′ ihg ej ′ |g ej i = i′ ,j ′ hei |ei′ ihei′ |g ej ′ ihej ′ |ej i for uni-
tary symmetry g.
• Some symmetries act on internal degrees of freedom, e.g. spin.
B. Symmetry as Unitary Operator: Fock space
• Implicit assumption: the vacuum is invariant under symmetry. Be careful about this!
• Symmetry operation ĝ defined on single particle Hilbert space induces symmetry op-
eration on Fock space, as the tensor product ĝ ⊗ ĝ ⊗ · · · ⊗ ĝ restricted in the the
(anit-)symmetrized many-body Hilbert space. E.g.: Homework #2, Problem 2.
• A more concise way is to consider the symmetry operations on the annihila-

tion/creation operators, which generates the Fock space.
• The rule of thumb: ψ̂ transforms as hψ|, ψ̂ † transforms as |ψi.
cj † = eî † gij .
cj = gij∗ eî , ge
– For complete orthonormal 1-particle basis ei : ge
Treating eî (eî † ) as a column(row) vector ê(ê† ), this can be written as vector-
e† = ê† · g.
e = g † · ê, gc
matrix products, gc
– The argument for this relation: |0i is invarient, |gej i = |ei i gij , the corresponding
cj † = eî † gij .
creation operators must be ge
P
• The operator of total particle number (both bosons and fermions) ê† · ê = i eî
†
eî is
invariant under symmetry transformation.
Q
• For fermions, the total product i eî † is invariant (up to a phase) under symmetry
Q Q
ci † = det(ĝ) i eî † .
transformation, i ge
C. Symmetry as Unitary Operator: Action on Operators
• Similar to the Heisenberg picture of time evolution, we can transfer the symmetry
g
operation on states to operations on operators: Ô → c
− gO.

• Convention #1: like the Heisenberg picture of time evolution, let the symmetry acts
only on the operator, with matrix element being the same as that of transforming only
c
states, hgψ|Ô|gφi = hψ|gO|φi, c = ĝ −1 Ôĝ.
then gO
• Convention #2: demand the matrix element to be invariant under symmetry opera-
c
tion, hgψ|gO|gφi c = ĝ Ôĝ −1 . This is more commonly used.
= hψ|Ô|φi, then gO
• A set of linear operators Ôi can also form a linear representation R(g) of the group,
di = P O
gO cj R(g)ji .
j
– Example: angular momentum operators L̂x,y,z form the adjoint representation of

SO(3) (spatial rotation).
D. Symmetry Generators as Conserved Observables
• Noether’s theorem: continuous symmetries correspond to conserved quantities,

R
for classical system described by the action S = L(q, q̇, t) dt,
if the action is invariant under ‘translation’ q(t) → q(t) + ǫf (t) for infinitesimal ǫ,
∂L d
then ‘momentum’ P = ∂ q̇
· f is conserved, dt
P = 0;
if the action is invariant under translation in time t → t + ǫ (L is in independent of t),
∂L d
then the Hamiltonian H = ∂ q̇
q̇ − L is conserved, dt
H = 0.
• Symmetry of a quantum system: the Hamiltonian is invariant under the action of g,

hgφ|Ĥ|gψi = hφ|Ĥ|ψi for all states φ, ψ, namely ĝ † · Ĥ · ĝ = Ĥ,
or [Ĥ, ĝ] = 0 for unitary symmetry.
• In quantum mechanics,
generators of continuous unitary symmetry corresponds to conserved observables.
– For generator X (element of Lie algebra), the corresponding observable is

−1
b = −✐ed
X ✐θX d d
· dθ e✐θX (this is independent of the value of θ where ddθ is taken).
– X̂ is Hermitian, and ddt X̂ = ✐[Ĥ, X̂] = 0.

† −1
Proof: consider ed✐θX
· Ĥ · ed
✐θX
= ed
✐θX
· Ĥ · ed
✐θX
= Ĥ, take dθd
at θ = 0, and use
−1 −1
the fact that X̂ = −✐ d ed ✐θX
|θ=0 = −✐ d ed
dθ
✐θX
· ed
✐θX
dθ
|θ=0 = ✐ed
✐θX
· d [ed
✐θX
dθ
]|θ=0 =
−1
✐ dθ [ed
d ✐θX
]|θ=0

• Symmetry quantum number carried by an operator Ô:

operator Ô carries quantum number λ of the symmetry generator X̂ if [X̂, Ô] = λÔ.
– Operators invariant under a continuous symmetry should commute with the sym-
metry generators (carry vanishing quantum number).
– Example: Ĥ = ~ω (b̂† b̂ + 1/2), and [Ĥ, b̂] = −~ω b̂, so the ‘energy’ (Ĥ quantum
number) of b̂ is (−~ω).
E. Symmetry and Level Degeneracy
• A symmetry g satisfies [Ĥ, ĝ] = 0, therefore g does not change energy eigenvalue.
Degenerate energy eigenstates form a representation space of the symmetry group.
Representation of g is block diagonalized in energy eigenbasis.
• Nondegenerate energy eigenstates are one-dimensional representations

(the state may only change phase under action of symmetry).
• Existence of non-commuting symmetry generators,

X̂ and Ŷ with [Ĥ, X̂] = [Ĥ, Ŷ ] = 0 and [X̂, Ŷ ] 6= 0,
usually implies degeneracy of energy levels.
– If [X̂, Ŷ ] = ✐z is a non-zero c-number, there must be degeneracy. Unitary operator

e✐X̂ changes eigenvalue of Ŷ by z. See Landau level example of Lecture #3.
– If [X̂, Ŷ ] is not a c-number, there may be a non-degenerate energy level,

IF the state is vanished by commutators of all order, 0 = [X̂, Ŷ ]|Ei =
[X̂, [X̂, Ŷ ]]|Ei = [Ŷ , [X̂, Ŷ ]]|Ei = . . . .
Example: ground state of electron in hydrogen atom (ignore spin) with
angular momentum L = 0, take X̂ = L̂x & Ŷ = L̂y .
• Usually label the degenerate levels by the representation of symmetry they belong to.
E.g. with spatial translation, label the states by momentum |pi;
with spatial rotation, label the states by angular momentum |L, Lz i, etc.

F. Examples: Translation
• Continuous translation in 1D open space: Ta : x → x + a, for all a ∈ ❘.

They form an Abelian group, Ta Ta′ = Ta+a′ , which is isomorphic to ❘.
R
– The related unitary operator is T̂a = |x + aihx| dx.
– Use momentum basis, this is simply T̂a = exp(− a p̂/~).✐

R R (−✐p (x+a)+✐p′ x)/~
∗ T̂a = |pihp|x + aihx|p′ ihp′ |dpdp′ dx = e 2π~
|pihp′ | dpdp′ dx
R R
✐
= δ(p − p′ )e−✐p a/~ |pihp′ | dpdp′ = e−✐p a/~ |pihp| dp = exp(− a p̂/~), where
R
p̂ = p|pihp| dp is the momentum operator.
∗ Exercise: use p̂ = − ✐ ∂x∂ , check that exp(−a ∂∂x )ψ(x) = ψ(Ta−1x) = ψ(x − a).
– The 1D Lie algebra is spanned by the single generator, the momentum p̂,
which is conserved in translation-invariant systems, e.g. Ĥ = p̂2 /2m.
– The momenum eigenstates |pi are 1D irreps. of this translation group,

with the representation matrix (and character) e−✐ap/~ .
G. Example: Discrete Translation
• Discrete translation (lattice translation): define T : x → x + a for a constant a 6= 0,

the discrete translation group is the cyclic group generated by T , {T n } for all n ∈ . ❩
✐
– The unitary operator is still T̂ = exp(− a p̂/~).
– There is no associated Lie algebra and conserved observables.
– Unitary irreps are 1D, T̂ = e✐θ with real θ mod 2π.
– Momentum eigenstates |p + 2π~

a
ni for n ∈ ❩ belong to the same 1D irrep. e− ap/~
✐
– Bloch’s theorem: for system with the above translation symmetry, [Ĥ, T̂ ] = 0,
The m-th energy eigenstate of irrep T̂ = e−✐ap/~ is a superposition of |p + nG~i,
|Em , pi =
P
n umn |p + nG~i, with n ∈ ❩
and G = 2π a
(the reciprocal lattice
P x
vector). The wavefunction is thus hx|Em , pi = n umn e2π✐n a e✐xp/~ = um (x) e✐xp/~ ,
P
the “Bloch function” um (x) = n umn e2π✐n (x/a) is periodic, um (x + a) = um (x).

– The Bloch function um (x) (or Fourier coefficients umn ), and

p
the “crystal momentum” p in first Brillouin zone (−π/a ≤ ~
< π/a in this case),
define the eigenstates(Bloch waves).
H. Examples: Point Group Oh
• In many materials there are the following oxygen octahedron structure:

one cation surrounded by six oxygens (or other anions),
the symmetry of the central atom is reduced
from continuous rotation to discrete rotation+reflections.
The point group (transformations leaving the central point invariant) is Oh .
The energy levels of electron orbitals on central cation
are classified by irreducible representations of Oh .
• Character table of Oh :
E 8C3 6C2′ 6C4 3C2 I 6S4 8S6 3σh 6σd basis of rep.
A1g 1 1 1 1 1 1 1 1 1 1 1, r2 s-orbital
A2g 1 1 −1 −1 1 1 −1 1 1 −1
√
Eg 2 −1 0 0 2 2 0 −1 2 0 (3z 2 − r2 , 3(x2 − y 2 )) some d-orbitals
T1g 3 0 −1 1 −1 3 1 0 −1 −1
T2g 3 0 1 −1 −1 3 −1 0 −1 1 (yz, zx, xy) some d-orbitals
A1u 1 1 1 1 1 −1 −1 −1 −1 −1
A2u 1 1 −1 −1 1 −1 1 −1 −1 1 xyz one f-orbital
Eu 2 −1 0 0 2 −2 0 1 −2 0
T1u 3 0 −1 1 −1 −3 −1 0 1 1 (x, y, z), (x, y, z)r2 p-orbitals,
(x3 , y 3 , z 3 ) some f-orbitals
T2u 3 0 1 −1 −1 −3 1 0 1 −1 (x(y 2 − z 2 ), y(z 2 − x2 ), z(x2 − y 2 )) some f-orbitals
Some group elements of Oh : action results of the (x, y, z) point.

E 8C3 6C ′ 6C 3C I 6S 8S 3σ 6σ
     2  4  2    4  6  h  d
x y y −y −x −x y −y x −y
                  
                  
y  z  x x −y  −y  −x −z  y  −x
                  
z x −z z z −z −z −x −z z

IV SELECTION RULE
IV. SELECTION RULE
A. Symmetry Constraints on Matrix Elements
• In general we want to consider the matrix element (fk )ij = hφi |fˆk |ψj i,
where i(j) indicates that φi (ψj ) is one of the degenerate energy levels of irrep. Rφ (Rψ ),
and k means that fˆk belongs to a set of operators forming irrep. Rf ,
then these matrix elements shall form a tensor product representation Rf ⊗ Rφ∗ ⊗ Rψ .
– fbk form a representation, in the sense that ĝ fbk ĝ −1 = fc

k′ Rf (g)k′ k .
– Rφ∗ is the conjugate representation of Rφ , Rφ∗ (g) = [Rφ (g)]∗ .

hgφi | = hφi |ĝ † = Rφ∗ (g)i′ i hφi′ |.
– (fk )ij form a tensor representation in the sense that

(fk )ij = hφi |fˆk |ψj i = hφi |ĝ −1 ĝ fˆk ĝ −1 ĝ|ψj i = Rφ∗ (g)i′ i Rf (g)k′ k Rψ (g)j ′ j hφi′ |fˆk′ |ψj ′ i =
Rf (g)k′ k Rφ∗ (g)i′ i Rψ (g)j ′ j (fk′ )i′ j ′ .
– Sum over g, by the orthogonality theorem, we have the . . .
• Selection rule: the matrix element (fk )ij = hφi |fˆk |ψj i will vanish
if the tensor representation Rf ⊗ Rφ∗ ⊗ Rψ , after decomposed into direct sum of irreps.,
does not contain trivial representation.
– Special case: if Rf , Rφ , Rψ are all 1-dim’l irrep., then Rf ⊗ Rφ∗ ⊗ Rψ is also a

1-dim’l irrep., then hφ|fˆ|ψi will vanish if Rf ⊗ Rφ∗ ⊗ Rψ is not the trivial irrep.
– Special case: if Rf is the trivial irrep. (there is only one fˆ, and it is invariant
under the group actions), and if Rφ and Rψ are both irrep., then
(i) hφi |fˆ|ψj i will vanish if Rφ and Rψ are “inequivalent” irrep.;
P
(ii) if Rφ = Rψ (exactly same matrices), then hφi |fˆ|ψj i = δij · dimR
1
φ
ˆ
i′ hφi′ |f |ψi′ i.
B. Examples: Selection Rule
• Parity selection rule (or usually “optical selection rule”):

consider group {1, I} generated by spatial inversion I with I 2 = 1,
it has only two irreps.: trivial (even) {1, 1} and odd representation {1, −1}.

IV SELECTION RULE
– States & operators are classified into “parity odd” (usually subscript u ) and “par-
ity even” (usually subscript g ) classes.
– Atomic orbitals of even(odd) angular momentum are parity even(odd).
– The matrix element is nonzero only for hψg |Ôg |ψg′ i, hψg |Ôu |ψu i, hψu |Ôg |ψu′ i,
ˆ g i = +|ψg i, I|ψ
hψu |Ôu |ψg i. Here I|ψ ˆ u i = −|ψu i, IÔ̂g Iˆ† = +Ôg , IÔ̂u Iˆ† = −Ôu ,
. . . . (Check the special cases under the “selection rule”)
– The optical transition (absorption/emission of one photon) probability amplitude

is proportional to hψfinal |E · r̂|ψinitial i (under lowest order of perturbation theory),
where E is the external electric field, the electric dipole operator r̂ is parity odd
(I : r → −r), so initial and final states should have opposite parity.
– Pseudo-vector: vectors that are even under inversion,

e.g. the angular momentum operator L = r × p, the magnetic moment M .
– Pseudo-scalar: scalars that are odd under inversion,

e.g. inner product of a pseudo-vector and a vector is a pseudo-scalar.
• Raman selection rule: absorb a photon of polarization E in and emit a photon of

polarization E out , the relevant matrix element is hfinal|(E out · r)(E in · r)|initiali.
– If the system have C4v symmetry (2D D4 symmetry, character table on page 7),
the initial state is of trivial representation A1 (e.g. s-orbital), E in,out are along x, y
directions respectively, then the final state must have the symmetry of function
xy, or B2 representation (e.g. dxy orbital).
C. Examples: Symmetry-allowed Hamiltonian
• In many cases you know the symmetry of your system, but don’t know the exact
Hamiltonian. The goal is to write down a Hamiltonian consistent with the symmetry.
– The general rule: find out representation ĝ of all symmetry generators, and de-
mand that ĝ −1 Ĥ ĝ = Ĥ.
– For continuous symmetry, Ĥ should commute with all symmetry generators, or

carry vanishing symmetry quantum number.

IV SELECTION RULE
• Example: translation symmetry, momentum conservation.

If the system has continuous translation symmetry, then for each monomial of opera-
tors in Ĥ, the sum of “momentum quantum numbers” of the factors must be zero,
† †
d † ψ(p),
e.g. ψ(p) \ ψ(p),
d ψ(−p) d ψ(p[ [ [ [
1 ) ψ(p2 ) ψ(p3 )ψ(p4 )δ(p1 + p2 − p3 − p4 ).
d † is the creation operator for momentum eigenstate |pi, and carries “mo-
Here ψ(p)
mentum quantum number” p (will change total momentum of the system by p); the
d carries “momentum quantum number” −p.
annihilation operator ψ(p)
• Example: particle number conservation U (1) symmetry.

ê†i are creation operators for orthonormal modes (bosons or fermions). The total
P d = e−✐θN̂ , for real θ
particle number operator, N̂ ≡ i ê†i êi , generates a U (1) group, g(θ)
mod 2π. From [N̂ , ê†i ] = +ê†i , the creation operators carry “particle number quantum
number” +1 (the annihilation operators carry “particle number quantum number”
d†
d Ĥ g(θ)
−1). Then if Ĥ has “particle number conservation symmetry”, namely g(θ)
for any θ, or equivalently [Ĥ, N̂ ] = 0, each term in Ĥ must have the same number of
creation operators and annihilation operators, e.g. ê†i êj , ê†i ê†j êk êℓ .
R †
• Example: point group symmetry on free particle Hamiltonian Ĥ = ψ̂ k ·H(k)· ψ̂ k dk,
where ψ̂ k is a column vector of annihilation operators, H(k) is a matrix.
g g R †
Under g: ψ̂ k → − R(g) · ψ̂ gk , Ĥ →
− ψ̂ gk · R(g)−1 · H(k) · R(g) · ψ̂ gk dk, then
H(gk) = R(g)−1 · H(k) · R(g).
– C4v symmetry generated by C4 : (x, y) → (−y, x), and σv : (x, y) → (x, −y),
†
consider the case

ψ̂ =(creation operators

for
 x
p , py orbitals),
0 1 1 0
then R(C4 ) =   = ✐σy , R(σv ) =   = σz ,
−1 0 0 −1
so σy H(kx , ky )σy = H(−ky , kx ), σz H(kx , ky )σz = H(kx , −ky ),
P
let 2 × 2 matrix H(k) = 3i=0 hi (k)σi , the real functions hi (k) must satisfy
h0 (kx , ky ) = h0 (−ky , kx ) = h0 (kx , −ky ) ∼ A1 irrep.,
h1 (kx , ky ) = −h1 (−ky , kx ) = −h1 (kx , −ky ) ∼ B2 irrep.,
h2 (kx , ky ) = h2 (−ky , kx ) = −h2 (kx , −ky ) ∼ A2 irrep.,
h3 (kx , ky ) = −h3 (−ky , kx ) = h3 (kx , −ky ) ∼ B1 irrep.,
this constrains the form of H matrix up to O(k 2 ) around k = (0, 0) as
H(kx , ky ) = (c0 + c1 k2 )σ0 + c2 · kx ky · σx + c3 · (kx2 − ky2 ) · σz , with real constant cs.

IV SELECTION RULE
D. Examples: Symmetry-assisted Diagonalization of Hamiltonian
• Given a Hamiltonian, first analyze its symmetry, then divide the Hilbert space into
smaller subspaces of different irreps of symmetry group.
• Example: Bose-Hubbard model on a 4-site square,

P
Ĥ = −t (b̂†1 b̂2 + b̂†2 b̂3 + b̂†3 b̂4 + b̂†4 b̂1 + h.c.) + U2 4i=1 n̂i (n̂i − 1),
solve the eigenstates of two bosons (10-dimensional Hilbert space).
– The apparent symmetry is C4v (2D D4 , character table on p.7). Generated by

C4 : b̂1 7→ b̂2 , b̂2 7→ b̂3 , b̂3 7→ b̂4 , b̂4 7→ b̂1 ; and
σ: b̂1 7→ b̂1 , b̂2 7→ b̂4 , b̂3 7→ b̂3 , b̂4 7→ b̂2 .
– The basis of creation operators can be chosen as (use projection operator):

c1 † : 1 P b̂† , B
A c1 † : 1 (b̂†1 − b̂†2 + b̂†3 − b̂†4 ), E
d x,y
†
: √12 (b̂†1 − b̂†3 ) & √12 (b̂†2 − b̂†4 )
2 i i 2
They correspond to single boson eigenstates of Ĥ.  

0 −1
c1 † →
Action of C4 : A c1 † ·(1), B
7 A c1 † 7→ B
c1 † ·(−1), (E
cx † , E
cy † ) 7→ (E
cx † , E
cy † )· .
1 0
 
1 0
c1 † →
Action of σ: A 7 A c1 † →
c1 † · (1), B c1 † · (1), (E
7 B cy † ) 7→ (E
cx † , E cy † ) · 
cx † , E .
0 −1
– The procedures of applying the “projection operator” are summarized in the
following tables,
g 1 C4 C43 C42 σ C42 σ C4 σ C43 σ
c
gb†1 b̂†1 b̂†2 b̂†4 b̂†3 b̂†1 b̂†3 b̂†2 b̂†4
c
gb†2 b̂†2 b̂†3 b̂†1 b̂†4 b̂†4 b̂†2 b̂†1 b̂†3
c
gb†3 b̂†3 b̂†4 b̂†2 b̂†1 b̂†3 b̂†1 b̂†4 b̂†2
c
gb†4 b̂†4 b̂†1 b̂†3 b̂†2 b̂†2 b̂†4 b̂†3 b̂†1
χA1 (g) 1 1 1 1 1 1 1 1
χA2 (g) 1 1 1 1 −1 −1 −1 −1
χB1 (g) 1 −1 −1 1 1 1 −1 −1
χB2 (g) 1 −1 −1 1 −1 −1 1 1
χE (g) 2 0 0 −2 0 0 0 0

IV SELECTION RULE
P c
g χ∗R (g)gb†i R = A1 R = A2 R = B1 R = B2 R = E
i=1 2(b̂†1 + b̂†2 + b̂†3 + b̂†4 ) 0 2(b̂†1 − b̂†2 + b̂†3 − b̂†4 ) 0 2(b̂†1 − b̂†3 )
i=2 2(b̂†2 + b̂†3 + b̂†4 + b̂†1 ) 0 2(b̂†2 − b̂†3 + b̂†4 − b̂†1 ) 0 2(b̂†2 − b̂†4 )
i=3 2(b̂†3 + b̂†4 + b̂†1 + b̂†2 ) 0 2(b̂†3 − b̂†4 + b̂†1 − b̂†2 ) 0 2(b̂†3 − b̂†1 )
i=4 2(b̂†4 + b̂†1 + b̂†2 + b̂†3 ) 0 2(b̂†4 − b̂†1 + b̂†2 − b̂†3 ) 0 2(b̂†4 − b̂†2 )
– Two boson states: symmetrized tensor product (use symbol ⊙) representations,
these states are classified into irreps as
(2)
A1 : A1 ⊙ A1 : √12 (Ac1 † )2 |0i; B1 ⊙ B1 : √1 (B cx † )2 + (E
c1 † )2 |0i; E ⊙ E: 1 [(E cy † )2 ]|0i,
2 2
† † † †
(2) c c 1 c 2
B1 : A1 ⊙ B1 : A1 B1 |0i; E ⊙ E: [(Ex ) − (Ey ) ]|0i, c 2
2
† † † †
(2)
B2 : E ⊙ E: 1 c c c c
(Ex Ey + Ey Ex )|0i = E cx † E
cy † |0i,
2
E (2) : A1 ⊙ E: (A cx † |0i, A
c1 † E cy † |0i); B1 ⊙ E: (B
c1 † E c1 † E cy † |0i).
c1 † E
cx † |0i, −B
The 10 × 10 problem is reduced(block-diagonalized) to

A1 irreps: 3 × 3, B1 irreps: 2 × 2, B2 irreps: 1 × 1, and E irreps: 2 × 2 (!).
– Note that E ⊗E = A1 ⊕A2 ⊕B1 ⊕B2 , however the A2 irrep here is anti-symmetric
with respect to the two factors and does not appear in boson system.
– Why E representation sub-problem is 2 × 2 instead of 4 × 4?

Note that (A1 Ex , A1 Ey ) and (B1 Ex , −B1 Ey ) are exactly the same 2-dim’l repre-
sentation,
 with  exactlythe same
 representation matrices,
0 −1 1 0
C4 →  , σ →  .
1 0 0 −1
By symmetry (check the special cases under the “selection rule”)
hA1 Ex |Ĥ|B1 Ex i = hA1 Ey |Ĥ| − B1 Ey i = hB1 Ex |Ĥ|A1 Ex i∗ = h−B1 Ey |Ĥ|A1 Ey i∗ ,
hA1 Ex |Ĥ|A1 Ex i = hA1 Ey |Ĥ|A1 Ey i, hB1 Ex |Ĥ|B1 Ex i = h−B1 Ey |Ĥ| − B1 Ey i, and
other matrix
 elements are zero. The 4 × 4 matrix  is two copies of the same 2 × 2
hA1 Ex |Ĥ|A1 Ex i, hA1 Ex |Ĥ|B1 Ex i
matrix,  
hB1 Ex |Ĥ|A1 Ex i, hB1 Ex |Ĥ|B1 Ex i
 
hA1 Ey |Ĥ|A1 Ey i, hA1 Ey |Ĥ| − B1 Ey i
= 
h−B1 Ey |Ĥ|A1 Ey i, h−B1 Ey |Ĥ| − B1 Ey i
c1 † A c1 † B
c1 + 2t B c1 + U
P
– Hamiltonian is Ĥ = −2t A 2 i n̂i (n̂i − 1).
(2)
– The matrix elements between A1 states are
(2) (2) P 1 † 2
Ĥ|A1 : A1 ⊙ A1 i = −4t|A1 : A1 ⊙ A1 i + U i 4 2 (b̂i ) |0i
√

IV SELECTION RULE
√
(2) (2) (2) 2 (2)
= −4t|A1 : A1 ⊙A1 i+U ( 41 |A1 : A1 ⊙A1 i+ 41 |A1 : B1 ⊙B1 i+ 4
|A1 : E ⊙Ei);
(2) (2) P
Ĥ|A1 : B1 ⊙ B1 i = +4t|A1 : B1 ⊙ B1 i + U i 4√1 2 (b̂†i )2 |0i
√
(2) (2) (2) 2 (2)
= +4t|A1 : B1 ⊙B1 i+U ( 41 |A1 : A1 ⊙A1 i+ 41 |A1 : B1 ⊙B1 i+ 4
|A1 : E ⊙Ei);
(2) P
Ĥ|A1 : E ⊙ Ei = U i 14 (b̂†i )2 |0i
√ √
2 (2) 2 (2) (2)
= U( 4
|A1 : A1 ⊙ A1 i + 4
|A1 i + 12 |A1 : E ⊙ Ei).
: B1 ⊙ B1 
√
U U 2U
−4t + 4 4
√
4 
(2)  U U 2U .
So the Hamiltonian in the A1 subspace is  4t + 
 √4 √
4 4 
2U 2U U
4 4 2
(2)
– The matrix elements between B1 states are
(2) P
Ĥ|B1 : A1 ⊙ B1 i = U i 14 (−1)i−1 (b̂†i )2 |0i
(2) (2)
= U ( 21 |B1 : A1 ⊙ B1 i + 21 |B1 : E ⊙ Ei);
(2) P
Ĥ|B1 : E ⊙ Ei = U i 14 (−1)i−1 (b̂†i )2 |0i
(2) (2)
= U ( 21 |B1 : A1 ⊙ B1 i + 21 |B1 : E ⊙ Ei).  
U U
(2) 2 2
So the Hamiltonian in the B1 subspace is .
U U
2 2
(2)
– The matrix elements between B2 states are
(2)
Ĥ|B2 : E ⊙ Ei = 0.
– The matrix elements between E (2) states are

Ĥ|E (2) : A1 ⊙ Ex i = −2t |E (2) : A1 ⊙ Ex i + U · √
2 2
1
[(b̂†1 )2 − (b̂†3 )2 ]|0i
= −2t |E (2) : A1 ⊙ Ex i + U ( 21 |E (2) : A1 ⊙ Ex i + 12 |E (2) : B1 ⊙ Ex i);
Ĥ|E (2) : B1 ⊙ Ex i = +2t |E (2) : B1 ⊙ Ex i + U · 2√1 2 [(b̂†1 )2 − (b̂†3 )2 ]|0i
= +2t |E (2) : B1 ⊙ Ex i + U ( 12 |E (2) : A1 ⊙ Ex i + 21 |E (2) : B 1 ⊙ Ex i). 
−2t + U2 U
So the Hamiltonian in the E (2)
subspace is two copies of  2 .
U U
2
2t + 2

Summary of Lecture #5: angular momentum and spin
• Get some understanding of the SO(3) and SU(2) group, their representations, and
their relations.
– Finite dimensional irreducible representations of SU(2) are labeled by half-integer

‘angular momentum’ j. Integer j representations are irreducible representations
of SO(3).
– Roughly speaking SU(2) is twice as large as SO(3), SO(3) ≃ SU(2)/❩2 .
• Be familiarized with orbital angular momentum and spin, and basic calculations in-
volving them.
– Be familiarized with ladder operators and calculations involving them.
– Basic tools for computation of g-factors and similar quantities.
• Get a clear understanding of time-reversal symmetry T̂ , especially its effect on spin.
– Time-reversal symmetry is anti-unitary, namely a unitary operator times the

complex conjugation operator.
– For half-odd-integer spin, T̂ 2 = −1. Such system with time-reversal symmetry

must have two-fold level degeneracy (Kramers theorem).
Landau & Lifschitz, Quantum Mechanics: Non-relativistic Theory, Chapter IV & VIII.
A. Auerbach, Interacting Electrons and Quantum Magnetism, Chapter 7.
• NOTE: Einstein convention for implicit summation over repeated indices is frequently
used.
• ⋆ indicates advanced topics (NOT required).

I SU(2) AND SO(3) GROUPS
I. SU(2) AND SO(3) GROUPS
A. Defining SO(3) Group

   
x x
  → 
← 
   
• SO(3) group: the group of proper rotations in 3-dimensional space,  y  → R •  y ,
   
z z
←
→ ←
→ ← → ←→ ←
→
with R T • R = 1 and det( R ) = 1. The notation R means it is a 3 × 3 matrix.
The symbol • means multiplication between 3-component vector and 3 × 3 matrix, or
multiplication between 3 × 3 matrices.
←→
– improper rotation (rotary reflection): those with det( R ) = −1.
Improper rotations are a proper rotation followed by inversion.
Example: S4 point group symmetry (do not confuse with

permutation group). The tetrahedron has no 4-fold axis,
but the 4-fold improper rotation S4 = I · C4−1 (clockwise 4-fold
rotation C4−1 followed by an inversion I about the center) is
its symmetry (vertices transform as 1 → 2 → 3 → 4 → 1).
– (Proper + improper rotations) form O(3) group.
• Some facts about SO(3) group:
– Any SO(3) rotation is a rotation of some angle θ around some axis n (3D unit
←
→ ←
→ ←→
vector), denoted by R n (θ) hereafter. [ R n (θ)]−1 = R n (−θ).
←→ ←→
– R n (θ) = R −n (−θ), two-fold redundancy for representing SO(3) by n and θ.
←→ ← → ←
→ ←
→
– R′ • R n (θ)• R′ −1 = R ← → (θ): R′−1 followed by rotation around n axis of angle
R′ •n
←→
θ then followed by R′ , is equivalent to a rotation around R′ • n axis of angle θ.
←→ P
– [ R n (θ)]i,j = ni nj + cos θ · (δi,j − ni nj ) − sin θ · c ǫijk nk , here i, j, k = x, y, z.
• Parametrizing SO(3) group: Euler angles.

any SO(3) rotation can be represented as
←
→ ←→ ←→
R z (α) • R y (β) • R z (γ), with three Euler angles
α ∈ [0, 2π), β ∈ [0, π), γ ∈ [0, 2π).

– Rotations around principal

 axis areexplicitly,   
cos θ − sin θ 0 cos θ 0 sin θ 1 0 0
←
→ 


 ←
→ 


 ←
→ 



R z (θ) =  sin θ

cos θ 0 ,

R y (θ) = 

0 1 0 ,

R x (θ) = 0 cos θ − sin θ .
 
0 0 1 − sin θ 0 cos θ 0 sin θ cos θ
– Representing group multiplication, Haar measure (volume element of group

space), et cetera, are not convenient in terms of Euler angles, or n and θ.
B. Defining SU(2) Group
• SU(2) group: the group of 2 × 2 special unitary matrices, U † · U = ✶, and det(U ) = 1.
– U(2) group: no unity determinant condition. Any U(2) matrix Ũ is of the form
of Ũ = e✐θ U with U ∈ SU(2), with θ = Arg(det Ũ )/2.
 
u v
• Parametrizing SU(2) group: U =  , with |u|2 + |v|2 = 1,
∗ ∗
−v u
the two complex number u, v give a faithful(1-to-1) parametrization of SU(2).
• Quaternion representation of SU(2): U = a0 σ0 − ✐(a1 σ1 + a2 σ2 + a3 σ3 ) = a0 σ0 − ✐a · σ,

P
with 4 real numbers (a0 , a1 , a2 , a3 ) = (a0 , a) with i a2i = a20 + a2 = 1.
– Relation to u, v: a0 = Re(u), a3 = −Im(u), a1 = −Im(v), a2 = −Re(v).
– Quaternion: numbers of the form a0 + a1 i + a2 j + a3 k, represented as (a0 , a),

where i2 = j2 = k2 = ijk = −1.
∗ Note the three “square roots of (-1)” i, j, k anti-commute,

ij = −ji = k, jk = −kj = i, ki = −ik = j.
∗ i, j, k can be represented by (−✐σ1,2,3 ). Exercise: check this.
– SU(2) group multiplication becomes quaternion number multiplication,

and looks like a rotation of 4-component vectors by a SO(4) matrix,
 
b0 b1 b2 b3
 
−b 
 1 b0 −b3 b2 
(a0 , a) ◦ (b0 , b) ≡ (a0 b0 − a · b, a0 b + b0 a + a × b) = (a0 , a1 , a2 , a3 ) 
−b2 b3
.
 b0 −b1 

−b3 −b2 b1 b0
– exp(−✐ 2θ n • σ) is represented by (cos 2θ , sin 2θ · n), for unit vector n and real θ.
– SU(2) manifold is 3-sphere S 3 : points ~a in 4D Euclidean space with |~a| = 1.

Group multiplications are rotations of S 3 in 4D space.

C. Defining SO(3) and SU(2) Groups
• Relation between SU(2) (quaternion) and SO(3):

θ ←→
the adjoint representation of SU(2) element e−✐ 2 n•σ is the 3×3 rotation matrix R n (θ).
θ θ ←→ θ θ P ←→
e−✐ 2 n•σ · (σ • A) · e✐ 2 n•σ = σ • R n (θ) • A. Or e−✐ 2 n•σ · σa · e✐ 2 n•σ = b σb · [ R n (θ)]ba .
– NOTE the two-to-one nature of this mapping from SU(2) to SO(3):

←
→
R n (θ) is a quadratic function of quaternion (ai ), so (ai ) and (−ai ), or SU(2)
matrices U and −U , represent the same SO(3) rotation. SO(3) ≃ SU(2)/❩2 .
– SO(3) manifold is real projective space RP3 = S 3 /❩2

(identify antipodal points, ~a and −~a with |~a| = 1, on S 3 ).
qP Q3
3 2
• Haar measure of SU(2): dµ(U ) = δ( i=0 ai − 1) i=0 dai .
This is the proper(up to constant factor) volume element of SU(2) group space.
– use 4D polar coordinates, (ai ) = r · (cos 2θ , sin 2θ (sin ϑ cos φ, sin ϑ sin φ, cos ϑ)),
this measure is δ(r − 1) · | ∂(a∂(r,θ,ϑ,φ)
0 ,a1 ,a2 ,a3 )
| · drdθdϑdφ = 21 sin2 2θ sin ϑ dθdϑdφ.
Integral over group space is over θ ∈ [0, 2π), ϑ ∈ [0, π], φ ∈ [0, 2π). This can be
written as 1
2
sin2 2θ dθ · d2 n, where d2 n means the surface area element on the unit
sphere S 2 of rotation axis n with |n| = 1 in 3D space.
D. SU(2) and SO(3) Groups and Lie Algebras
• The Lie algebras of SU(2) and SO(3) are essentially the same, su(2) = so(3).
– Both su(2) and so(3) are three dimensional, with three basis(generators) Jx,y,z .
– Jx,y,z satisfy the commutation relation [Ja , Jb ] = ✐ǫabc Jc (Einstein convention),

this is sometimes written as J × J = ✐J . Here a, b, c = x, y, z.
     
0 0 0 0 0 ✐ 0 −✐ 0
←
→ 


 ←
→ 


 ←
→ 



– For SO(3), J x = 0 0 −✐,
 
Jy=  0 0 0,
 
Jz=  ✐ 0 0.
 
0 ✐ 0 −✐ 0 0 0 0 0
←→ ←
→ ←→
Or ( J a )bc = −✐ǫabc . And R n (θ) = exp(−✐θ J • n).
σx,y,z
– For SU(2) matrix group, Jx,y,z = 2
, J are usually denoted by S here.
– If [Ja , Mb ] = ✐ǫabc Mc , by the Baker-Hausdorff formula and the above results,

P ←
→
e−✐θn•J · Mb · e✐θn•J = c Mc · [ R n (θ)]cb .

• Group elements are exp(−✐θJ • n), with real θ and real unit vector n.
←→ ←→
– For SO(3), θ has period 2π (e−✐2π J •n = 1 ).
– For SU(2), θ has period 4π. Roughly means SU(2) is twice as large as SO(3)
(note the 2-to-1 mapping between them).
– SO(3) & SU(2) matrices are fundamental representations of the SO(3) & SU(2)
group respectively. SO(3) matrices are adjoint representation of both groups.
←
→
– NOTE: exp(−✐θ′ J • n′ ) exp(−✐θJ • n) exp(✐θ′ J • n′ ) = exp(−✐θJ • R n′ (θ′ ) • n),
←→ ′ ′
where R n′ (θ′ ) is SO(3) rotation matrix (adjoint representation of e−✐θ J·n ).
– Therefore, every element of SU(2) [SO(3)] is conjugate to exp(−✐θJz ), (because

any rotation axis n can be rotated to z-direction). The conjugacy classes are
determined by θ (infinite many classes). Note that θ and −θ give the same class,
therefore the character of any representation must be even functions of θ.
E. Schwinger Boson Representations of SU(2) & SO(3)
• SU(2)
  matrices
 act  on 
a 2-dim’l complex  linear
 space as
z z1 uz1 + vz2 z′
 1 −U
→U · =  ≡  1
z2 z2 −v ∗ z1 + u∗ z2 z2′
 
z1
• View this space as the 1-boson states (z1 b̂†↑ + z2 b̂†↓ )|0i = (b̂†↑ |0i, b̂†↓ |0i) ·  ,
z2
where b̂↑,↓ are two orthonormal boson modes. The transformation of (z1 , z2 ) can be
U
→ (b̂†↑ , b̂†↓ ) · U .
also viewed as a unitary transformation on the basis, (b̂†↑ , b̂†↓ ) −
• The single-boson transformation induces transformation in n-boson space according

(z1 b̂†↑ +z2 b̂†↓ )n U (z1′ b̂†↑ +z2′ b̂†↓ )n
to √
n!
|0i −
→ √
n!
|0i, note that the norm (|z1 |2 + |z2 |2 )n = (|z1′ |2 + |z2′ |2 )n
does not change, because this is unitary transformation in n-boson space.
• Label the occupation basis |n̂↑ = n↑ , n̂↓ = n − n↑ i as |j = n2 , m = n↑ − n2 i.

The above state can be expanded in the occupation basis as
(z1 b̂†↑ +z2 b̂†↓ )n Pn n↑ n−n↑
√
√
n!
|0i = z
n↑ =0 1 2 z n!
(b̂† )n↑ (b̂†↓ )n−n↑ |0i
(n↑ )!(n−n↑ )! ↑
P2j n↑ 2j−n↑
q
n!
P j+m j−m q (2j)!
= n↑ =0 z1 z2 n↑ !(n−n↑ )!
|j, n↑ − ji = m z1 z2 (j+m)!(j−m)!
|j, mi,
P
where m means summing over m = −j, −j + 1, · · · , j − 1, j.

U P (j)
• U induces a unitary transformation on the basis |j, mi −
→ m′ |j, m′ i Dm′ m (U ).
The D matrix can be solved by considering
P j+m j−m q (2j)! U P ′ j+m ′ j−m
q
(2j)!
m z1 z2 (j+m)!(j−m)!
→ m (z1 )
|j, mi − (z2 ) (j+m)!(j−m)!
|j, mi, and
P j+m j−m q (2j)! U P ′ ′
q
(j)
m z1 z2 (j+m)!(j−m)!
→ m,m′ z1j+m z2j−m (j+m(2j)!
|j, mi − ′ )!(j−m′ )! Dmm′ (U ) |j, mi.
Plug in z1′ = uz1 + vz2 and z2′ = −v ∗ z1 + u∗ z2 ,

Pj+m k j+m−k k j+m−k j+m
expand (z1′ )j+m = k=0 u v z1 z2 k
,
P ′
and (z2′ )j−m = kj−m ∗ k′ ∗ j+m−k′ k′ j−m−k j−m
′ =0 (−v ) (u ) z1 z2 k ′ , match terms (k + k ′ = j + m′ ).
The matrix element of the (2j + 1) × (2j + 1) representation is thus

q ′ )!(j−m′ )! Pj+m j+m k j+m−k
(j)
Dmm′ (U ) = (j+m j−m ′ ′
(j+m)!(j−m)! k=0 k j+m′ −k
u v (−v ∗ )j+m −k (u∗ )j−m−(j+m −k) .
n

Note: invalid binomial coefficients m with m < 0 or m > n shall be zero.
• D(j) is a (2j + 1)-dimensional irreducible unitary representation of SU(2) group.

Its “irreducibility” will not be proved here.
(j)
• Dmm′ (U ) is a homogeneous polynomial of degree 2j in terms of u, v, u∗ , v ∗ .
For integer j, U and −U produce the same D matrix, this is then an irreducible unitary
representation of SO(3).
q
(ℓ) 4π
• Relation to spherical harmonics: D0m (e✐φJz e✐θJy ) = Y m (θ, φ),
2ℓ+1 ℓ
for integer ℓ.
R (j ′ ) (j)
• Orthogonality theorem: [Dµ′ µ (U )]∗ · Dm′ m (U ) dµ(U ) = |SU(2)| δ ′ δ ′ ′δ ,
2j+1 j j µ m µm
R
dµ(U ) is the Haar measure, |SU(2)| = dµ(U ) is the volume of SU(2) group space.
P2j sin[(j+ 21 )θ]
• Character of j-representation: χj (e−✐θn•J ) = χj (e−✐θJz ) = µ=0 e
−✐(j−µ)θ
= sin 21 θ
.
θ θ θ
– Proof: U = e−✐ 2 σz induces z1 → e−✐ 2 z1 and z2 → e✐ 2 z2 . Then
P q
′ j+m ′ j−m (2j)!
(z
m 1 ) (z 2 ) (j+m)!(j−m)!
|j, mi
P −✐ θ [(j+m)−(j−m)] j+m j−m q (2j)!
= me 2 z1 z2 (j+m)!(j−m)!
|j, mi.
(j)
Therefore Dm′ m = δm′ m e−✐mθ is a diagonal matrix, and its character is
P −✐mθ P2j ✐(j−µ)θ e✐jθ −e✐(−j−1)θ 1 1
e✐(j+ 2 )θ −e✐(−j− 2 )θ sin[(j+ 1 )θ]
m e = µ=0 e = 1−e−✐θ
= e✐θ/2 −e−✐θ/2
= sin 1 2θ .
2
R 2π 2 θ
– Exercise: check orthogonality relation 0 χj ′ (θ)∗ χj (θ) sin 2 dθ = πδj ′ j .

II ANGULAR MOMENTUM AND SPIN
II. ANGULAR MOMENTUM AND SPIN
A. Angular Momentum as SO(3) Generators
←
→
• The action of spatial rotation R n (θ) on a wavefunction of 3D coordinates ψ(r), by
←→ ←→
our convention of symmetry action, is (Rn (θ)ψ)(r) = ψ( R n (θ)−1 • r) = ψ(e✐θ J ·n • r).
Here J are the fundamental representation matrices of SO(3).
• Take d
✐ dθat θ = 0, this produces the action of symmetry generator nd · J on ψ, nd
·J :
←→ ∂ ∂ ∂
ψ 7→ −((n · J ) • r) · ∂r ψ = −na (−✐ǫabc )rc ∂r b
ψ = n · (−✐r × ∂r )ψ.
b = −✐r × ∂ = r̂ × p̂/~.
Therefore J ∂r
b
This is the orbital angular momentum (divided by ~) and usually denoted as L.
2
• L̂x,y,z are hermitian. L̂ ≡ L̂2x + L̂2y + L̂2z is hermitian and positive semi-definite.
b ×L
• Exercise: check L b = ✐L,
b [L
ba , r̂b ] = ✐ǫabc r̂c , [L b 2, L
ba , p̂b ] = ✐ǫabc p̂c , [L bx,y,z ] = 0.
• Ladder operators: L̂± = L̂x ± ✐L̂y , obviously (L̂+ )† = L̂− .
– [L̂z , L̂+ ] = +L̂+ , [L̂z , L̂− ] = −L̂− , namely L̂± changes L̂z eigenvalues by ±1.
– [L̂+ , L̂− ] = 2L̂z , and

b 2 = L̂2 + (1/2)(L̂+ L̂− + L̂− L̂+ ) = L̂z (L̂z − 1) + L̂+ L̂− = L̂z (L̂z + 1) + L̂− L̂+ .
L z
• Eigenstates of orbital angular momentum:

2 2
suppose |L̂ = α, L̂z = βi is the simultaneous eigenstate of L̂ and L̂z , then
2 2
– L̂± |L̂ = α, L̂z = βi ∝ |L̂ = α, L̂z = β ± 1i, from the commutation relations.
2
– Use the above formula of L̂ , notice that L̂+ L̂− and L̂− L̂+ are both positive
2
semi-definite. Then α = hα, β|L̂ |α, βi ≥ hα, β|L̂z (L̂z ± 1)|α, βi = β · (β ± 1).
– The sequence of |α, βi with different β (differ by integers) must be truncated

with a minimal βmin and maximal βmax , s.t. L̂+ |α, βmax i = 0 and L̂− |α, βmin i = 0.
q
Then α = βmax ·(βmax +1) = βmin ·(βmin −1). So βmax = −βmin = − 12 + α + 14 ≡ ℓ,
α = ℓ · (ℓ + 1), and 2ℓ = βmax − βmin is a non-negative integer.
2
– |L̂ = ℓ(ℓ + 1), L̂z = mi are usually denoted by |L = ℓ, Lz = mi, or just |ℓ, mi,
with m = −ℓ, −ℓ + 1, . . . , ℓ − 1, ℓ.

– In polar coodinates (r, θ, φ), L̂a does not depend on radius r.

∂ ∂ sin θ ∂
L̂z = −✐ ∂φ , L̂± = ±e±✐φ ( ∂θ ± ✐ cos θ ∂φ
).
2
– If a wavefunction ψ(r) is the simultaneous eigenstate of L̂ and L̂z , then it must be
R(r)·Yℓm (θ, φ), where R is the radial wavefunction, Yℓm is the spherical harmonics.
Here ℓ must be integer, due to the periodicity with respect to φ with period 2π.
• Matrix elements in orthonormal |L = ℓ, Lz = mi basis:

b 2 |ℓ, mi = ℓ(ℓ + 1) |ℓ, mi,
L bz |ℓ, mi = m |ℓ, mi,
L
Condon-Shortley convention:
p p
b± |ℓ, mi = ℓ(ℓ + 1) − m(m ± 1)|ℓ, m ± 1i = (ℓ ∓ m)(ℓ ± m + 1)|ℓ, m ± 1i.
L
– The magnitude of matrix element for L b± can be computed by,

b± |ℓ, mi, L
(L b± |ℓ, mi) = hℓ, m|L
b†± L
b± |ℓ, mi = hℓ, m|L
b∓ L
b± |ℓ, mi
b2 − L
= hℓ, m|L b z (L
bz ± 1)|ℓ, mi = ℓ(ℓ + 1) − m(m ± 1) = (ℓ ∓ m)(ℓ ± m + 1)
– Condon-Shortley convention fixes relative phases between |ℓ, mi with different m.
B. Spin-1/2
• “Real” spin - the intrinsic angular momentum - of a particle (e.g. spin-1/2 moment of
electron) is usually the combined effect of relativity and quantum mechanics. However
there are many pseudo-spin-1/2 systems, which are just two-state systems.
– Examples of pseudospin: electron on one of two-sublattices of graphene.
• Spin-1/2 wavefunctions & spin coherent state:

Generators of rotation in spin-1/2 Hilbert space are “spin operators” S = σ2 .
Sz eigenstates form a complete orthonormal basis |Sz = ± 12 i.
1
A generic spin-1/2 wavefunction is z1 |Sz = +
2 
i + z2 |Sz = − 12 i (with |z1 |2 + |z2 |2 = 1).
z1
This two-component “spinor wavefunction”   corresponds to a “magnetic moment”
z2
 
z1
along n = (z1∗ , z2∗ ) · σ ·   direction, this is the spin coherent state |ni.
z2
Exercise: check that n defined above is a real unit vector, check that (n· Ŝ)|ni = 21 |ni.
– For n = (sin θ cos φ, sin θ sin φ, cos θ), the spinor wavefunction is usually chosen
as (z1 [n], z2 [n])T = (e−✐φ/2 cos 2θ , e✐φ/2 sin 2θ )T .

– Note that under φ → φ + 2π this spinor wavefunction (zs) changes sign (!)
R
– Resolution of identity: |nihn| sin θdθdφ
2π
=✶
ˆ . |ni form overcomplete basis.
– ⋆ Path integral of spin system can be defined in terms of classical vector n using
these basis. However Berry phase must be included: e.g. Auerbach, Chapter 10.
θ ′
−✐ n ·σ
• Rotation in spin-1/2 space  is implemented by SU(2) matrix e 2 :
z1 [n]
θ ′
spinor e−✐ 2 n ·σ ·   corresponds to moment along ←
→
R n′ (θ) • n direction.
z2 [n]
– Rotation by 2π angle operator is −✶ (sign change of spinor wavefunction!).

This is true for all half-odd-integer spin/angular momentum system.
• ⋆ Higher spin coherent states: see e.g. Auerbach, Chapter 7.

a spin-S moment along n direction can be thought as 2S number of spin-1/2 moments
along n, |niS ∼ |ni 1 ⊗· · ·⊗|ni 1 (2S spin- 12 ). The 2S spin-1/2 has to be symmetrized.
2 2
(z1 b̂†↑ +z2 b̂†↓ )n
It is in fact the state √
n!
|0i used before in Schwinger boson representation,
if we identify boson occupation basis |j, mi as spin-S basis |S, Sz i.
P q
– |niS = Sm=−S (S+m)!(S−m)!
(2S)!
z1S+m z2S−m |S, Sz = mi,
where z1,2 are components of spin- 21 spinor (z1 , z2 )T for moment along n.
The (2S + 1) coefficients are spin-S spinor wavefunction for moment along n.
b
– (n · S)|ni b is spin operator of spin-S.
= S|niS , where S
S
R
– Resolution of identity: |niS hn|S (2S+1)4π
sin θdθdφ
= ✶.
C. Addition of Angular Momentum
• system of angular momentum j1 + system of angular momentum j2 :

the tensor product states |j1 , m1 i ⊗ |j2 , m2 i transform under SU(2) rotation as
U P (j ) (j )
→ m′ ,m′ |j1 , m′1 i ⊗ |j2 , m′2 i Dm1′ m1 (U ) Dm2′ m2 (U ).
|j1 , m1 i ⊗ |j2 , m2 i −
1 2 1 2
This (2j1 + 1)(2j2 + 1)-dimensional tensor product representation is usually reducible.
• Label the angular momentum j irrep by j. Then the tensor product representation
j1 ⊗ j2 = |j1 − j2 | ⊕ (|j1 − j2 | + 1) ⊕ · · · ⊕ (j1 + j2 ). - Clebsch-Gordon.
This is just the multiplication table of irreps for SU(2)[SO(3)].

– “Proof”: assume without loss of generality j1 ≥ j2 , use orthogonality of charac-

sin[(j+ 12 )θ]
ters, and χj = , show that χj1 ⊗j2 = χ(j1 −j2 )⊕···⊕(j1 +j2 ) ,
sin( θ2 )
P
Method #1: define x = e✐θ , then χj = 2j µ=0 x
j−µ
, represent χj1 ⊗j2 = χj1 · χj2 and
P
χ(j1 −j2 )⊕···⊕(j1 +j2 ) = J χJ both as Laurent series of x, then compare coefficients.
sin[(j + 1 )θ]·sin[(j + 1 )θ]
Method #2: χj1 ⊗j2 = 1 2 2 2
sin( θ2 )·sin( θ2 )
= 21 cos[(j1 −jsin(
2 )θ]−cos[(j1 +j2 +1)θ]
θ
)·sin( θ2 )
2
Pj1 +j2 1 cos[Jθ]−cos[(J+1)θ] Pj1 +j2 sin[(J+ 12 )θ]
= J=j1 −j2 2 sin( θ )·sin( θ ) = J=j1 −j2 sin( θ ) = χ(j1 −j2 )⊕···⊕(j1 +j2 ) .
2 2 2
• The total angular momentum operator J b =J b 1 ⊗✶+✶⊗J b 2 (just write J

b 1 +J
b 2 ) satisfies
b ×J
J b = ✐J b . By a unitary transformation of basis, J
b can be block-diagonalized, each
block is the angular momentum operator for J between |J1 − J2 | and J1 + J2 . The
unitary transformation is given by the . . .
J 1 J2 J
• Clebsch-Gordon coefficient: hJ1 , m1 ; J2 , m2 |J, mi (also denoted by Cm 1 m2 m
),
expansion coefficient of total angular momentum J and Jˆz = m basis state |J, mi on
tensor product basis |J1 , m1 ; J2 , m2 i ≡ |J1 , m1 i ⊗ |J2 , m2 i, .
P
– Definition: |J, mi = m1 ,m2 |J1 , m1 ; J2 , m2 ihJ1 , m1 ; J2 , m2 |J, mi.
– Orthogonality: as (2j1 + 1)(2j2 + 1)-dimensional unitary matrix,

P ′ ′
J,m hJ1 , m1 ; J2 , m2 |J, mihJ, m|J1 , m1 ; J2 , m2 i = δm1 m1 δm2 m2 ,
′ ′
P ′ ′
m1 ,m2 hJ , m |J1 , m1 ; J2 , m2 ihJ1 , m1 ; J2 , m2 |J, mi = δJ ′ J δm′ m .
(j)
– Decomposition of the tensor product representation, Dm′ ,m (U )
P (j ) (j )
= m′ ,m′ ,m1 ,m2 hJ, m′ |J1 , m′1 ; J2 , m′2 i·Dm1′ ,m1 (U )·Dm2′ ,m2 (U )·hJ1 , m1 ; J2 , m2 |J, mi.
1 2 1 2
1
P (J) (J ) (J )
– Then by orthogonality theorem, |SU (2)| U ∈SU (2) [Dm′ ,m (U )]∗ Dm′1,m1 (U )Dm′2,m2 (U )
1 2
1
= 2J+1
hJ1 , m′1 ; J2 , m′2 |J, m′ ihJ, m|J1 , m1 ; J2 , m2 i.
– Selection rule: hJ1 , m1 ; J2 , m2 |J, mi is nonzero only if m1 + m2 = m (Jˆz conser-

vation), and |J1 − J2 | ≤ J ≤ J1 + J2 and J1 + J2 − J is integer.
– Recursion relation: apply J∓ = J1,∓ + J2,∓ on |J, m ± 1i, use the above definition
of C-G coefficient, we have, hJ1 , m1 ; J2 , m2 |J, mi
q q
(J1 −m1 )(J1 +m1 +1) (J2 −m2 )(J2 +m2 +1)
= (J−m)(J+m+1)
hJ1 , m1 + 1; J2 , m2 |J, m + 1i + (J−m)(J+m+1)
hJ1 , m1 ; J2 , m2 + 1|J, m + 1i
q q
(J1 −m1 +1)(J1 +m1 ) (J2 −m2 +1)(J2 +m2 )
= (J−m+1)(J+m)
hJ1 , m1 − 1; J2 , m2 |J, m − 1i + (J−m+1)(J+m)
hJ1 , m1 ; J2 , m2 − 1|J, m − 1i
– Consider m = J + 1 (so |J, mi = 0) for the above relation after the last “=”.
q
We have, hJ1hJ
,m1 −1;J2 ,J−m1 +1|J,Ji
1 ,m1 ;J2 ,J−m1 |J,Ji
= − (J2 −(J−m1 )+1)(J2 +(J−m1 ))
(J1 −m1 +1)(J1 +m1 )
.

This solves all hJ1 , m1 ; J2 , m2 = J − m1 |J, m = Ji up to an overall phase factor.

m < J case can be obtained by recursion relation after the first “=”.
D. Wigner-Eckart Theorem
(k)
• If (2k + 1) operators T̂q (q = −k, −k + 1, · · · , k) transform under rotation as the
|J = k, Jz = qi angular momentum basis states, namely under SU(2) rotation U ,
(k) U (k) P (k) (k)
T̂q − → Û T̂q Û † = q′ T̂q′ Dq′ q (U ),
(k)
then the matrix element of T̂q between angular momentum j ′ & j states is
(k)
hj ′ , m′ |T̂q |j, mi = hj ′ , m′ |j, m; k, qihj ′ ||T̂ (k) ||ji,
hj ′ ||T̂ (k) ||ji is “reduced matrix element” and does not depend on m′ , q, m,
dependence on m′ , q, m are all in the Clebsch-Gordon coefficient hj ′ , m′ |j, m; k, qi.
– In some convention a factor √ 1′ is on the right-hand-side.

2j +1
(k)
– Such operators T̂q are usually called irreducible tensor operators.
Their commutator with rotation generators are [J b , T̂q(k) ] = P ′ T̂ (k) ′ b
q q ′ hk, q |J |k, qi.
p
In particular [Jbz , T̂q ] = q T̂q , [Jb± , T̂q ] = (k ∓ q)(k ± q + 1)T̂q±1 .
(k) (k) (k) (k)
P ˆ ˆ (k) (k)
– a=x,y,z [Ja , [Ja , T̂q ]] = k(k +1)T̂q . This can be used to determine the “angular
momentum quantum number” k of an operator.

(k) P (k)
– Proof: hj ′ , m′ |T̂q |j, mi = |SU1(2)| U ∈SU (2) hj ′ , m′ |Û † Û T̂q Û † Û |j, mi
P P (j ′ ) (k) (j) (k)
= |SU1(2)| U ∈SU (2) n′ ,q′ ,n [Dn′ ,m′ (U )]∗ Dq′ ,q (U )Dn,m (U )hj ′ , n′ |T̂q′ |j, ni
(use orthogonality theorem for C-G coefficients)
P (k)
= n′ ,q′ ,n 2j ′1+1 hj ′ , m′ |j, m; k, qihj, n; k, q ′ |j ′ , n′ ihj ′ , n′ |T̂q′ |j, ni.
P (k)
So hj ′ ||T̂ (k) ||ji = 2j ′1+1 n′ ,q′ ,n hj, n; k, q ′ |j ′ , n′ ihj ′ , n′ |T̂q′ |j, ni.
• Special case: T̂a transform like a vector, [Jâ , T̂b ] = ✐ǫabc T̂c , (a, b, c = x, y, z).
b b
Projection theorem: hj, m′ |T̂a |j, mi = hj,n|J·T |j,ni hj, m′ |Jâ |j, mi. (a = x, y, z)
j(j+1)
(1) (1) (1)

– Define T̂0 ≡ T̂z , T̂±1 ≡ √1 (∓T̂x − ✐T̂y ),
2
and similarly Jˆq (q = −1, 0, 1), they are
irreducible tensor operators with angular momentum quantum number k = 1.
b ·T
– J b ≡ Jˆx T̂x + Jˆy T̂y + Jˆz T̂z = P1 (Jˆq(1) )† T̂q(1) .
q=−1
b ·T
– hj, n|J b |j, ni is independent of n. So T
b is proportional to J
b in J = j space.
The proportionality constant can depend on j though.

III TIME-REVERSAL SYMMETRY
– Proof: use Wigner-Eckert theorem, and results for C-G coefficient,

′ ˆ(1)
hj,m |J |j,mi
hj, m′ |j, m; k = 1, qi = √ q (solve m′ = J case first), then
j(j+1)
(1) P ′ ′ |j,n′ i (1)
hj, m′ |T̂q |j, mi = n′ ,q′ ,n hj,m |j,m;k=1,qihj,n;k=1,q
2j+1
hj, n′ |T̂q′ |j, ni
hj,m′ |Jˆq |j,mi P
(1)
= 1 ˆ(1) † ′ ′ (1)
j(j+1) n′ ,q ′ ,n 2j+1 hj, n|(Jq ′ ) |j, n ihj, n |T̂q ′ |j, ni
P bT b |j,ni
(1) hj,n|J·
= hj, m′ |Jˆq |j, mi · 1 · n
j(j+1) 2j+1
.
E. Examples:
• Landé g-factor:
consider a particle with orbital angular momentum L and spin S,
the coupling to Zeeman magnetic field B is H b + gS S),
b = −B · µB (gL L b
e~
where µB = is a constant derived from particle charge e and mass m.
2m
Define total angular momentum J b =L b + S.

b Then [J b 2 , Ĥ] = 0. In the subspace of
total angular momentum J [namely subspace of J b 2 = J(J + 1)], this Hamiltonian is

b , where gJ is the Landé g-factor.
b J = −B · µB gJ J
H
c = (gL L
– Define M b + gS S).
b Exercise: check [Jba , M
cb ] = ✐ǫabc M b 2, M
cc , and [J c ] = 0.
bM c |J,mi
– According to the projection theorem, M c = hJ,m|J· b.
J
J(J+1)
b 2 + gS S
b 2 + (gL + gS )L b 2 + gS S b 2 + (gL + gS ) Jb −Lb −Sb
2 2 2
Use Jb ·M
c = gL L b ·S
b = gL L
2
gL −gS b 2 b 2 gL +gS b 2
= (L − S ) +
2
J , 2
(gL −gS )[L(L+1)−S(S+1)]+(gL +gS )J(J+1)
then we have gJ = 2J(J+1)
.
e~
– For electron, gL = 1, gS ≈ 2, µB = 2me
is the Bohr magneton.
III. TIME-REVERSAL SYMMETRY
• Effects of time-reversal in classical physics: changes sign of

velocity, momentum, angular momentum, electric current, magnetic moment, mag-
netic field, . . . .
• Time-reversal for non-relativistic wavefunction in quantum mechanics:

d d
Take complex conjugate of ✐ dt ψ = Ĥψ, it becomes ✐ d(−t) ψ ∗ = Ĥ ∗ ψ ∗ . In this case,
time-reversal is just complex conjugation on wavefunction, T̂ : t → −t, ψ → ψ ∗ .

– If the Hamiltonian is “real”, the system has time-reversal symmetry:

ψ ∗ (−t) is the solution of the same Schrödinger equation as ψ(t).
– With time-reversal symmetry, non-degenerate energy eigenstates ψ(t) are

(real function)×e−✐φ e−✐Et .
‘proof’: ψ(t) and ψ ∗ (−t) are of same energy. So ψ ∗ (t = 0) = e2✐φ ψ(t = 0), then
e✐φ ψ(t = 0) is real.
• Complex conjugation operator K̂: anti-linear operator (K̂λ|ψi = λ∗ K̂|ψi) defined in

coordinate basis as K̂|xi = |xi. Note that K̂† = K̂−1 = K̂. For anti-linear operator K̂,
the Hermitian conjugate is defined by (K̂† ψ, φ) = (ψ, K̂φ)∗ .
– “Real states” are states invariant under time-reversal: K̂|φi = |φi.

R R
– hx|K̂|ψi = hx|K̂ ( ψ(x′ )|x′ i dx′ ) = hx| ( ψ ∗ (x′ )|x′ i dx′ ) = ψ(x)∗ .
R e✐px R e−✐px
– K̂|pi = K̂ ( √ 2π
|xi dx) = √
2π
|xi dx = |−pi.
– For linear operator Ô, K̂Ô = Ô∗ K̂, or K̂ÔK̂−1 = Ô∗ .

Note: hφ|Ô∗ |ψi = (hφ|Ô|ψi)∗ only for real states.
– Any anti-linear operator Ô = ÔK̂ · K̂, where ÔK̂ is a linear operator.

Any anti-unitary operator is of the form Û K̂ where Û is a unitary operator.
A. Time-reversal Symmetry and Angular Momentum
• Time-reversal operation on states should change sign of measured angular momentum,

b Tb |J, mi = −hJ, m′ |J
hJ, m′ |Tb † J b |J, mi.
– Consider Jbz , hJ, m′ |Jbz |J, mi = δm′ m m, the above relation shows that Tb |J, mi =
cm |J, −mi, where cm is a phase factor.
– Consider Jbx , hJ, m′ |Jbx |J, mi

p p
= 12 (δm′ ,m+1 (J − m)(J + m + 1) + δm′ +1,m (J − m′ )(J + m′ + 1)).
hJ, m′ |Tb † Jbx Tb |J, mi = c∗m′ cm hJ, −m′ |Jbx |J, −mi
p p
= c∗m′ cm 21 (δ−m′ ,−m+1 (J + m)(J − m + 1) + δ−m′ +1,−m (J + m′ )(J − m′ + 1)).
Comparing terms, e.g. for m′ = m + 1, we must have cm+1 = −cm .
– Finally Tb |J, mi = c (−1)J−m |J, −mi, where c is a constant phase.

– For integer J, Condon-Shortley convention is [c = (−1)J ]

Tb |J, mi = (−1)m |J, −mi.
– For spin-1/2, the usual convention is c = 1, Tb | ↑i = | ↓i, Tb | ↓i = −| ↑i.
– Tb Tb |J, mi = Tb c (−1)J−m |J, −mi = c∗ c (−1)J−m (−1)J+m |J, mi = (−1)2J |J, mi.
For half-odd-integer J states, Tb 2 = −✶.
• Time-reversal operations on operators: time-reversal operator Tb = U

bT K,
b
bT is some unitary operator, K
where U b is complex conjugation.
We need (U b )∗ (U
bT )ij (J b )iℓ . However U
b † )kℓ = −(J bT would depend on basis choice (!).
jk T
bT = e✐πJby (Condon-Shortley convention). In this basis,

– Using |J, Jz i basis, U
Jby are purely imaginary in this basis, so changes sign under complex conjugation;
Jbx,z matrix elements are real, but e✐πJy Jbx,z e−✐πJy = −Jbx,z .
b b
bT = ✐σ y .
– For spin-1/2, the common convention is U
– However if “real” basis are used, e.g., for L = 1, use p-orbital basis, |px i =
|L=1,m=−1i−|L=1,m=1i
√
2
|py i = |L=1,m=−1i+|L=1,m=1i
, √
− 2✐
, |pz i = |L = 1, m = 0i, the
b are all purely imaginary. Time-reversal is just complex
matrix elements of L
conjugation on matrix elements, without further unitary transformation.
B. Time-reversal Symmetry: Kramers Degeneracy
• Kramers theorem (Kramers degeneracy):

for system with time-reversal symmetry and of half-odd-integer angular momentum,
all energy levels are (at least) doubly degenerate.
– Suppose energy eigenstate |Ei is non-degenerate, time-reversal symmetry dic-

tates that Tb |Ei is also energy E eigenstate, so Tb |Ei = e✐φ |Ei, then Tb Tb |Ei =
e✐φ e−✐φ |Ei = |Ei, this contradicts Tb 2 = −✶ for half-odd-integer spin.
– In fact, the inner product (|Ei, T |Ei) vanishes. By the definition of anti-unitary
operator, (|Ei, T |Ei) = (T |Ei, T T |Ei)∗ = (T |Ei, −|Ei)∗ = −(|Ei, T |Ei).
– For system with odd number of electrons and time-reversal symmetry, the energy
levels must be doubly degenerate. Because odd number of spin-1/2 can only
produce half-odd-integer total spin.

Summary of Lecture #6: perturbation theory
The Goals and The Requirements
• Understand the generic and formal approach to perturbative expansion for time-
independent perturbations.
– Viewpoint #1: perturbative expansion as a (asymptotic) series expansion of en-

ergy eigenvalues and eigenstates.
– Viewpoint #2: perturbative expansion as a sequence of unitary transformations,

trying to separate coupled modes/degrees of freedom.
• Understand the basic tools for dealing with time-dependent perturbations:

the interaction picture for Hamiltonian(Schrödinger picture) ĤS (t) = Ĥ0 + V̂S (t).
– Definitions: |ψ(t)iI = e✐Ĥ0 t/~ |ψ(t)iS , ÔI (t) = e✐Ĥ0 t/~ ÔS e−✐Ĥ0 t/~ .
Relation to Schrödinger picture: hψ(t)|S ÔS |φ(t)iS = hψ(t)|I ÔI (t)|φ(t)iI .
– Time evolution: ✐~ dt
d
|ψ(t)iI = V̂I (t)|ψ(t)iI .
✐
Rt ′ ′
|ψ(t)iI = ÛI (t)|ψ(t = 0)iI = ÛI (t)|ψ(t = 0)iS , formally ÛI (t) = T e− ~ 0 V̂I (t ) dt ,
Dyson series: ÛI (t) = ✶ + −~✐ t1 =0 V̂I (t1 ) dt1 +( −~✐ )2 t1 =0 t21=0 V̂I (t1 )V̂I (t2 ) dt1 dt2 +
Rt Rt Rt
...
– Note the similarity and difference to the Heisenberg picture (lecture notes #3).
Landau & Lifschitz, Quantum Mechanics: Non-relativistic Theory, Chapter VI.

I TIME-INDEPENDENT PERTURBATION THEORY FOR DISCRETE LEVELS
I. TIME-INDEPENDENT PERTURBATION THEORY FOR DISCRETE LEVELS
A. Nondegenerate Perturbation Theory: The Goal
(0)
• Given a Hamiltonian Ĥ0 with nondegenerate energy eigenvalues En and correspond-
(0)
ing normalized eigenstates |ψn i (n = 0, 1, · · · ), compute the energy eigenvalues and
eigenstates of Ĥ = Ĥ0 + λV̂ , in terms of series of the “small parameter” λ.
– In the |ni basis, Ĥ0 is a diagonal matrix (H0 )mn = En δmn with nondegenerate
(0) (0)
diagonal entries. V̂ is a generic matrix with matrix element Vmn = hψm |V̂ |ψn i.
– Diagonal and off-diagonal matrix elements of V̂ play different roles: diagonal Vnn
shifts energy of n-th level without changing eigenstate; off-diagonal Vmn mixes
n-th and m-th eigenstates and shifts their energies.
 q
E0 V12
– “Level repulsion”: eigenvalues of  ∗  is E0 +E1 ± ( E0 −E 1 2
) + |V12 |2 , the
V12 E1 2 2
p
difference is (E0 − E1 )2 + 4|V12 |2 ≥ |E1 − E0 |, adding off-diagonal perturbation
tends to increase level distance.
– It is usually assumed that for sufficiently small λ,

the n-th eigenstates |n, λi of Ĥ is adiabatically
(0)
connected to |ψn i (no level-crossing with increasing λ).
– Assume the n-th eigenvalue and n-th eigenstate are,
(0) (1) (2)
En = En + λEn + λ2 En + · · · ,
(0) (1) (2)
|ψn i = |ψn i + λ|ψn i + λ2 |ψn i + · · · . Note that |ψn i here is not normalized,
(k>0) (0)
and it is usually assumed that all perturbations |ψn i are orthogonal to |ψn i.
B. Formal Perturbative Expansion
• Define projection operators P̂n = |ψn ihψn | and Q̂n = ✶ − P̂n =

(0) (0) P (0) (0)
m6=n |ψm ihψm |.
(0)
Note that Ĥ0 P̂n = P̂n Ĥ0 = En P̂n , [Ĥ0 , Q̂n ] = 0 and P̂n Q̂n = Q̂n P̂n = 0.
• Consider the stationary state Schrödinger equation: (Ĥ0 + λV̂ )|ψn i = En |ψn i.
Use |ψn i = P̂n |ψn i + Q̂n |ψn i, assume Q̂n |ψn i is
a small perturbation to be solved. Rearrange terms,
(0)
(En − Ĥ0 )Q̂n |ψn i = (En − En )P̂ |ψn i + λV̂ |ψn i (∗).
Apply Q̂n on both sides of (∗), Q̂n (En − Ĥ0 )Q̂n · Q̂n |ψn i = 0 + λQ̂n V̂ |ψn i.

(0)
• The non-trivial eigenvalues of Q̂n (En − Ĥ0 )Q̂n are En − Em with m 6= n, assume all
1
these eigenvalues are nonzero, then this operator has an “inverse” Ĝn = Q̂n E Q̂n .
n −Ĥ0
(0)
defined on the space orthogonal to |ψn i, Ĝn · Q̂n (En − Ĥ0 )Q̂n = Q̂n . Note that
Ĝn Q̂n = Q̂n Ĝn = Ĝn , and Ĝn P̂n = P̂n Ĝn = 0.
• We can now “solve” the perturbation as Q̂n |ψn i = Ĝn Q̂n λV̂ |ψn i = Ĝn λV̂ |ψn i. Then
|ψn i = P̂n |ψn i + Ĝn λV̂ |ψn i, or (✶ − Ĝn λV̂ )|ψn i = P̂n |ψn i, or formally
|ψn i = (✶ − Ĝn λV̂ )−1 P̂n |ψn i = ∞ k
P
k=0 (Ĝn λV̂ ) P̂n |ψn i [1].
• Take inner product with P̂n |ψn i on both sides of (∗), we can “solve” the energy shift,
(0)
(En − En )hψn |P̂n |ψn i = hψn |P̂n λV̂ |ψn i = ∞ k
P
k=0 hψn |P̂n λV̂ (Ĝn λV̂ ) P̂n |ψn i [2].
(0) (0)
• P̂n |ψn i ∝ |ψn i, usually just choose P̂n |ψn i = |ψn i. Then hψn |ψn i ≥ hψn |P̂n |ψn i = 1.
(0) (0) (0)
• Example: 3-level problem, under the basis of (|ψ1 i, |ψ2 i, ψ3 i), Ĥ0 is diagonal. 
For
   
1 0 0 0 0 0 0 0 0
Q̂n=1 = ✶3×3 − P̂n=1 =

 
1
   
n = 1, P̂n=1 =  0 0 0 ,  0 1 0 , Ĝn=1 = .
    0 0 
 (0)
     E1 −E2 
1
 
0 0 0 0 0 1 0 0 (0)
E1 −E3
C. Formal Perturbative Expansion: Summary
• Summary:
|ψn i = (✶ − λĜn V̂ )−1 P̂n |ψn i = ∞ k k
P
k=0 λ (Ĝn V̂ ) P̂n |ψn i [1],
(0)
(En −En )hψn |P̂n |ψn i = ∞ k 1
P
k=0 hψn |P̂n λV̂ (Ĝn λV̂ ) P̂n |ψn i [2], where Ĝn = Q̂n E n −Ĥ0
Q̂n .
• Note that [1] and [2] do not really solve |ψn i and En in terms of known quantities.
Because Ĝn contains the unknown En .
(0)
• 1st-order perturbation: approximate |ψn i = P̂n |ψn i + O(λ) = |ψn i + O(λ). Then
(0) (0) (0)
En − En = λhψn |V̂ |ψn i + O(λ2 ) = λVnn + O(λ2 ).
(0) (0)
• 2nd-order perturbation: |ψn i = |ψn i + λĜn V̂ |ψn i + O(λ2 )
(0) (0) (0) (0)
= |ψn i + λQ̂n (0)1 Q̂n V̂ |ψn i + O(λ2 ) = |ψn i + λ m6=n (0)Vmn (0) |ψm i + O(λ2 ).
P
En −Ĥ0 En −Em
(0)
En − En = λVnn + λ2 m6=n V(0) nm Vmn 3
P
(0) + O(λ ).
En −Em
Note that off-diagonal Vnm always lowers ground state energy (level repulsion).
• For higher order perturbation, more accurate En should be used in Ĝn . For ex-
(0) (0) (0)
ample, the 3rd-order energy shift is En − En = λVnn + hψn |λV̂ Ĝn λV̂ |ψn i +

(0) (0) (0) (0)

hψn |λV̂ Ĝn λV̂ Ĝn λV̂ |ψn i + O(λ4 ). In the hψn |λV̂ Ĝn λV̂ |ψn i term, you need to use
(0) 1
λ1 -order approximation of Ĝn , then you need to use En ≈ En + λVnn , and (0) ≈
En −Em
1 1 λVnn (0) (0)
(0) (0) ≈ (0) (0) − (0) (0) in Ĝn . In the hψn |λV̂ Ĝn λV̂ Ĝn λV̂ |ψn i term,
En +λVnn −Em En −Em (En −Em )2
th
you can use the 0 -order approximation of Ĝn .
1
• Wavefunction re-normalization: hψn |ψn i = Zn
, Zn ≤ 1 is the weight of unperturbed
(0)
state |ψn i in |ψn i. Perturbative expansion is good when Zn is close to unity.
∂
Zn is related to Zn = (0) En .
∂En
P∞ P∞ k+k′ (0) ′ (0)
– ‘Proof’: Z1n = hψn |ψn i = k=0 k′ =0 λ hψn |(V̂ Ĝn )k (Ĝn V̂ )k |ψn i = 1 +
P∞ P∞ k+k′ (0) ′ (0) (0)
k=1 k′ =1 λ hψn |(V̂ Ĝn )k (Ĝn V̂ )k |ψn i. Note that Ĝn |ψn i = 0, so the
k = 0, k ′ 6= 0 or k ′ = 0, k 6= 0 terms do not contribute.
∂ (0) (0)
Take on both sides of [2] (fix Em for m 6= n, |ψm i, V̂ ), note that ∂ (0) Ĝn =
(0)
∂En ∂En
∂ 1
P (0) (0) −1 ∂ 2 ∂
(0) (Q̂n Q̂ ) = m6=n |ψm ihψm | (0) En = −(Ĝn ) (0) En , then
En −Ĥ0 n (0)
∂En (En −Em )2 ∂EnP ∂En
∂ (0) ∂ (0) ∞ k ∂ k (0)
(0) E n − 1 = λhψ n | V̂ (0) |ψ n i = λhψ n | V̂ k=1 λ (0) ( Ĝ n V̂ ) |ψn i
∂En ∂En ∂En
(0) (0)
= − ∂ (0) En · hψn |V̂ ∞
P k+1
Pk−1 k′ k−k′
∂En k=1 λ k′ =0 (Ĝn V̂ ) Ĝn (Ĝn V̂ ) |ψn i. Therefore
Pk−1 (0) (0)
( ∂ (0) En )−1 = 1 + ∞ k k′ k−k′
P
∂En k=2 λ k′ =1 hψn |(V̂ Ĝn ) (Ĝn V̂ ) |ψn i = hψn |ψn i.
• Non-degenerate perturbation is a good approximation when the off-diagonal terms are

(0) (0)
much smaller than original energy differences, |λVn,m | ≪ |En − Em | (n 6= m).
D. Example: Harmonic Oscillator Plus Linear Potential
p̂2 2 (0)
• Ĥ0 = 2m
+ mω
2
x̂2 = ~ω(b̂† b̂ + 12 ), with non-degenerate energy levels En = (n + 21 )~ω,
(0) (0) (0)
and unperturbed eigenstates |ψn i = √1n! (b̂† )n |ψ0 i, and b̂|ψ0 i = 0,
q
~ †
q
~ √ √
• Perturbation λV̂ = λx̂ = λ 2mω (b̂ + b̂ ), Vnm = 2mω ( mδn+1,m + nδn,m+1 ).
• 1st-order perturbation vanishes. 2nd-order perturbation correction to energy is

(0) 2 2 (0) (0)
λ2 n+1 λ2
En ≈ En + m6=n λEn|V−E nm | n
P
m
= E n + 2mω
( −ω
− ω
) = E n − 2mω 2.
p̂2 mω 2 2
• This happens to be the exact result. Ĥ0 + λV̂ = 2m
+ 2
x̂ + λx̂
p̂2 mω 2 λ 2 λ2 λ2
= 2m
+ 2
(x̂ + mω 2
) − 2mω 2
= ~ω(b̂′† b̂′ + 12 ) − 2mω 2
, with b̂′ = b̂ + √ λ
2~mω 3
.
• Due to inversion symmetry Iˆ : |xi 7→ | − xi, I(

ˆ Ĥ0 + λV̂ )Iˆ† = Ĥ0 − λV̂ . Eigenvalues
of Ĥ0 + λV̂ must be even function of λ. Therefore all odd-order perturbations vanish.

• Exercise: compute the 4th order correction and check that it vanishes.
E. Degenerate Perturbation Theory
(0) (0)
• If energy level En has g-fold degenerate orthonormal eigenstates |ψnα i, α = 1, · · · , g,
define P̂n = α |ψnα ihψnα | and Q̂n = ✶ − P̂n . The expansion is the same, except...
P (0) (0)
P (0)
• P̂n |ψn i = α cα |ψnα i is a linear combination with unknown coefficients cα to be solved.
• |ψn i = (✶ − λĜn V̂ )−1 P̂n |ψn i =

P P∞ (0)
β k=0 λk (Ĝn V̂ )k |ψnβ i cβ [1′ ].
(0)
• Take inner product with |ψnα i on both sides of (∗), we have the “secular equation”,
(0) (0) (0)
(En − En ) · (cα ) = β hψnα |V̂ ∞ k+1
(Ĝn V̂ )k |ψnβ i · (cβ ) [2′ ].
P P
k=0 λ
P (0)
• 1st-order perturbation: |ψn i = α cα |ψnα i + O(λ), the secular equation is
(0)
(En − En ) · cα = β λVnα,nβ · cβ + O(λ2 ), diagonalize the g × g matrix
P
(0) (0)
Vnα,nβ = hψnα |V̂ |ψnβ i to get the energy shift and cα .
P (0) P λVm,nα (0) 2
• 2nd-order perturbation: |ψn i = α cα (|ψnα i + m6=n E (0) −E (0) |ψm i) + O(λ ), the
n m
(0) P P λ2 Vnα,m Vm,nβ 3
secular equation is (En − En )cα = β (λVnα,nβ + m6=n E (0) −E (0) ) · cβ + O(λ ).
n m
• If 1st order perturbation completely removes degeneracy, we can use normalized cα

( α |cα |2 = 1) from 1st order perturbation here to compute 2nd-order energy shift,
P
(0) c∗ V Vm,nβ cβ
En − En = λ α,β c∗α Vnα,nβ cβ + λ2 m6=n α nα,m + O(λ3 ).
P P
(0) (0)
En −Em
(0) (0) (0) (0)

• Almost-degenerate perturbation: for En 6= Em , if |λVn,m | ≫ |En − Em |, we need
to use degenerate perturbation theory,

and treat the original energy difference as
E0 λV
perturbation. For example, for Ĥ = , when |λV | ≫ |E1 − E0 |, we should
λV E1
 
define Ĥ0 = E0 ✶2×2 , and

0 λV
  as perturbation.
λV E1 − E0
F. Example: Heisenberg Exchange from Hubbard Model
• For two sites i = 1, 2, on each site we have spin-1/2 electron modes ĉis , s =↑, ↓. The
unperturbed Hamiltonian is onsite Coulomb repulsion (Hubbard interaction), Ĥ0 =
U (n̂1↑ n̂1↓ + n̂2↑ n̂2↓ ). Consider electron number i,s n̂is = i,s ĉ†is ĉis = 2 subspace.
P P

• The spectrum of Ĥ0 is illustrated on the right.

(0)
The ground states are 4-fold degenerate, E0 = 0,
(0)
|ψ0α i = (ĉ†1↑ ĉ†2↑ |0i, ĉ†1↑ ĉ†2↓ |0i, ĉ†1↓ ĉ†2↑ |0i, ĉ†1↓ ĉ†2↓ |0i),
(0)
The excited states are two-fold degenerate, E1 = U ,
(0)
|ψ1α i = (ĉ†1↑ ĉ†1↓ |0i, ĉ†2↑ ĉ†2↓ |0i).
†
+ ĉ†2s ĉ1s ). 1st-order perturbation
P
• Consider electron hoppings, V̂ = −t s=↑,↓ (ĉ1s ĉ2s
vanishes. We have to solve the 2nd-order secular equation. For the ground states
(0) P P hψ(0) |V̂ |ψ1γ
(0) (0) (0)
ihψ1γ |V̂ |ψ0β i
subspace, (E0 − E0 )cα = β γ 0α (0) (0) · cβ . The 4 × 4 matrix is
E0 −E1
 
0 0 0 0
 
2t2 0 1 −1 0 † †
√1 (ĉ ĉ − ĉ†1↓ ĉ†2↑ )|0i, with
 
−U  0 −1 1 0
. The lowest energy state is spin singlet 2 1↑ 2↓
 
0 0 0 0
2
energy − 4tU , the remaining three spin triplet states have zero energy.
• This energy difference between spin singlet and triplets is effectively captured by
2
Heisenberg exchange, 4tU (Ŝ 1 · Ŝ 2 − 41 ), where Ŝ i = 12 s,s′ ĉ†is (σ)ss′ ĉis′ .
P
• Degenerate perturbation may be avoided by symmetry analysis. Define unitary op-

ˆ ĉ1s ↔ ĉ2s , which swaps the two sites. Then IˆĤ Iˆ† = Ĥ, and Iˆ2 = ✶. Iˆ
erator I:
generates a Z2 group {✶, I} ˆ with two irreps (Γ1,2 ), RΓ1 (I) = 1 and RΓ2 (I) = −1.
ˆ Then we can divide the
Define Ŝz ≡ 21 i (n̂i↑ − n̂i↓ ), then Ŝz commutes with Ĥ and I.
P
Hilbert space by Ŝz eigenvalues and irreps of the Z2 group, and solve the perturbation
theory in each subspace.
Ŝz Iˆ states Ĥ0 V̂
1 −1 ĉ†1↑ ĉ†2↑ |0i (0) (0)
−1 −1 ĉ†1↓ ĉ†2↓ |0i (0) (0)
  
† †
√1 (ĉ ĉ
2 1↑ 2↓
− ĉ†1↓ ĉ†2↑ )|0i 0 0 0 −2t
0 1   
† †
√1 (ĉ ĉ
2 1↑ 1↓
+ ĉ†2↑ ĉ†2↓ )|0i 0 U −2t 0
  
† †
√1 (ĉ ĉ
2 1↑ 2↓
+ ĉ†1↓ ĉ†2↑ )|0i 0 0 0 0
0 −1   
† †
√1 (ĉ ĉ
2 1↑ 1↓
− ĉ†2↑ ĉ†2↓ )|0i 0 U 0 0

II PERTURBATIVE EXPANSION AS UNITARY TRANSFORMATIONS
II. PERTURBATIVE EXPANSION AS UNITARY TRANSFORMATIONS
A. Prelude: A Harder Way to Solve a 2 × 2 Problem

   
E0 0 0 V
• Given Ĥ = Ĥ0 + λV̂ =  +λ  = E0 +E1 σ0 + E0 −E1 σz + λV σx , in orthonormal
0 E1 V 0 2 2
(0) (0)
basis (|ψ0 i, |ψ1 i). Try to “decouple” the two levels by unitary transformations.
(1) (1) (0) (0)

• Consider (|ψ0 i, |ψ1 i) = e✐λŜ0 (|ψ0 i, |ψ1 i), Hamiltonian in this basis is Ĥ (1) =
e✐λŜ0 Ĥe−✐λŜ0 = Ĥ0 + λ([✐Ŝ0 , Ĥ0 ] + V̂ ) + λ2 ( 12 [✐Ŝ0 , [✐Ŝ0 , Ĥ0 ]] + [✐Ŝ0 , V̂ ]) + · · · , demand
that Ĥ (1) has no off-diagonal term of order O(λ). Then we need [✐Ŝ0 , Ĥ0 ] + V̂ = 0,
V
which is Ŝ0 = σ .
E0 −E1 y
2 2 3 3 4 4
• Ĥ (1) = Ĥ0 + Eλ0 −E
V
1
σz − 43 (Eλ0 −E
V
1)
λ V
0 1
5 2
2 σx − (E −E )3 σz + O(λ ), eigenvalues up to O(λ ) are
λ2 V 2 λ2 V 2
E0 + E0 −E1
and E1 − E0 −E1
.
3 Ŝ 3 Ŝ
• This procedure can be continued: define Ĥ (2) = e✐λ 1
Ĥ (1) e−✐λ 1
, and choose Ŝ1 to
cancel the off-diagonal terms in Ĥ (1) of order λ3 , Ĥ (2) is then diagonal up to O(λ4 ),
the diagonal entries give approximate eigenvalues up to O(λ4 ), and the new basis
(2) (2) 3 Ŝ (0) (0)
(|ψ0 i, |ψ1 i) = e✐λ 1
e✐λŜ0 (|ψ0 i, |ψ1 i) are the approximate eigenstates up to O(λ3 ).
• Note: Because Ĥ (1) contains off-diagonal terms of only order λ3 and higher, the differ-
ence between Ĥ (2) diagonal terms and Ĥ (1) diagonal terms will be at least of order λ6 ,
1
coming from [✐λ3 Ŝ1 , (order λ3 off-diagonal terms of Ĥ (1) )] and 2
[✐λ3 Ŝ1 , [✐λ3 Ŝ1 , Ĥ0 ]].
Therefore the diagonal term of Ĥ (1) is already accurate up to order λ4 .
B. The General Idea
• Given Ĥ = Ĥ0 + λV̂ , Ĥ0 has known energy levels En (may be degenerate). Try to use
a sequence of unitary transformations to remove “off-diagonal terms”.
• Here “off-diagonal terms” means terms connecting eigenstates of Ĥ0 with different
energy. Therefore the procedure only block-diagonalizes the Hamiltonian. Terms
(diagonal and off-diagonal) connecting eigenstates of Ĥ0 with the same energy, namely
the matrix in the secular equation of degenerate perturbation theory, will be produced.

II PERTURBATIVE EXPANSION AS UNITARY TRANSFORMATIONS
• Define the projection operators P̂n projecting onto En eigenstate space. Then
✶ = n P̂n . Define V̂nm = P̂n V̂ P̂m (related to matrix element Vnm by V̂nm =
P
(0) (0) P
Vnm |ψn ihψm |), then V̂ = n,m V̂nm . Note that [Ĥ0 , V̂nm ] = (En − Em )V̂nm .
• Consider Ĥ (1) = e✐λŜ0 Ĥe−✐λŜ0 , we want to cancel V̂nm terms with n 6= m in Ĥ (1) , then
we need [✐Ŝ0 , Ĥ0 ] + n6=m V̂nm = 0. The solution is ✐Ŝ0 = n6=m EnV̂−E
P P nm
m
.
V̂mn ]+λ2 [✐Ŝ0 , n V̂nn ]+O(λ3 ) = Ĥ0 +λ n V̂nn +

V̂nn + λ2 [✐Ŝ0 ,
2
• Ĥ (1) = Ĥ0 +λ
P P P P
n m6=n
λ2 V̂n′ m V̂mn V̂n′ m V̂mn 2 V̂mn V̂nn −V̂mm V̂mn
+ O(λ3 ). The diagonal
P P
2 n′ 6=m,m6=n ( En′ −Em − Em −En ) + λ m6=n Em −En
terms are Ĥ0 + λ n V̂nn + λ2 m6=n V̂Enm V̂mn

+ O(λ3 ), same as series expansion result.
P P
n −Em
• This abstract procedure can be carried out to arbitrary order of λ (by computers).
, we demand that [✐λ2 Ŝ2 , Ĥ0 ] cancels the order
2 Ŝ 2 Ŝ
For next order, Ĥ (2) = e✐λ 2
Ĥ (1) e−✐λ 2
λ2 off-diagonal terms of Ĥ (1) . Note that [Ĥ0 , V̂n′ m V̂mn ] = (En′ − En )V̂n′ m V̂mn . Then
✐Ŝ2 = n′ 6=m,m6=n,n′ 6=n 2(E ′1−En ) ( V̂En′′m−E
V̂mn V̂ V̂
− Enm′ m−Emn 1 V̂mn V̂nn −V̂mm V̂mn
P P
m n
)+ m6=n Em −E n Em −En
. But
n n
the generated corrections to diagonal terms will be O(λ4 ). So diagonal terms in Ĥ (1)
is already accurate up to O(λ3 ).
C. Example: d-d Hoppings Mediated by d-p Hybridization
• Consider two transition metal(TM) ions connected by an oxygen,

the d-electrons can hop between the two TM ions
through oxygen p-orbitals. Ignore spin here.
Ĥ0 = ǫd (dˆ†1 dˆ1 + dˆ†2 dˆ2 ) + ǫp ĉ† ĉ. Here dˆ1,2 are d-electrons, ĉ is p-electron.
• The perturbation is d-p hybridization, V̂ = tdp (dˆ†1 ĉ − dˆ†2 ĉ + h.c.). NOTE the signs.
Define V̂+ = tdp (dˆ†1 ĉ−dˆ†2 ĉ) and V̂− = (V̂+ )† . Then [Ĥ0 , V̂± ] = ±(ǫd −ǫp )V̂± , V̂ = V̂+ +V̂− .
Note V̂ has no diagonal component (always change eigenvalues of Ĥ0 ).
• Ĥ (1) = e✐Ŝ0 Ĥe−✐Ŝ0 , and demand [✐Ŝ0 , Ĥ0 ] + V̂ = 0. Then ✐Ŝ0 = 1

ǫd −ǫp
(V̂+ − V̂− ).
Ĥ (1) ≈ Ĥ0 + 21 [✐Ŝ0 , V̂ ] = Ĥ0 + 1
[V̂
2(ǫd −ǫp ) +
− V̂− , V̂+ + V̂− ] = Ĥ0 + 1
[V̂ , V̂− ].
ǫd −ǫp +
• [V̂+ , V̂− ] = t2dp 2i,j=1 (−1)i+j [dˆ†i ĉ, ĉ† dˆj ] = t2dp 2i,j=1 (−1)i+j (dˆ†i dˆj − δij c† c)
P P
P ˆ† ˆ
= t2 (dp i d d − 2ĉ† ĉ) − t2 (dˆ† dˆ + h.c.).
i i dp 1 2
tdp 2
• The 2nd-order perturbation generates effective d-d hopping − ǫd −ǫ p
(dˆ†1 dˆ2 + h.c.).

III TIME-DEPENDENT PERTURBATION THEORY
D. Example: Spin Interactions from Hubbard Model (Not Required)
• Reference: A.H. MacDonald, et al., Phys. Rev. B 37, 9753 (1988).
• Consider the Hubbard model, Ĥ0 = U (n̂1↑ n̂1↓ + n̂2↑ n̂2↓ ), V̂ = −t s=↑,↓ (ĉ†1s ĉ2s + h.c.).
P
Define V̂+1 = −t s=↑,↓ (n̂1,−s ĉ†1s ĉ2s (1 − n̂2,−s ) + h.c.),

P
V̂−1 = −t s=↑,↓ ((1 − n̂1,−s )ĉ†1s ĉ2s n̂2,−s + h.c.),

P
V̂0 = −t s=↑,↓ ((1 − n̂1,−s )ĉ†1s ĉ2s (1 − n̂2,−s ) + n̂1,−s ĉ†1s ĉ2s n̂2,−s + h.c.),
P
P
then [Ĥ0 , V̂m ] = mU V̂m , and V̂ = m V̂m .
Note that by definition V̂m changes the number of “double-occupancy” by m.
1
• We want [✐Ŝ0 , Ĥ0 ] + V̂+1 + V̂−1 = 0, then ✐Ŝ0 = U
(V̂+1 − V̂−1 ). Ĥ (1) = e✐Ŝ0 Ĥe−✐Ŝ0 ≈
Ĥ0 + V̂0 + 12 [✐Ŝ0 , V̂+1 + V̂−1 ] + [✐Ŝ0 , V̂0 ]
• The diagonal terms (those commute with Ĥ0 ) are Ĥ0 + V̂0 + U1 [V̂+1 , V̂−1 ].
• Consider the action of these terms on the H0 ground states (single-occupancy states,
with n̂1 = n̂1↑ + n̂1↓ = 1 and n̂2 = n̂2↑ + n̂2↓ = 1), then only − U1 V̂−1 V̂+1 is
effective, because V̂−1 acting on single-occupancy states will vanish (V̂−1 will de-
crease the number of double-occupancy by 1). V̂+1 can move an electron from
site i to site j, create a double-occupancy on site j, then V̂−1 must move an elec-
tron from site j back to site i, to remove this double-occupancy. The effect of
2 P 2 P
this term is then − tU s,s′ ,i6=j (ĉ†is ĉjs )(ĉ†js′ ĉis′ ) = tU [ s,s′ ,i6=j ĉ†is ĉ†js′ ĉjs ĉis′ − s,i6=j n̂is ]
P
2 P 2
= tU [ i6=j (2Ŝ i · Ŝ j + 21 ) − 2] = 4tU (Ŝ 1 · Ŝ 2 − 14 ).
P † † 1
• Exercise: check the spin exchange s,s′ ĉis ĉjs′ ĉjs ĉis′ = 2Ŝ i · Ŝ j + 2 , in the single-
occupancy subspace. Here Ŝ i = 21 s,s′ ĉ†is (σ)ss′ ĉis′ .

P
III. TIME-DEPENDENT PERTURBATION THEORY
A. The interaction picture
• Given the Schrödinger picture time-dependent Hamiltonian ĤS (t) = Ĥ0 + V̂S (t), where
Ĥ0 is independent of time t, The time evolution is ✐~ ddt |ψ(t)iS = [Ĥ0 + V̂S (t)]|ψ(t)iS .
The goal is to “eliminate” Ĥ0 from this equation.

• Define the “interaction picture” states |ψ(t)iI = e✐Ĥ0 t/~ |ψ(t)iS .

Define the “interaction picture” operators ÔI (t) = e✐Ĥ0 t/~ ÔS e−✐Ĥ0 t/~ .
Note the similarity to the Heisenberg picture, and
hψ(t)|S ÔS |φ(t)iS = hψ(t)|I ÔI (t)|φ(t)iI .
• Time evolution of states: ✐~ ddt |ψ(t)iI = V̂I (t)|ψ(t)iI .
– “Proof”: ✐~ ddt |ψ(t)iI = (✐~ ddt e✐Ĥ0 t/~ )|ψ(t)iS + e✐Ĥ0 t/~ · ✐~ dt
d
|ψ(t)iS
= e✐Ĥ0 t/~ (−Ĥ0 )|ψ(t)iS + e✐Ĥ0 t/~ [Ĥ0 + V̂S (t)]|ψ(t)iS = e✐Ĥ0 t/~ V̂S (t)|ψ(t)iS
= e✐Ĥ0 t/~ V̂S (t)e−✐Ĥ0 t/~ · e✐Ĥ0 t/~ |ψ(t)iS = V̂I (t)|ψ(t)iI .
• Define the unitary time evolution operator in the interaction picture ÛI (t), |ψ(t)iI =
ÛI (t)|ψ(t = 0)iI = ÛI (t)|ψ(t = 0)iS . Then ✐~ ddt = V̂I (t)ÛI (t). The formal solution is
Rt
the time-ordered exponential ÛI (t) = T exp[− ~✐ 0 V̂I (t′ ) dt′ ], or ...
• Dyson series: ÛI (t) = ✶ + −~✐

Rt Rt R t1
V̂ (t ) dt1 + ( −~✐ )2
t1 =0 I 1 t1 =0 t2 =0
V̂I (t1 )V̂I (t2 ) dt1 dt2 + . . .
• Relation to Schrödinger picture time evolution operator: ÛI (t) = e✐Ĥ0 t/~ ÛS (t).
• Transition probability:
Let |ii and |f i be normalized eigenstates of Ĥ0 , Ĥ0 |ii = Ei |ii, Ĥ0 |f i = Ef |f i.
Start at state |ii at t = 0, evolve over time t, the probability of final state |f i is
✐ ✐
P (i → f, t) = |hf |US (t)|ii|2 = |hf |e− ~ Ĥ0 t ÛI (t)|ii|2 = |hf |e− ~ Ef t ÛI (t)|ii|2
= |hf |ÛI (t)|ii|2
B. Transition probability: constant perturbation


 Ĥ , t < 0,
0
• ĤS (t) =
 Ĥ0 + V̂ , t > 0.
V̂ is hermitian and t-independent. This is a constant perturbation turned on at t = 0.
• Keep up to 1st order term in the Dyson series of ÛI (t).

Rt
hf |ÛI (t)|ii = hf |ii + −~✐ t1 =0 e✐(Ef −Ei )t1 /~ hf |V̂ |ii dt1 + O(V 2 )
−✐
= hf |ii + ~
~ ✐(Ef −Ei )
(e✐(Ef −Ei )t/~ − 1)Vf i + O(V 2 ). Here Vf i ≡ hf |V̂ |ii.
E −E
4 sin2 ( f2~ i t)
The probability is (for f 6= i case), P (i → f, t) = (Ef −Ei )2
|Vf i |2

4 sin2 ( ∆E t)
• Transition rate: P (i → f, t)/t = t·(∆E)2
2~
|Vf i |2 , where ∆E = Ef − Ei .
2
Take t → +∞ limit, use lima→+∞ sinax(ax)
2 = πδ(x), then we have
• Fermi’s golden rule: transition rate is

P (i→f,t) 2π
Γ(i → f ) ≡ limt→+∞ t
= ~
δ(Ef − Ei )|Vf i |2 .
• Exercise: keep up to 2nd order terms in Dyson series, redo the calculation.
• The meaning of the δ(Ef − Ei ):

P
The total transition rate to leave the initial state is, Γi ≡ f, f 6=i Γ(i → f ). Then
P (i → i, t) ∼ (1 − Γi t) ∼ e−Γi t . The lifetime of state i is Γ1i . Formally the energy of
state i has imaginary part −✐ 2Γ~ i . Γi equals to dEf ρ′ (Ef )Γ(i → f ), where ρ′ (Ef ) is
R
the density of state, ρ′ (E) = f ′ , f ′ 6=i δ(E − Ef ′ ). Here f ′ may be a continuous label,
P
P
then f ′ is an integral. The Fermi golden rule means that the decay rate of the initial
2π ′
state is (to lowest order of perturbation), ~
ρ (Ei ) · (average of |Vf i |2 for Ef = Ei ).
C. Transition probability: harmonic perturbation


 Ĥ , t < 0,
0
• ĤS (t) =
 Ĥ0 + V̂ e✐ωt + V̂ † e−✐ωt , t > 0.
ω is a nonzero real constant. V̂ may not be hermitian, but is t-independent.
• Keep up to 1st order term in the Dyson series of ÛI (t).

Rt
hf |ÛI (t)|ii = hf |ii + −~✐ t1 =0 e✐(Ef −Ei )t1 /~ (e✐ωt Vf i + e−✐ωt Vf†i ) dt1 + O(V 2 )
†

Vf i ✐ V ✐
= hf |ii − (Ef −Ei +~ω) (e ~ (Ef −Ei +~ω)t − 1) + (Ef −Ei −~ω) (e ~ (Ef −Ei −~ω)t − 1) + O(V 2 ).
f i
The probability is (for f 6= i case),

4 sin2 ( ∆E+~ω t) 4 sin2 ( ∆E−~ω t)
P (i → f, t) = (∆E+~ω)2
2~
|Vf i |2 + (∆E−~ω) 2~
2 |Vf†i |2
4 sin( ∆E+~ω t) sin( ∆E−~ω t)
+ 2~
(∆E)2 −~2 ω 2
2~
(Vf∗i Vf†i e−✐ωt + (Vf†i )∗ Vf i e✐ωt ).
Take t → +∞ limit, the transition rate is given by
• Fermi’s golden rule: transition rate is

P (i→f,t)
Γ(i → f ) ≡ limt→+∞ t
= 2π
~
[δ(Ef − Ei + ~ω)|Vf i |2 + δ(Ef − Ei − ~ω)|Vf†i |2 ].
• Exercise: keep up to 2nd order terms in Dyson series, redo the calculation.

• Detailed balance (see Sakurai, section 5.7): note that Γ(i → f ) equals to Γ(f → i).
Roughly speaking, absorption(of energy ~ω) rate = emission rate.
D. Relation to time-independent theory
• Ĥ = Ĥ0 + V̂ .
(0) (0) (0) (0)
Denote normalized eigenstates of Ĥ0 as |ψn i, Ĥ0 |ψn i = En |ψn i, and
normalized eigenstates of Ĥ as |ψn i, Ĥ|ψn i = En |ψn i.
(0) (0)
For simplicity denote hψn |V̂ |ψm i as Vnm .
(0) (0) (0) (0)

• Consider hψn |ÛI (t)|ψn i = hψn |e✐Ĥ0 t/~ ÛS (t)|ψn i
(0) (0) P (0) (0) (0)
= e✐En t/~ hψn |( m |ψm ihψm |e−✐Em t/~ )|ψn i = m |hψn |ψm i|2 e−✐Em t/~ e✐En t/~ .
P
(0)
Consider non-degenerate case, |ψn i has large overlap with only |ψn i,
(0) (0) (0) (0) (0)
hψn |ÛI (t)|ψn i = Zn e−✐(En −En )t/~ + m, m6=n |hψn |ψm i|2 e−✐Em t/~ e✐En t/~
P
(0) (0)
(0) (0)
≈ Zn + Zn −~✐ (En − En )t + Zn 21 ( −✐(En −E n )t 2
) + m, m6=n |hψn |ψm i|2 e−✐Em t/~ e✐En t/~ .
P
~
(0)
Here Zn = |hψn |ψn i|2 is close to unity.
(0) (0) (0) (0)

• Compute hψn |ÛI (t)|ψn i by Dyson series. Up to 2nd order this is hψn |ÛI (t)|ψn i
Rt (0) (0) Rt Rt (0) (0)
= 1 + −~✐ t1 =0 dt1 hψn |V̂I (t1 )|ψn i + ( −~✐ )2 t1 =0 dt1 t21=0 dt2 hψn |V̂I (t1 )V̂I (t2 )|ψn i
Rt Rt (0) (0) (0) (0)
= 1+ −~✐ Vnn t+( −~✐ )2 t1 =0 dt1 t21=0 dt2 m e✐En t1 /~ Vnm e−✐Em t1 /~ ·e✐Em t2 /~ Vmn e−✐Em t2 /~
P
(0) (0) (0) (0)
Rt ✐(E −E )t /~
= 1 + −~✐ Vnn t + 21 ( −✐V~nn t )2 + ( −~✐ )2 m6=n Vnm Vmn t1 =0 dt1 e✐(En −Em )t1 /~ e m(0) n (0)1 −1
P
✐(Em −En )/~
= 1 + −~✐ (Vnn + m6=n V(0) nm Vmn 1 −✐Vnn t 2

P
(0) ) · t + 2 ( ~
)
En −Em
(0) (0)
✐(En −Em )t/~
− ( −~✐ )2 m6=n Vnm Vmn (0) e (0) −1
P
(0) (0)
✐(Em −En )/~·✐(En −Em )/~
= [1 − m6=n V(0)nm Vmn ] + −~✐ (Vnn + m6=n V(0) nm Vmn 1 −✐Vnn t 2

P P
(0) (0) ) · t + 2 ( ~
)
(En −Em )2 En −Em
(0) (0)
✐(E −E )t/~
− ( −~✐ )2 m6=n Vnm Vmn (0) e (0)n m (0) (0) .
P
✐(Em −En )/~·✐(En −Em )/~
Compare with previous results, we see that up to 2nd order,
(0)
• En − En = Vnn + m6=n V(0) nm Vmn
P
(0) ,
En −Em
Zn = 1 − m6=n V(0)nm Vmn
P
(0) 2 .
(En −Em )
Consistent with time-independent perturbation theory.

Summary of Lecture #7: scattering theory
• Understand the basic tools in scattering theory:

perturbation theory applied to continuous spectrum of
the Hamiltonian of (free particle Ĥ0 + scattering potential V̂ )
– For sufficiently short-ranged potential, scattering state energy is the same as

plane wave state energy for Ĥ0 .
1
– Lippmann-Schwinger equation: |ψi = |ki + E0 −Ĥ0 +✐ǫ
V̂ |ψi.
Note the +✐ǫ produces correct t → ±∞ limit.
1
– Born approximation: |ψi ≈ |ki + E0 −Ĥ0 +✐ǫ
V̂ |ki.
– Scattering matrix (S-matrix): unitary time evolution operator in interaction pic-

ture.
• Note: we will assume that the space is three dimensional.
• Understand some basic concepts in scattering theory
– Scattering cross section, σ: total scattered particle current under unit incoming
particle current density.
dσ
– Differential scattering cross section, dΩ
: scattered particle current dσ into the
solid angle element dΩ, divided by dΩ, under unit incoming current density.
4π
– Optical theorem: σ = k
Im[f (k, k)].
Total cross section is related to the forward scattering amplitude.
• We will not be dealing with inelastic scattering in class.
Landau & Lifschitz, Quantum Mechanics: Non-relativistic Theory, Chapter XVII.

II LIPPMANN-SCHWINGER EQUATION
I. SETUP OF SCATTERING PROBLEM
• Rough picture (time-dependent):

particle beam (plane wave packet) |ψ(t = 0)i coming in, interacting with V̂ ;
scattered particles (scattered wave) |ψ(t → +∞)i going out.
• Rough picture (time-independent):

a (short-ranged) scattering potential V̂ as a perturbation to free particle Hamilto-
nian Ĥ0 . The unperturbed state is momentum eigenstate |ki (k is wavevector, ~k is
momentum). The perturbed state contains scattered waves.
– difficulties in treating scattering theory as perturbation theory: infinite degen-

eracy, because there are infinitely many |k′ i with the same energy (|k′ | = |k|)
under Ĥ0 as |ki.
2
~
• We will consider non-relativistic particles only. Ĥ0 = − 2m ∇2 .
• Particle number (probability) current is Re[ψ ∗ (r)(−✐ m~ ∇)ψ(r)].
A. Scattering cross section
dσ
• Differential cross section: dΩ
.
dσ is the particle current being scattered into the solid angle element dΩ, under unit
incoming particle current density.
dσ
R
• Cross section: σ = dΩ
dΩ.
Total scattered particle current, under unit incoming particle current density.
• Incoming plane wave |ki is not normalizable. Probability is not easily defined. To
overcome this, one can use free wave packet instead of plane wave. See e.g. Sakurai,
Chapter 7.
II. LIPPMANN-SCHWINGER EQUATION
• Take the time-independent approach, (Ĥ0 + V̂ )|ψi = E|ψi.

II LIPPMANN-SCHWINGER EQUATION
• Assume that V̂ is sufficiently short-ranged, namely for position r at a large distance

from the scatterer, V̂ (r) ≈ 0. And |ψi has the plane wave component |ki. Consider
the Schrödinger equation around r (where we can ignore V̂ ), we conclude that E =
~2 k 2
E0 = 2m
.
• Formally (see also series expansion approach to time-independent perturbation),

1
|ψi = |ψ0 i + E−Ĥ0
V̂ |ψi, where |ψ0 i is Ĥ0 eigenstate, usually taken as |ki.
1
We however need to avoid the singularity of E−Ĥ0
.
1
• Lippmann-Schwinger equation: |ψi = |ki + E−Ĥ0 +✐ǫ
V̂ |ψi, where ǫ → 0+.
• Understanding +✐ǫ (not rigorous):

Lippmann-Schwinger equation can be rewritten as (Ĥ0 + V̂ )|ψi = (E + ✐ǫ)|ψi − ✐ǫ|ki.
d
Consider Schrödinger equation ✐~ ddt |ψi = (Ĥ0 + V̂ )|ψi, we have ✐~ dt (e✐Et/~−ǫt/~ |ψi) =
e✐Et/~−ǫt/~ −1
−e✐Et/~−ǫt/~ · ✐ǫ|ki, solution is e✐Et/~−ǫt/~ |ψi = |ψ(t = 0)i − ✐E−ǫ
· ✐ǫ|ki, or
ǫ ǫ
|ψi = e−✐Et/~+ǫt/~ (|ψ(t = 0)i + E+✐ǫ
|ki) − E+✐ǫ
|ki.
Then |ψi becomes the plane wave when t → −∞, mimic the scattering process.
• Insert resolution of identity into Lippmann-Schwinger equation

hr|ψi = hr|ki + d3 r ′ d3 p hr| ~2 k2 1~2 p2 |pihp|r ′ ihr ′ |V̂ |ψi.
R R
2m
− 2m +✐ǫ
1
Note that hr|ψi = ψ(r), hr|pi = (2π)3/2
e✐p·r ,
hr ′ |V̂ = V (r ′ )hr ′ | (assume V̂ depends on
✐p·(r−r ′ )
position only), then ψ(r) = (2π)13/2 e✐k·r + d3 r ′ d3 p ke2 −p2 +✐ǫ · (2π)2m ′ ′
R R
3 ~2 · V (r )ψ(r ). The
integration over p can be done. Use polar coordinates and define the +z direction
✐p·(r−r ′ ) R∞ Rπ R 2π ✐p|r−r ′ | cos θ
along r − r ′ . We have d3 p ke2 −p2 +✐ǫ = 0 p2 dp 0 sin θ dθ 0 dφ e k2 −p2 +✐ǫ =
R
′ ′ ′
R∞ 2 R1 e✐p|r−r | cos θ
R∞ 2 2π (e✐p|r−r | −e−✐p|r−r | )
0
p dp −1
d(cos θ) 2π 2 2
k −p +✐ǫ
= 0
p dp ✐p|r−r | ′ 2 2
k −p +✐ǫ
=
R∞ ✐p|r−r ′ | ✐z|r−r ′ | 2
2π e 2π e 2π ✐k|r−r ′ |
−∞
pdp ✐|r−r ′ | k 2 −p2 +✐ǫ = 2π ✐ Resz=+√k2 +✐ǫ (z · ✐|r−r ′ | · k 2 +✐ǫ−z 2 ) = − |r−r ′| e .
Finally
′
1 2m e✐k|r−r |
e✐k·r d3 r ′ · V (r ′ )ψ(r ′ ).
R
• ψ(r) = (2π)3/2
− ~2 4π|r−r ′ |
• Further approximation: |r| ≫ |r ′ |, the position where we measure the scattered wave is
′
very far(compared to the range of V̂ ) from the scattering potential, |r −r ′ | ≈ |r|− r·r |r|
,
✐kr R ′ ′
ψ(r) ≈ (2π)13/2 e✐k·r − 2m e
~2 4πr
d3 r ′ e−✐k ·r V (r ′ )ψ(r ′ ), where k′ = |k| |r|
r
. This is usually
written as

IV BORN APPROXIMATION
✐kr

• ψ(r) ≈ 1
(2π)3/2
e✐k·r + f (k′ , k) e r , where k′ = |k| |r|
r
. This looks like superposition of
incoming plane wave and outgoing spherical wave (with angle dependent amplitude).
3/2 R 3 ′ −✐k′ ·r′
f (k′ , k) = − (2π)
4π
2m
~ 2 dr e V (r ′ )ψ(r ′ ), has the unit of length.
• The outgoing particle current in solid angle element dΩ along k′ direction at distance
✐kr ✐kr
r is Re{[ (2π)13/2 f (k′ , k) e r ]∗ (−✐ m~ ∂r
∂
)[ (2π)13/2 f (k′ , k) e r ]} · r2 dΩ = ~k 1
m (2π)3
|f (k′ , k)|2 dΩ.
The incoming particle current density is Re[( (2π)13/2 e✐k·r )∗ (−✐ m~ ∇)( (2π)13/2 e✐k·r )] =
~k 1
m (2π)3
. The differential cross section is the ratio of the two,
• dσ
dΩ
= |f (k′ , k)|2
III. OPTICAL THEOREM
• Not very rigorous derivation (see e.g. Prof. Murayama’s notes for rigorous one):
Consider a sphere of radius r, the total outgoing particle current should be zero,
✐kr

∂
)ψ(r)] = 0, where ψ = (2π)13/2 e✐k·r + f (k′ , k) e r .
R 2
r dΩ Re[ψ ∗ (r)(−✐ m~ ∂r
Rπ R 2π
Use polar coordinates and define +z direction along k, 0 = r2 0 sin θdθ 0 dφ
2
~ 1
m (2π)3
{k cos θ + k |f (θ,φ)|
r2
+ Re[ k cos
r
θ ∗
f (θ, φ)e✐k(cos θ−1)r + ( kr + −✐
r2
)f (θ, φ)e−✐k(cos θ−1)r ]},
Rπ R 2π Rπ R 2π
so σ = 0 sin θdθ 0 dφ |f (θ, φ)|2 = −r· 0 sin θdθ 0 dφ Re[cos θ·f ∗ (θ, φ)e✐k(cos θ−1)r +
−✐
(1 + kr
)f (θ, φ)e−✐k(cos θ−1)r ].
Consider the case kr ≫ 1 and average over r, the contribution to the integral comes
mainly from θ ≈ 0 region, and φ dependence of f is small there. Define x = 1 −
R2 −2✐kr
cos θ, σ ≈ −r · 0 dx 2π Re[f ∗ (0, 0)e−✐krx + f (0, 0)e✐krx ] = −r · 2π Re[f ∗ (0, 0) e −✐kr−1 +
2✐kr −1
f (0, 0) e ✐kr
], ignore the fast oscillating factor e±2✐kr (for averaging over r), we have
4π
• Optical theorem: σ = k
Im[f (k, k)].
Total cross section is related to the forward scattering amplitude.
IV. BORN APPROXIMATION
• Like in 1st order approximation in the series expansion approach to time-independent

perturbation theory, approximate the unknown |ψi on the right-hand-side of
Lippmann-Schwinger equation by the incoming plane wave, we have the

V PARTIAL WAVE EXPANSION
1
• Born approximation: |ψi = |ki + E−Ĥ0 +✐ǫ
V̂ |ki
3/2 ′ ✐k·r ′
• f (k′ , k) = − (2π) 2m ′ e 2m ′
d3 r ′ e−✐k ·r V (r ′ ) (2π) d3 r ′ V (r ′ ) e−✐q·r ,
R R
4π ~2 3/2 = − 4π~2
where q = k′ − k is the momentum transfer. So f (k′ , k) is proportional to the Fourier

transform of V (r) at momentum transfer wavevector.
• If the scattering potential V (r) is central (depends on r = |r| only),

f (k′ , k) = − 4π~
2m
r dr sin θdθ dφ V (r) e−✐qr cos θ
R 2 R R
2
−✐qr ✐qr
2m
r dr 2π e −✐−e 2m
dr r sin(qr)
R 2 R
= − 4π~ 2 qr
V (r) = − ~ 2 q
V (r)
• One can apply higher order approximation to the Lippmann-Schwinger equation,

and obtain the Born expansion, |ψi = |ki+ E−Ĥ1 +✐ǫ V̂ |ki+ E−Ĥ1 +✐ǫ V̂ 1
E−Ĥ0 +✐ǫ
V̂ |ki+. . . .
0 0
V. PARTIAL WAVE EXPANSION
• Consider the case of central potential V (r) = V (r), then f (k′ , k) is only function of
the angle θ between k′ and k, define the k direction as +z direction,
✐kr
ψ(r) ∼ e✐kz + f (θ) e r .
• Expand the plane wave into spherical waves,

e✐kz = ∞ l
P
l=0 (2l + 1)✐ jl (kr)Pl (cos θ),
where Pl is Legendre polynomial, jl is spherical Bessel function.
1
P∞
• For large r, e✐kz ∼ 2✐kr l=0 (2l + 1)(e✐kr − (−1)l e−✐kr )Pl (cos θ),
P∞
• Expand f (θ) into Legendre polynomials, f (θ) = l=0 (2l + 1)fl Pl (cos θ).
dΩ |f |2 = 4π 2
R P
• σ= l (2l + 1)|fl | ,
Rπ 2
where we have used 0
sin θdθ Pl′ (cos θ)Pl (cos θ) = δ′ .
2l+1 l l
4π 4π
P
• By optical theorem, σ = k
Im[f (0)] = k l (2l + 1)Im(fl ),
where we have used Pl (1) = 1.
• Comparing the two results, |fl |2 = k1 Im(fl ), or |1 + 2✐kfl |2 = 1.

Define phase shift δl such that 1 + 2✐kfl = e2✐δl , then fl = k1 e✐δl sin δl .
1
+ 1)Pl (cos θ)[e2✐δl e✐kr − (−1)l e−✐kr ].
P
• ψ(r) ∼ 2✐kr l (2l
4π
+ 1) sin2 δl .
P
σ= k2 l (2l

VI SCATTERING MATRIX (S-MATRIX)
q
Yl (θ, φ)Rl (r), where Yl (θ) = 2l+1
0 0
P
• Expand ψ(r) = l 4π
Pl (cos θ),
h 2
i
then the radial wavefunction satisfies − 1r ddr2 r + l(l+1)
r 2 + 2m
~ 2 V (r) Rl (r) = k 2 Rl (r),
1 1
and Rl (r) ∼ 2✐kr
√ [e2✐δl e✐kr − (−1)l e−✐kr ] as r → ∞.
4π(2l+1)
• To solve δl , solve the radial wavefunction Rl (r) first, expand its asymptotic (r → ∞)
form by spherical Bessel functions jl (kr) and nl (kr), Rl (r) ∼ jl (kr) cos δl +nl (kr) sin δl .
– Reminder about spherical Bessel function: jl (ρ) and nl (ρ),

h 2
i
solutions to − ρ1 ddρ2 ρ + l(l+1)
ρ 2 R(ρ) = R(ρ).
jl (ρ) = (−ρ)l ( ρ1 ddρ )l ( sinρ ρ ), nl (ρ) = −(−ρ)l ( ρ1 ddρ )l ( cosρ ρ ).
– Asymptotic behavior:
ρl
ρ → 0, jl (ρ) ∼ (2l+1)!!
, nl (ρ) ∼ − (2l−1)!!
ρl
.
ρ → ∞, jl (ρ) ∼ sin(ρ−lπ/2)
ρ
, nl (ρ) ∼ − cos(ρ−lπ/2)
ρ
.
VI. SCATTERING MATRIX (S-MATRIX)
• Ŝ is the unitary time evolution operator in interaction picture, under the limit that
the initial time ti → −∞ and the final time tf → +∞, Ŝ ∼ limti →−∞,tf →+∞ ÛI (tf , ti ).
• From Dyson series, Ŝ = ✶ + −✐

R tf R tf R t1
~ ti
dt1 V̂I (t1 ) + ( −~✐ )2 ti
dt1 ti
dt2 V̂I (t1 )V̂I (t2 ) + . . . ,
with V̂I (t) = e✐Ĥ0 t/~ V̂ e−✐Ĥ0 t/~ .
• The limit ti → −∞ and tf → +∞ will formally cause non-convergent integrals. To

avoid this, assume that the potential (in Schrödinger picture) V̂ depends on time and
decays to zero as t → ±∞, V̂ (t) = V̂ e−ǫ|t| with ǫ > 0.
R +∞
• The 1st order term in Dyson series is −~✐ −∞ dt1 e✐Ĥ0 t1 /~ V̂ e−ǫ|t| e−✐Ĥ0 t1 /~
R +∞
= k,k′ −~✐ −∞ dt1 |kie✐Ek t1 /~ hk|V̂ |k′ ie−ǫ|t1 | e−✐Ek′ t1 /~ hk′ |
P
R0 R +∞
= k,k′ −~✐ |kihk|V̂ |k′ ihk′ |( −∞ dt1 e✐(Ek −Ek′ )/~·t1 +ǫt1 + 0 dt1 e✐(Ek −Ek′ )/~·t1 −ǫt1 )
P
= k,k′ −~✐ |kihk|V̂ |k′ ihk′ |( ✐(Ek −E1 ′ )/~+ǫ − ✐(Ek −E1 ′ )/~−ǫ )
P
k k
P −✐ ′ ′ 2ǫ
= k,k′ ~ |kihk|V̂ |k ihk | (Ek −E ′ )2 /~2 +ǫ2
k
∼ k,k′ −~✐ |kihk|V̂ |k′ ihk′ | · 2πδ( k ~ k ) = − k,k′ |kihk|V̂ |k′ ihk′ | · 2π ✐δ(Ek − Ek′ ),
P E −E ′ P
2ǫ
note that formally limǫ→+0 x2 +ǫ2
= 2πδ(x).

VI SCATTERING MATRIX (S-MATRIX)
R∞ R t1
• The 2nd order term in Dyson series is ( −~✐ )2 −∞ dt1 −∞ dt2 V̂I (t1 )V̂I (t2 )
∞ t
= ( −~✐ )2 −∞ dt1 −∞ dt2 k1 ,k2 ,k3 |k1 ie✐Ek1 t1 /~ hk1 |V̂ |k2 ie−ǫ|t1 | e−✐Ek2 t1 /~
R R 1
P
× e✐Ek2 t2 /~ hk2 |V̂ |k3 ie−ǫ|t2 | e−✐Ek3 t2 /~ hk3 |

= ( −~✐ )2 k1 ,k2 ,k3 |k1 ihk1 |V̂ |k2 ihk2 |V̂ |k3 ihk3 |
P
hR i
0 R t1 R∞ R0 R t1
× −∞ dt1 −∞ dt2 + 0 dt1 ( −∞ dt2 + 0 dt2 ) e✐(Ek1 −Ek2 )t1 /~−ǫ|t1 | e✐(Ek2 −Ek3 )t2 /~−ǫ|t2 |
R0 ✐(Ek −Ek )t1 /~+ǫt1
= ( −~✐ )2 k1 ,k2 ,k3 |k1 ihk1 |V̂ |k2 ihk2 |V̂ |k3 ihk3 | × [ −∞ dt1 e ✐(Ek1 −E3k )/~+ǫ
P
2 3
R∞ ✐(E −Ek )t1 /~−ǫt1 ✐(E −Ek )t1 /~−ǫt1 ✐(E −Ek )t1 /~−ǫt1
e k1 2 e k1 3 −e k1 2
+ 0 dt1 ( ✐(Ek −Ek )/~+ǫ + ✐(Ek −Ek )/~−ǫ
)]
2 3 2 3
= ( −~✐ )2 k1 ,k2 ,k3 |k1 ihk1 |V̂ |k2 ihk2 |V̂ |k3 ihk3 |
P
1 1 1
× [ ✐(Ek (
−Ek3 )/~+ǫ ✐(Ek1 −Ek3 )/~+ǫ
− ✐(Ek −Ek2 )/~−ǫ
)
2 1
1 1 1
− ✐(Ek
(
−Ek3 )/~−ǫ ✐(Ek1 −Ek3 )/~−ǫ
− ✐(Ek −Ek2 )/~−ǫ
)].
2 1
Consider limǫ→+0 x+1✐ǫ = −✐πδ(x)+P( x1 ) where P means Cauchy principal value, above
result is − k1 ,k2 ,k3 |k1 ihk1 |V̂ |k2 i Ek −E1 k +✐ǫ hk2 |V̂ |k3 ihk3 | × 2π ✐δ(Ek1 − Ek3 )
P
1 2
= − k1 ,k3 |k1 ihk1 |V̂ E −1Ĥ +✐ǫ V̂ |k3 ihk3 | × 2π ✐δ(Ek1 − Ek3 )
P
k1 0
• Finally hk|(Ŝ − ✶)|k′ i = −2π ✐δ(Ek − Ek′ )hk|(V̂ + V̂ 1

Ek −Ĥ0 +✐ǫ
V̂ + . . . )|k′ i.
Note the similarity to Born expansion.
• From k′ hk|Ŝ † |k′ ihk′ |Ŝ|ki = hk|ki (because Ŝ is unitary), we have

P
k′ [hk|k ihk |(Ŝ − ✶)|ki + hk|(Ŝ − ✶)|k ihk |ki + |hk |(Ŝ − ✶)|ki| ] = 0, or
P ′ ′ † ′ ′ ′ 2
− k′ 4πδ(Ek − Ek′ )Im[hk|k′ iT (k′ , k)] = 2 ′ 2

P P
k′ [2πδ(Ek − Ek )] |T (k , k)| , where
′
T (k′ , k) = hk′ |(V̂ + V̂ 1

V̂
+ . . . )|ki. This is formally the
Ek′ −Ĥ0 +✐ǫ
optical theorem −Im[T (k, k)] = k′ πδ(Ek − Ek′ )|T (k′ , k)|2 .
P
In 3D space, f (k′ , k) = − 2m · 2π 2 · T (k′ , k), and k′ δ(Ek − Ek′ ) · · · = dΩ 2m k

P R
~2 ~2 2
...,
this recovers the previous optical theorem.
• For central potential problem, the S-matrix is block-diagonal in the basis of angular
momentum ℓ free spherical waves, the matrix element of Ŝ between inward spherical
wave and outward spherical wave is exp(2✐δℓ ) where δℓ is the phase shift.
See e.g. Prof. Murayama’s lecture notes for more details.

VII EXAMPLES
VII. EXAMPLES
A. Coulomb potential scattering: Born approximation
ZZ ′ e2
• V (r) = V (r) = r
, Z and Z ′ are charges (in unit of elementary charge e) of
the point-like scatterer and the incoming particle respectively, Gauss unit is used (to
change to SI unit, replace e2 by e2 /4πǫ0 ).
• Under Born approximation:

1 ∞ 2m 1 ∞
f (k′ , k) = − 2m dr (ZZ ′ e2 ) sin(qr).
R R
2
~ q 0
dr rV (r) sin(qr) = − 2
~ q 0
This however does not seem to converge.
• Consider a short-ranged Yukawa potential V (r) = (ZZ ′ e2 ) exp(−µr)

r
.
Under Born approximation:
1 ∞
f (k′ , k) = − 2m dr (ZZ ′ e2 ) exp(−µr) sin(qr) = − 2m 1
(ZZ ′ e2 ) q2 +µ
R
~2 q 0 ~2 2.
Note that q = k′ − k, so q = 2k sin(θ/2) where θ is the scattering angle (between k′

and k),
f (θ) = − 2m
~2
1
(ZZ ′ e2 ) 4k2 sin2 (θ/2)+µ 2.
dσ
• Differential cross section: dΩ
= |f (θ)|2 = [ 2m
~2
1
(ZZ ′ e2 )]2 ( 4k2 sin2 (θ/2)+µ 2
2) .
dσ
d(cos θ)[ 2m 1
(ZZ ′ e2 )]2 ( 2k2 (1−cos )2
R R R
• Total cross section: σ = sin θdθ dφ dΩ = 2π ~2 θ)+µ2
= [ 2m
~2
(ZZ ′ e2 )]2 4k2 µ4π2 +µ4
• Finally take the limit that µ → 0,

dσ
dΩ
= [ 2m
~2
(ZZ ′ e2 )]2 q14 = [ 2m
~2
(ZZ ′ e2 )]2 16k4 sin1 4 (θ/2) ;
σ → ∞.
• σ diverges because Coulomb potential is a long-ranged, no matter how off-target the

incoming particle is from the scatterer, it will be affected.
• Form factor:
if the scatterer’s charge has a distribution ρ(r ′ ) [Z =
R 3 ′
d r ρ(r ′ )], V (r) =
R 3 Z ′ e2
d r |r−r′ | ρ(r ′ ). This is a convolution of Coulomb potential with ρ(r ′ ), under Born
approximation, the scattering amplitude is the Fourier transform of V , so is the prod-
uct of Fourier transforms of Coulomb potential and ρ,

VII EXAMPLES
′ 2
f (θ) = − 2m Z e
d3 r ′ ρ(r ′ ) exp(−✐q · r ′ ) = fpoint scatterer (θ) · F (q),
R
~2 q 2
where the form factor F (q) = Z1 d3 r ′ ρ(r ′ ) exp(−✐q · r ′ ).

R
By measuring differential cross section and comparing with that of point scatterer,
one can measure |F (q)|2 and obtain information about the charge distribution ρ.
• Bragg scattering: e.g. X-ray or neutron diffraction on crystal,

if the scatterer has a periodic density distribution, e.g. electrons in crystals, ρ(r ′ ) =
ρ(r ′ + a1 ) = ρ(r ′ + a2 ) = ρ(r ′ + a3 ) when the arguments are within the paral-
lelepiped spanned by La1 , La2 , La3 . Under Born approximation, the scattering am-
plitude is f (q) = fpoint scatterer (q) d3 r ′ ρ(r ′ ) exp(−✐q · r ′ ). The total form factor is
R
F (q) = unit cell d3 r ′ ρ(r ′ ) exp(−✐q · r ′ ) L−1

R P PL−1 PL−1
x=0 y=0 z=0 exp[−✐q · (xa1 + ya2 + za3 )]
1−e−L✐q·a1 1−e−L✐q·a2 1−e−L✐q·a3
= Funit cell (q) · 1−e−✐q·a1 1−e−✐q·a2 1−e−✐q·a3
, differential cross section is proportional to
sin 2 (Lq·a /2) sin2 (Lq·a /2) sin2 (Lq·a /2)
|F (q)|2 = |Funit cell (q)|2 sin2 (q·a11/2) sin2 (q·a22/2) sin2 (q·a33/2) . The differential cross section
is peaked when q · a1,2,3 = 0 mod 2π, namely q is a reciprocal lattice vector (Bragg’s
law). The peak width (in k-space) is proportional to 1/L namely inverse of sample’s
linear size.
B. Hard sphere: s-wave scattering


 0, r > a;
• V (r) =
 V0 , r < a.
• For small momentum k ≪ 1/a, scattering happens mainly in the s-wave channel
2
(l = 0). The radial Schrödinger equation is [− ddr2 + 2m
~2
V (r)](r R0 (r)) = k 2 (r R0 (r)).
q
• Repulsive case: V0 > 0, define K = 2mV ~2
0
.
 sin(√k 2 − K 2 r), r < a;


– For k > K, r R(r) =

 A sin(ka + δ0 ), r > a.
√
The phase shift δ0 = arc tan[ √k2k−K 2 tan( k 2 − K 2 a)] − ka.
For k ≫ K, δ0 vanishes (as K 2 /k 2 ) and so does the cross section.
 sinh(√K 2 − k 2 r), r < a;

– For k < K, r R(r) =

 A sin(ka + δ0 ), r > a.
√
The phase shift δ0 = arc tan[ √K 2k−k2 tanh( K 2 − k 2 a)] − ka.

VII EXAMPLES
For k ≪ K, δ0 ∼ ka[ tanh(Ka)

Ka
− 1].
dδ0
Define the s-wave scattering length, a0 = − limk→0 dk
, in this case a0 = a [1 −
tanh(Ka)
Ka
]
> 0. For infinitely repulsive potential (V0 → +∞, K → +∞), a0 = a.
q
• Attractive case: V0 < 0, define K = − 2mV ~2
0
.
 sin(√K 2 + k 2 r), r < a;


– r R(r) =
 A sin(ka + δ0 ), r > a.
√
The phase shift δ0 = arc tan[ √K 2k+k2 tan( K 2 + k 2 a)] − ka.
dδ0 tan(Ka)
The scattering length a0 = − limk→0 dk
= a [1 − Ka
].
For small Ka ≪ 1, this is negative.
– For Ka = π/2 + nπ, n = 0, 1, . . . , the scattering length

 a0 diverges. This is the
 sin(Kr), r < a;
condition to have a zero energy bound state r R(r) =
 constant, r > a.
k
√
1+✐ √ tan( k2 +K 2 a)
2✐δ0 −2✐ka k2 +K 2
– The S-matrix element e = e 1−✐ √ 2k
√
tan( k2 +K 2 a)
can have a pole at
√ k +K 2
2 2 2 2
k = ✐κ, if κ = − tan(√KK 2−κ
−κ2 a)
. κ
This is the condition to have a energy − ~2m bound
 sin(√K 2 − κ2 r), r < a;

state r R(r) =
 A exp(−κr), r > a.
• In summary,
the scattering length is positive(usually negative) for repulsive(attractive) potential;
scattering length can diverge if a bound state energy approaches zero;
poles of S-matrix (with complex k) indicate bound states.

Summary of Lecture #8: introduction to relativistic
quantum mechanics
• Have some basic understanding about the Dirac equation:

p
– One way to reconcile special relativity E = m2 c4 + c2 p2 and
quantum mechanics ψ ∼ exp[✐(p · x − E · t)/~].
– ✐~ ddt ψ = (cα · p̂ + mc2 β)ψ.

αi and β anti-commute, are hermitian,
  and square
 to identity.

0 σi σ0 0
Use Dirac’s convention αi =   and β =  .
σi 0 0 −σ0
• Get to know some basic consequences of the Dirac equation:
– Orbital angular momentum L = x × p is not conserved.

~
Angularmomentum
 J = L + S is conserved. Spin Sa = ǫ αα.
4✐ abc b c
σa 0
Sa = ~2   under Dirac’s convention.
0 σa
– Zeeman coupling term − mq S · B with g = 2.
– Spin-orbit coupling: under potential V (x), the low energy effective Hamiltonian
include − 2mq2 c2 Ŝ · (E × p̂), where E = − ∂∂r V .
– Zitterbewegung: Dirac particle cannot be “at rest” in the classical sense. Even
with p = 0 and without external potential, it seems to be oscillating, x(t) =
2t 2
x(0) + ~
2mc
(α sin 2mc
~
+ ✐αβ cos 2mc
~
t
).
M.E. Peskin, D.V. Schroeder, An Introduction to Quantum Field Theory, Chapter 3.
J.J. Sakurai, Advanced Quantum Mechanics, Chapter 3.
P.A.M. Dirac, The Principle of Quantum Mechanics, Chapter XI.

II KLEIN-GORDON EQUATION
I. REMINDER ABOUT BASICS OF SPECIAL RELATIVITY
• Lorentz group: O(1, 3) group,

(ct, x, y, z) → (ct, x, y, z) · Λ4×4 , preserving (ct)2 − x2 − y 2 − z 2 .
• Use metric ηµν = diag(1, −1, −1, −1) where diag means diagonal matrix. Λ·η ·ΛT = η.
• Contra-variant 4-vectors: index µ = 0, 1, 2, 3,

4-position xµ = (ct, x, y, z);
4-momentum pµ = ( Ec , px , py , pz ), E is energy;
4-potential Aµ = ( φc , Ax , Ay , Az ), φ and A are electrostatic and vector potentials;
4-current density j µ = (cρ, j), ρ and j are density and current density.
• Covariant 4-vectors are obtained by contravariant ones as vν = v µ ηµν . Einstein’s

∂
convention of implicit summation over repeated indices is used. ∂µ ≡ ∂xµ
. Greek
indices take values of 0, 1, 2, 3, Roman indices take values of 1, 2, 3 (or x, y, z).
• Proper time element dτ is given by (dτ )2 = dxµ ηµν dxν .
• Lorentz invariant quantities:

for example pµ pµ = E 2 /c2 − p2 = m2 c2
• Classical mechanics and electromagnetism can be cast into Lorentz covariant form.
Maxwell’s equations: ǫµνρ ∂µ Fνρ = 0, ∂ µ Fµν = µ0 jν where µ0 is vacuum permeability.
Fµν = ∂µ Aν − ∂ν Aµ = −Fνµ is electromagnetic field tensor. F0a = Ea /c for a = 1, 2, 3
and E is electric field; Fab = −ǫabc Bc for a, b, c = 1, 2, 3 and B is magnetic field.
• A moving particle in electric field will experience an effective magnetic field, the mag-
netic moment of the particle will precess (Thomas precession).
II. KLEIN-GORDON EQUATION
d
• The plane wave states are ψ ∼ e✐(p·x−E·t)/~ , suggesting p → −✐~∇ and E → ✐~ dt .
p2
• For non-relativistic particle, dispersion relation is E = 2m
, suggesting non-relativistic
~2
d
Schrödinger equation ✐~ dt ψ = − 2m ∇2 ψ.

III DIRAC EQUATION
• The first attempt to reconcile with relativistic dispersion relation E 2 − p2 c2 = m2 c4 is

2
the Klein-Gordon equation ~2 (− c12 ddt2 + ∇2 )ψ = m2 c2 ψ, or (~2 ∂ µ ∂µ + m2 c2 )ψ = 0.
• The ψ in Klein-Gordon equation cannot be treated as probability amplitude.
ψ ∗ ψ d3 x is conserved, and ψ ∗ ψ is non-negative, so can

R
– In non-relativistic case,
be interpreted as probability density.
d d
– Under Klein-Gordon equation, (✐ψ ∗ dt ψ ∗ ) d3 x is conserved, however the
R
ψ − ✐ψ dt
integrand is not positive semidefinite and cannot be interpreted as probability
density.
III. DIRAC EQUATION
• The trouble with probability interpretation of Klein-Gordon equation is the appear-

ance of 2nd time derivative. Then it is tempting to consider the dispersion relation,
p
E = p2 c2 + m2 c4 and replace E and p as before. However quantum mechanics
should allow linear superposition of wave functions, the square root does not allow
that. Dirac formally solved the square root as linear function of p and obtain
d
• Dirac equation: ✐~ dt ψ = (cα · p̂ + mc2 β)ψ.
• To be consistent with E 2 = p2 c2 + m2 c4 , we need

{αi , αj } = 2δij , {αi , β} = 0, and β 2 = 1.
Then αi and β must be matrices, and ψ is a vector.
   
0 σi σ 0
• Dirac’s convention: αi =   and β =  0 . σ0,1,2,3 are Pauli matrices.
σi 0 0 −σ0
 
2
d
mc cσ · p̂
Dirac equation becomes ✐~ dt ψ=  · ψ.
cσ · p̂ −mc2
• ψ † · ψ can be interpreted as probability density.
• Lorentz covariant form of Dirac equation: (✐~γ µ ∂µ − mc2 )ψ̃ = 0,

where γ µ = (β, βα) and ψ̃ = βψ. Note that {γ µ , γ ν } = η µν .
Under Lorentz transformations (including spatial rotations) the vector wave function
ψ̃ shall transform non-trivially (not only changing argument t and x).

III DIRAC EQUATION
• Conserved quantities for Ĥ = cα · p̂ + mc2 β,
– momentum p̂ = −✐~∇.
– angular momentum J = L + S, L = x × p̂ is orbital angular momentum,

~
Sa = ǫ αα
4✐ abc b c
is “spin” angular momentum. L (or S) alone is not conserved.
Spin has eigenvalues ± ~2 (Dirac particles are spin-1/2).
[Ja , Jb ] = ✐~ǫabc Jc , [La , Lb ] = ✐~ǫabc Lc , [Sa , Sb ] = ✐~ǫabc Sc , [La , Sb ] = 0.
Spatial rotation corresponds to exp(−✐θn · J /~).
– helicity J · |p|
p
= S · |p|
p
, projection of (spin) angular momentum onto direction of
momentum. Helicity can be ± ~2 .
• Weyl equation: when mass m = 0, the Dirac equation can be block-diagonalized into
two 2 × 2 equations ✐~ ddt ψ = ±(cσ · p̂)ψ. ± signs correspond to right(left)-handed
Weyl fermions.
• Majorana fermions: it is possible to choose α to be real and β to be purely imaginary

(e.g. β = σ1 ⊗ σ2 , α1 = σ1 ⊗ σ3 , α2 = −σ3 ⊗ σ0 , α3 = −σ1 ⊗ σ1 ), in this case the Dirac
equation divided ✐ on both sides becomes real, so has real wave function solutions.
• Zitterbewegung:
consider the Heisenberg picture equation of motion, d
dt
x = ~✐ [Ĥ, x] = cα, so the veloc-
ity operator has eigenvalue ±c and different components of velocity do not commute!
Consider the case that p = 0, further use of Heisenberg equations of motion produces
d 2d 2 4 2t 2
( dt ) dt x = − 4m~2c d
dt
x. solution is x(t) = x(0) + ~
2mc
(α sin 2mc
~
+ ✐αβ cos 2mc
~
t
).
Dirac particle “at rest” seems to be oscillating very rapidly.
A. Solutions to Dirac equation

 
2
mc cσ · p
• let ψ(t, x) = e✐(p·x−E·t)/~ u(p). Then   · u(p) = E u(p)
cσ · p −mc2
Note that σ · p has eigenvalues ±|p|, denote the corresponding eigenvectors as χ± (p),
(σ · p)χ±
(p) = ±|p|χ± (p), if p = |p|(sin θ cos φ,
sin θ sin φ, cos θ), one can choose
cos(θ/2) −e−✐φ sin(θ/2)
χ+ (p) =  , χ− (p) =  , the two solutions to u(p) with
✐φ
e sin(θ/2) cos(θ/2)

III DIRAC EQUATION
q  q 
E+mc2 E+mc2
χ (p) χ− (p)
q 2mc2 +  q 2mc2
p
E= p2 c2 + m2 c4 are u+ (p) = and u− (p) =  ,
E−mc2 E−mc2
2mc2
χ+ (p) − 2mc2
χ− (p)
the subscript ± of u indicate helicity states.
• Negative energy solutions:  

2
−mc cσ · p
let ψ(t, x) = e−✐(p·x−E·t)/~ v(p). Then  For  · v(p) = E v(p).
2
cσ · p mc
q   q 
E−mc2 E−mc2
2 χ+ (p) − χ (p)
E = p2 c2 + m2 c4 , v+ (p) = q 2mc 2  and v− (p) =  q 2mc2 − .
p
E+mc E+mc 2
2mc2
χ+ (p) 2mc2
χ− (p)
Because of the way we define ψ here, this seems nagative energy −E solution of the
Dirac equation, with momentum −p. This ψ should be interpreted as the conjugate
of the “positron” wave function.
B. Non-relativistic limit of Dirac equation
• The Dirac equation of charge q particle in electromagnetic field is

d
✐~ dt ψ = [cα · (−✐~∇ − qA) + mc2 β + qV ]ψ.
V is electrostatic potential, A is vector potential, both depend on space and time.
• Consider the case that |p| ≪ mc. Separate the “diagonal” and “off-diagonal” parts of
Hamiltonian, Ĥ = Ĥ0 + Ĥ1 where Ĥ0 = mc2 β + qV and Ĥ1 = cα · (−✐~∇ − qA).
• Eliminate the off-diagonal parts Ĥ1 by sequences of unitary transformations,

d
ψ ′ = e✐Ŝ ψ, then ✐~ dt ψ ′ = Ĥ ′ ψ ′ where Ĥ ′ = e✐Ŝ Ĥe−✐Ŝ − ✐~e✐Ŝ ∂∂t e−✐Ŝ
= Ĥ + [✐Ŝ, Ĥ] + 12 [✐Ŝ, [✐Ŝ, Ĥ]] + 16 [✐Ŝ, [✐Ŝ, [✐Ŝ, Ĥ]]] + 1
24
[✐Ŝ, [✐Ŝ, [✐Ŝ, [✐Ŝ, Ĥ]]]] + ...
∂
−~ ∂t ∂
Ŝ − ~ 12 [✐Ŝ, ∂t ∂
Ŝ] − ~ 61 [✐Ŝ, [✐Ŝ, ∂t Ŝ]] + . . .
• Ĥ1 anti-commutes with β. Choose ✐Ŝ = βα · (p̂ − qA)/2mc = β Ĥ1 /2mc2 , so that
β
[✐Ŝ, mc2 β] = −Ĥ1 , note that [✐Ŝ, Ĥ1 ] = mc2
(Ĥ1 )2 , [✐Ŝ, [✐Ŝ, Ĥ1 ]] = − m21c4 (Ĥ1 )3 , . . . ;
β
Ĥ ′ = mc2 β + (qV + 2mc2
[Ĥ1 , qV ] − (2mc1 2 )2 [Ĥ1 , [Ĥ1 , qV ]] + . . . )
β
+( 2mc 2
2 (Ĥ1 ) −
1
3m2 c4
(Ĥ1 )3 + 8mβ2 c4 (Ĥ1 )3 + . . . )
β ∂ ✐~ ∂
+(✐~ 2mc 2 ∂t Ĥ1 − 8m2 c4 [Ĥ1 , ∂t Ĥ1 ] + ...)
1
• The diagonal terms up to m2
are (no need for further unitary transformation)
(Ĥ1)2 1 ✐~
β (mc2 + 2mc2
)+qV − 8m2 c4
[Ĥ1 , [Ĥ1 , q V ]] − 8m2 c4
[Ĥ1 , ∂∂t Ĥ1 ].

III DIRAC EQUATION
(Ĥ1 )2
• 2mc2
= 1
2m
αi (p̂i − qAi ) · αj (p̂j − qAj ) = 1
2m
(δij + ✐ 2Ŝ~k ǫijk )(p̂i − qAi )(p̂j − qAj )
1 1
= 2m
[(p̂ − qA)2 − 2Ŝk ǫijk · q ∂i Aj ] = 2m
[(p̂ − qA)2 − 2q Ŝ · B],
this gives the non-relativistic kinetic energy and the Zeeman term with Landé g-factor
g = 2.
• For the last two terms, [Ĥ1 , q V ] = −✐~qcα · (∇V ), [Ĥ1 , ∂∂t Ĥ1 ] = [Ĥ1 , −qcα · ∂
∂t
A],
✐~
[Ĥ1 , [Ĥ1 , q V ]] = [Ĥ1 , −✐~qcα · (∇V )], so − 8m12 c4 [Ĥ1 , [Ĥ1 , q V ]] − 8m2 c4
[Ĥ1 , ∂∂t Ĥ1 ]
= − 8m✐2 c4 [Ĥ1 , ✐~qcα · (−∇V − ∂
∂t
A)] = − 8m✐2 c4 [Ĥ1 , ✐~qcα · E]
✐~c q2
= − 8m 2 c4 [αi (p̂i − qAi ), αj Ej ]
= − 8m✐~q2 c2 ((δij + ✐ 2Ŝ~k ǫijk )(p̂i − qAi )Ej − (δij − ✐ 2Ŝ~k ǫijk )Ej (p̂i − qAi ))
2
~ q
= − 8m 2 c2 (∇ · E) +
q
ǫ Ŝ E (p̂ − qAi ) − 4m✐~q2 c2 (Ŝk ǫijk ∂i Ej )
2m2 c2 ijk k j i
2
~ q
= − 8m 2 c2 (∇ · E) −
q
2m2 c2
Ŝ · (E × P̂ ) − 4m✐~q2 c2 Ŝ · (∇ × E).
The 1st term is the Darwin term. The 2nd term is the spin-orbit coupling. The last
term vanishes for static electromagnetic field.
• For static central potential V (r) = V (r), E = −∇V = − rr ∂∂r V (r), the spin-orbit
coupling becomes − 2mq2 c2 Ŝ · (E × P̂ ) = q 1 dV
2m2 c2 r dr
Ŝ · (r × P̂ ) = q 1 dV
2m2 c2 r dr
Ŝ · L̂.

Lectures

Uploaded by

Copyright:

Available Formats

Lectures

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lectures

Uploaded by

Copyright:

Available Formats

Summary of Lecture #1: fundamental concepts

Goals and Requirements:

Hilbert space H linear space equipped with an inner product

• By the end of this lecture, you should feel comfortable about

• NOTE: statements with ⋆ are advanced topics/challenge questions/extra exercises

Advanced Quantum Mechanics, Fall 2018 1/15

I. THE HILBERT SPACE

• ψ(x) is a complex-valued function defined on some “coordinate” space V (x ∈ V ).

• Normalized ψ(x) has “dimension”(unit) of “volume”−1/2 (usually not dimensionless).

• We may need non-normalizable wavefunctions when the space V is not compact

B. The Hilbert Space

– Being a linear space: if ψ1 and ψ2 are elements of H (legitimate wavefunc.), then

Advanced Quantum Mechanics, Fall 2018 2/15

• There is a natural inner product(“overlap”) of two wavefunctions φ and ψ,

– Hermiticity: (φ, ψ) = (ψ, φ)∗ .

– Linearity: (φ, λ1 ψ1 + λ2 ψ2 ) = λ1 · (φ, ψ1 ) + λ2 · (φ, ψ2 ), for λ1 , λ2 ∈ ❈.

– Positive definiteness: (ψ, ψ) ≥ 0, and (ψ, ψ) = 0 if and only if “ψ = 0”.

– The Gram determinant det[(ψi , ψj )] ≥ 0.

– n = 2 case is the Cauchy-Schwarz inequality, (ψ1 , ψ1 )(ψ2 , ψ2 ) ≥ |(ψ1 , ψ2 )|2 .

C. Combining Hilbert Spaces: Direct Sum & Tensor Product

• Direct sum of two Hilbert spaces 

– Note: you need consistent definition of the “volumes” of V1,2 .

• Tensor product of two Hilbert spaces H1 (V1 ) ⊗ H2 (V2 ):

– Entanglement: wavefunctions in H1 ⊗ H2 may not be a direct product ψ1 ⊗ ψ2 .

– Identical particles: will be treated later.

Advanced Quantum Mechanics, Fall 2018 3/15

D. The Dirac Notation

• ‘kets’ |ψi: element(quantum state) in Hilbert space corresponding to wavefunction ψ.

– ‘bra’ is a linear functional: hψ|λ1 φ1 + λ2 φ2 i = λ1 hψ|φ1 i + λ2 hψ|φ2 i.

– ‘bras’ form an anti-linear space: λ∗1 hψ1 | + λ∗2 hψ2 | = hλ1 ψ1 + λ2 ψ2 |.

– Any ‘continuous’ linear functional f : H 7→ ❈, φ 7→ f (φ), corresponds to a

• Short-hand notation: |ψ1 ihψ2 | with ψ1 ∈ H1 , ψ2 ∈ H2 is a linear operator: H2 7→

Advanced Quantum Mechanics, Fall 2018 4/15

E. Complete Orthonormal Basis

• (Discrete) Orthonormal basis ei (i = 0, 1, . . . ) satisfy hei |ej i = δij .

– For finite dimensional Hilbert space, this is just

– We will see the resolution of identity in terms of overcomplete basis later.

• Example: Fourier series.

• Basis of composite Hilbert space:

– the (n + m) basis of H1 ⊕ H2 can be chosen as

– the (n × m) basis of H1 ⊗ H2 can be chosen as |ei i ⊗ |e′j i, namely

Advanced Quantum Mechanics, Fall 2018 5/15

II. QUANTUM MECHANICAL OPERATORS

A. Quantum Mechanical Operators

• Anti-linear operators: replace the last condition of linear operators by

– Example: the operator of “taking complex conjugate” K: φ(x) 7→ φ(x)∗ .

• Hermitian conjugate (adjoint) of linear operators: Ô† is a linear operator satisfying

– (λÔ)† = λ∗ Ô† , (ÂB̂)† = B̂ † Â† .

– Hermitian operators: those satisfy Ô† = Ô.

– Any operator is the sum of its Hermitian & anti-Hermitian part:

– Matrix represention under non-orthogonal/overcomplete basis can also be useful.

– Expectation value of Ô in state ψ: hψ|Ô|ψi/hψ|ψi.

Advanced Quantum Mechanics, Fall 2018 6/15

– Eigenvalues of Hermitian operators are real.

• (Not required) Singular value decomposition (SVD):

– In complete orthonormal basis |ei i, the above relations becomes

• Projection operators: operators P̂ : H 7→ H, satisfying P̂ P̂ = P̂ .

– ✶ − P̂ is also a projection operator. Exercise: check that (✶ − P̂ )(✶ − P̂ ) = (✶ − P̂ ).

• Inverse of an operator: Â−1 must satisfy Â−1 · Â = ✶ and Â · Â−1 = ✶