Contents 1

Concepts Fondamentaux de la Physique

Introduction to Second Quantization

Christophe Mora

Contents
1 Preliminaries 1
1.1 Single-particle Hilbert space . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Many-particle Hilbert space . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Basics of second quantization 2

2.1 Occupation number representation . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 Creation and annihilation operators . . . . . . . . . . . . . . . . . . . . . . . 3
2.3 Bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Representation of operators 4
3.1 Change of basis and the field operator . . . . . . . . . . . . . . . . . . . . . 4
3.2 Representation of one-body and two-body operators . . . . . . . . . . . . . . 5

Part of the complexity in the many-body problem - systems involving many particles -
comes from the indistinguishability of identical particles, fermions or bosons. Calculations
in first quantization thus involve the cumbersome (anti-)symmetrization of wavefunctions.
Second quantization is an efficient technical tool that describes many-body systems in a
compact and intuitive way.

1 Preliminaries
Before entering the details of second quantization, it is worth drawing a clear distinction
between the single-particle and the many-particle Hilbert spaces.

1.1 Single-particle Hilbert space

Consider a single particle described by the hamiltonian ĥ acting on the Hilbert space H1 .
H1 is generated by the complete set of eigenfunctions |λi (λ = k, σ, ν, . . .)

ĥ|λi = ελ |λi,

the eigenvalues ελ . The identity operator in H1 is given by the completeness relation

withP
1 = λ |λihλ|.
Examples:
1. single particle in free space, ĥ = −~2 ∇2 /(2m). The eigenfunctions are labeled by the
ik·r
wavevectors k with ψk (r) = hr|ki = e√V and the energies εk = ~2 k 2 /2m.

2. spin 1/2 in a magnetic field, ĥ = −BS z . The Hilbert space has dimension 2, generated
by the eigenstates | ↑i and | ↓i of the spin operator S z .
1.2 Many-particle Hilbert space 2

The basis of two-particle states, given by the set of (anti-)symmetrized functions, + for
bosons and − for fermions,
1
ψλ,ν (1, 2) = √ [ϕλ (1) ϕν (2) ± ϕλ (2) ϕν (1)] ,
2

is built out of the single-particle states ϕλ (1) = h1|λi. The corresponding Hilbert space1 is
denoted F2 .

1.2 Many-particle Hilbert space

We first discuss fermions. Following the two-particle case, the set of antisymmetrized Slater
determinants
1 X
ψλ1 ,...,λN (1, . . . , N ) = √ (−1)P ϕλ1 (P1 ) . . . ϕλN (PN ), (1)
N ! P ∈SN

where the summation runs over all permutations of {1, . . . , N }, forms the basis2 of the
Hilbert space FN . In the bosonic case, the basis is obtained from symmetrized states, i.e.
Eq. (1) where (−1)P is replaced by 1.
The hamiltonian may describe independent particles in which case
N
X
Ĥ = ĥ(i) ,
i=1

where each piece ĥ(i) acts only on the particle i.

Examples:

p2i /(2m).
P
1. for particles in free space, Ĥ = i

2. for an assembly of N distinguishable spins in a magnetic field, Ĥ = −B N z

P
i=1 Si . The
N
Hilbert space has dimension 2 and symmetrization is not required.
P (i)
Interactions between particles can be added, Ĥ = i ĥ + V̂ , where V̂ includes all
multi-particle interactions. For example, Coulomb interactions read
N X N
X e2 1
V̂Coulomb = . (2)
i=1 j>i
4πε 0 |ri − rj |

2 Basics of second quantization

So far, we have introduced and discussed the many-body problem in the language of first
quantization. Second quantization corresponds to a different labelling of the basis of states
Eq. (1) together with the introduction of creation and annihilation operators that connect
spaces with different numbers of particles.
1
Restricted to (anti-)symmetrized wavefunctions, F2 is a subset of the larger space H1 ⊗ H1 .
2
A different choice for the set of single-particle states |λi gives, using Eq. (1), a different many-particle
basis that nevertheless spans the same Hilbert space FN .
2.1 Occupation number representation 3

2.1 Occupation number representation

Since identical particles are indistinguishable, it is not possible, for a state of the form
Eq. (1), to ascribe a definite single-particle state to a given particle. Therefore, instead of
focusing on the wavefunction of each particle individually, one can reverse the perspective
and characterize the states of Eq. (1) by the set of single-particle states {λ1 , . . . , λN } that
are occupied by particles, all other single-particle states being empty.
In terms of notations, |{nλ }i represents |ψλ1 ,...,λN i3 with, for fermions, nλ = 1 for λ = λi ,
i = 1 . . . N , and nλ = 0 otherwise. The state can be written schematically as
λ1 λ2 λN
|{nλ }i = |0 . . . 1 . . . 0 . . . 1 . . . 0 . . . (. . .) 1 i, (3)
where it is explicitly specified on the right-hand-side which states are occupied and which
state are empty.
Bosonic states have similar expressions although the occupation numbers nλ can take
values larger than 1, for example
λ1 λ2 λN
|{nλ }i = |0 . . . 5 . . . 0 . . . 1 . . . 0 . . . (. . .) 7 i, (4)
for nλ1 = 5, nλ2 = 1, . . ., nλN = 7.

2.2 Creation and annihilation operators

P
The constraint on the number of particles, λ nλ = N , can be released by working in the
extended Hilbert space
+∞
M
F= FN ,
N =0
called the Fock space. Here, F1 = H1 is the single-particle Hilbert space, F0 contains a
unique vacuum state, often noted |0i, in which no particle is present.
In the Fock space, creation operators are introduced that raise the number of particles
in a given single-particle state by 1. For fermions, it reads
λ1 λ2 λN λ1 λ2 λN
c†λ1 |0 . . . 0 . . . 0 . . . 1 . . . 0 . . . (. . .) 1 i = |0 . . . 1 . . . 0 . . . 1 . . . 0 . . . (. . .) 1 i,
while particle creation in a single-particle state that is already occupied gives zero,
λ1 λ2 λN
c†λ2 |0 . . . 1 . . . 0 . . . 1 . . . 0 . . . (. . .) 1 i = 0.

The annihilation operator cλ , lowering the number by 1, is the hermitian conjugate of c†λ .
The full basis of the Fock space F is in fact generated by creation operators applied on the
vacuum state, namely |nλ1 = 1, . . . , nλN = 1i = c†λ1 . . . c†λN |0 >.
The antisymmetric properties of the basis states (Slater determinants) |nλ1 . . . nλN i are
ensured by the anticommutation relations
{cα , cβ } = cα cβ + cβ cα = 0, {cα , c†β } = δα,β . (5)

The product n̂λ = c†λ cλ gives the number of fermions occupying the state |λi,
c†λ cλ |{nα }i = nλ |{nα }i
where nλ = 0 or 1.
3
ψλ1 ,...,λN (1, . . . , N ) = h1, . . . , N |ψλ1 ,...,λN i.
2.3 Bosons 4

2.3 Bosons
There are only slight differences in the way second quantization works for fermions and for
bosons. In the case of bosons, the basis states are symmetrized functions and the number of
bosons in a given single-particle state is not restricted. These properties are ensured by the
commutation relations

[bα , bβ ] = bα bβ − bβ bα = 0, [bα , b†β ] = δα,β , (6)

with the (annihilation) creation operators (bα ) b†α . From Eq. (6), one can prove4 that
√
b†λ |nλ i = nλ + 1|nλ + 1i
√ (7)
bλ |nλ i = nλ |nλ − 1i

such that n̂λ = b†λ bλ is indeed the number operator, n̂λ |nλ i = nλ |nλ i.

3 Representation of operators
The complexity associated with wavefunction (anti)symmetrization has been reduced, in the
formalism of second quantization, to the surprisingly simple commutation relations, Eq. (5)
for fermions and Eq. (6) for bosons. Had the usual operators of the theory complicated
expressions in terms of creation/annihilation operators, this would not be very useful. How-
ever, as we shall see below, the hamiltonian as well as standard operators do have simple
expressions in second quantization.

3.1 Change of basis and the field operator

Starting with the expression |λi = c†λ |0i, one can insert the closure relation 1 =
P
λ |λihλ|
to derive the transformation law for the creation/annihilation operators
X X
c†α = hλ|αi c†λ , cα = hα|λi cλ , (8)
λ λ

from one basis to another. Hence, the change of basis only requires the calculation of matrix
elements hα|λi involving single-particle states.
By convention, the field operator Ψ(r) in a continuous problem is associated to the basis
of position states |ri, X
Ψ(r) = hr|λi cλ . (9)
λ

Using Eq. (5) and Eq. (6), one finds the commutation relation

{Ψ(r), Ψ(r0 )} = 0, {Ψ(r), Ψ† (r0 )} = δ(r − r0 ), fermions,

(10)
0 † 0 0
[Ψ(r), Ψ(r )] = 0, [Ψ(r), Ψ (r )] = δ(r − r ), bosons.

The total number of particles (fermions or bosons) is then given by

X † Z
N̂ = cλ cλ = dd r ρ̂(r) (11)
λ

4
The states |nλ i are chosen to be normalized to 1.
3.2 Representation of one-body and two-body operators 5

where the local density operator ρ̂(r) = Ψ† (r)Ψ(r) has been introduced.

Example: The transformation to the Fourier momentum representation reads

1 X ik·r
Ψ(r) = √ e ck , (12)
V k

wherePck destroys a particle with momentum k. The total number of particle is given by
N̂ = k c†k ck .

3.2 Representation of one-body and two-body operators

Single-particle or one-body operators have the form Ô(1) = N (1)
P
i=1 ô [i] in first quantization,
(1)
where ô [i] is a single-particle operator acting on the ith particle. In the language of second
quantization, they take the form
X
Ô(1) = hα|ô(1) |βi c†α cβ , (13)
α,β

d1 d2 ϕ∗α (1)h1|ô(1) |2iϕβ (2).

R
with the matrix elements hα|ô(1) |βi =

Examples:
p2i /(2m), describing independent particles, reads
P
1. The kinetic energy operator T̂ = i
in second quantization
X ~2 k 2 †
T̂ = c ck , (14)
k
2m k
i.e. it is diagonal in the momentum basis. An alternative expression involving the field
operator is

(~∇/i)2 ~2
Z Z
†
d
T̂ = d r Ψ (r) Ψ(r) = dd r ∇Ψ† (r) · ∇Ψ(r). (15)
2m 2m

2. Tight-binding models are simplified band models for electrons in solids where only
neighboring sites hybridize. A particularly simple example is given by the hamiltonian
X † 
Ĥ = −t ci cj + c†j ci , (16)
hi,ji

where c†i creates an electron on site i and hi, ji denotes neighboring sites. The product
c†i cj describes intuitively the hopping of an electron from site j to site i: one electron
is annihilated on site j while a novel electron appears on sitePi. The hamiltonian
Eq. (16) is diagonalized by going to the Fourier space ck = √1Ns i e−ik·ri ci (Ns is the
P †
number of sites of the lattice), with the result Ĥ = k εk ck ck . In one dimension,
εk = −2t cos(ka), a being the lattice spacing.
We now consider a two-body operator such as the Coulomb interaction of Eq. (2). In
first quantization, it has the form
1 X (2)
Ô(2) = ô [i, j], (17)
2 i6=j
References 6

where ô(2) [i, j] accounts for pair interactions. In second quantization, it reads5
1 X
Ô(2) = hα β|ô(2) |γδic†α c†β cδ cγ , (18)
2 α,β,γ,δ

with the matrix elements

Z
(2)
hα β|ô |γδi = d1 d2 ϕ∗α (1)ϕ∗β (2)ô(2) [1, 2]ϕγ (1)ϕδ (2). (19)

Example: Electron-electron Coulomb interaction is given in second quantization by

e2
Z
1X
V̂Coulomb = dr1 dr2 Ψ† (r1 )Ψ†σ2 (r2 )Ψσ2 (r2 )Ψσ1 (r1 ), (20)
2 σ ,σ 4πε0 |r1 − r2 | σ1
1 2

in terms of the field operator Ψσ (r). Here the spin σ of electrons has been included. After
going to the Fourier momentum representation of Eq. (12), one obtains the alternative
expression
1 X X
V̂Coulomb = v(q) c†k1 +q,σ1 c†k2 −q,σ2 ck2 ,σ2 ck1 ,σ1
2V σ ,σ q,k ,k
1 2 1 2

with the Fourier transform of the Coulomb pair potential v(q) = e2 /(ε0 q 2 ).

References
[1] H. Bruus and K. Flensberg, Many-Body Quantum Theory in Condensed Matter Physics
(Oxford University Press, Oxford, 2004).

[2] A. Altland and B. Simons, Condensed Matter Field Theory (Cambridge University Press,
Cambridge, 2010).

5
Note the ordering of indices which, in the product of annihilation operators, is reversed with respect to
the ordering in the matrix element.