Quantum many–particle systems (second

quantization)
(Advanced quantum mechanics)
October 1, 2011

References:
• F. Schwabl, Advanced Quantum Mechanics.

1 States and observables for a many–particle

system
1.1 State space
In the case of a system composed of many distinguishable particles (or, in
general, of many sub-systems), the Hilbert space of states is constructed as
follows. First, one chooses an orthonormal basis of single-particle states for a
particle (subsystem) α, composed of states |iα iα (we will use Latin letters for
states and Greek letters for particles). Then, the basis of the space of states
of the whole system of N particles is formed, by definition, by the vectors
|i1 i1 |i2 i2 . . . |iα iα . . . |iN iN ≡ |i1 i2 . . . iα . . . iN i,
where the second, short notation assumes that the particles (subsystems) are
ordered in a standard way. These states are called product states and the
basis is called the canonical basis or product state basis. The product states
are assumed to be mutually orthogonal,
hi01 i02 . . . i0α . . . i0N |i1 i2 . . . iα . . . iN i = δi1 i01 δi2 i02 . . . δiα i0α . . . δiN i0N .
These vectors are naturally interpreted as representing the states in which the
particle 1 is in the state |i1 i etc. By virtue of the superposition principle, the
allowed states of the system of N particles are all the superpositions of these
basis states1 . As we will see, in the case of identical particles most of these
vectors do not represent valid physical states. Therefore, it seems reasonable to
refer to them as configurations, reserving the term states for the physically
allowed ones. For identical particles we demand that all the single-particle
bases are the same.
In this part of the lecture we will study the consequences of the fact that
states and observables of a system of many identical (indistinguishable) parti-
cles should transform in a specific way if we exchange two or more particles.
1
Note that the dimension of the resulting Hilbert space is the product of the dimensions
of the single-particle spaces. More on compound quantum systems in the last part of the
lecture, if time permits.

1
Such a reordering of elements of a given finite set is called a permutation.
In order to move ahead with the quantum theory we first need to study some
basic properties of these operations2 .

1.2 Some facts about permutations

First, let us introduce some notation. A general, unspecified permutation will
be denoted by P . A possible representation of a permutation is
 
1 2 3 4 5 6
P= ,
4 5 1 6 2 3

which means that the element 4 takes the place of the element 1, the element
5 takes the place of the element 2 etc. A special case of a permutation is a
cycle, in which the elements are substituted “in a chain”. An example is P134
(also written as (134)), which denotes a permutation in which 1 is replaced by
3, 3 by 4 and 4 by 1,
 
1 2 3 4 5 ...
P= .
3 2 4 1 5 ...

It can be shown that each permutation can be represented by a sequence of

disjoint cycles (disjoint means that each element appears only in one cycle).
For instance,
 
1 2 3 4 5 6
P= = (163)(24)(5).
6 4 1 2 5 3

Here the right hand side denotes a sequence of cycle permutations that are
performed one after another from right to left (although for disjoint cycles the
order does not matter).
A very special case of a permutation, which is important for our lecture, is
a two-element cycle, that is, a permutation that just exchanges two elements,
say i and j. Such a permutation is called a transposition and will be denoted
by Pij or (ij). Note that Pij2 = E, where E is the identity (a “do nothing”
permutation). An important fact is that any permutation can be represented
as a sequence of transpositions (not necessarily disjoint). This can be done in
many different ways. However, it can be shown that in any such decomposition
of a given permutation the number of transpositions will be either always
odd or always even. This allows us to assign a parity to every permutation:
a permutation is said to be even (or to have a parity of +1) if it can be
decomposed into an even number of transpositions and to be odd (parity −1)
otherwise. The parity of a permutation P is denoted as (−1)P .
It is obvious that a composition (or product) of two permutations (that is,
performing two permutations P1 and P2 one after the other) is equivalent to
2
A complete discussion and proofs of the facts quoted here can be found, e.g., in
F.W. Byron, R.F. Fuller, Mathematics of Classical and Quantum Physics or, in Polish, in
R. Gonczarek, Teoria grup w fizyce, http://www.dbc.wroc.pl/Content/992/gonczarek.pdf.

2
performing some other single permutation P3 , which is written as P3 = P2 P1 .
Moreover, combining the permutations is associative, that is, performing the
product of P1 and P2 , and then P3 is the same as performing P1 , and then the
product of P2 and P3 ,
P3 (P2 P1 ) = (P3 P2 )P1 .
In addition, the set of all permutations contains a neutral element (identity) E
and for each permutation P one can find its inverse (that is, one that restores
the initial order of elements). The inverse to P is denoted by P −1 and is for-
mally defined by P −1 P = PP −1 = E. Taken together, these properties mean
that the set of permutations has the formal structure of a group. The group
of permutations (called also “symmetric group”) of N elements is denoted by
SN and the number of its elements (the order of the group) is N !.
An important theorem states that if {P1 , P2 , . . . , PN ! } are all the elements
of the group SN and each appears only once in this sequence than for any
permutation P ∈ SN , the sequence {PP1 , PP2 , . . . , PPN ! } also contains all
the elements of the group and each of them appears only once. The same is
true for the sequence {P1 P, P2 P, . . . , PN ! P}.

1.3 Permutation of particles

Consider an N -particle configuration |i1 i1 |i2 i2 . . . |iN iN . If we perform a per-
mutation on the particles described by this configuration we will obtain a
different configuration. For instance, for a cyclic permutation Pαγβ ,
Pαγβ
|i1 i1 . . . |iiα . . . |jiβ . . . |kiγ . . . |iN iN −→ |i1 i1 . . . |iiγ . . . |jiα . . . |kiβ . . . |iN iN .
Thus, permutation of particles corresponds to a transformation of configura-
tions, which defines an operator P defined on the set of configurations. We
will demand that this operator is linear, which allows us to extend it onto
arbitrary superpositions of configurations, that is, on the whole Hilbert space.
Thus, for our example we write
Pαγβ |i1 i1 . . . |iiα . . . |jiβ . . . |kiγ . . . |iN iN = |i1 i1 . . . |iiγ . . . |jiα . . . |kiβ . . . |iN iN
= |i1 i1 . . . |jiα . . . |kiβ . . . |iiγ . . . |iN iN ,
where in the final step the particles were ordered in a standard way. The
representation of an arbitrary permutation P is defined in an analogous way.
Note that P denotes an operation of permuting a certain set of elements, while
P is an operator in the Hilbert space of the quantum system. We say that the
operators P form a representation of the group SN 3
One defines two operators acting on the N -particle state vectors: the sym-
metrizer S+ and the anti-symmetrizer S− ,
1 X
S± |i1 . . . iN i = √ (±1)P P |i1 . . . iN i.
N ! P ∈SN

3
Some more mathematics is needed to make this statement absolutely strict. See the
texts on group theory recommended above.

3
From the theorem quoted at the end of Sec. 1.2 it easily follows that
P S+ = S+ P = S+ and P S− = S− P = (−1)P S− . (1)
In the proof of the second fact, we note that P Pi has the same parity as Pi if
P is even and opposite parity if P is odd. An obvious special case of this is
Pij S− = −S− .
We will now discuss a few properties of the permutation operator acting
on the quantum states of the N -particle system in the way defined above.
The permutation operator conserves the scalar product, that is, for any
states |Ψi, |Φi,
hΨ|Φi = hP Ψ|P Φi.
In view of the linearity of the permutation operator and of the scalar prod-
uct it is sufficient to verify this fact for two product configurations |Ψi =
|i1 . . . iα . . . iβ . . . iN i and |Φi = |i01 . . . i0α . . . i0β . . . i0N i. Moreover, as every per-
mutation can be composed of transpositions, one only has to show this equality
for transpositions. One has
Pαβ |Ψi = |i1 . . . iβ . . . iα . . . iN i, Pαβ |Φi = |i01 . . . i0β . . . i0α . . . i0N i,
hence
hP Ψ|P Φi = δi1 i01 . . . δiβ i0β . . . δiα i0α . . . δiN i0N = hΨ|Φi.
As a consequence, P is a unitary operator. Indeed, using the previous fact,
we have
hΦ|P Ψi = hP −1 Φ|P −1 P Ψi = hP −1 Φ|Ψi.
This means that P −1 = P † , that is, P is unitary.
If the particles are identical (indistinguishable) then, classically, the system
with particles i and j having the positions ri , rj and the momenta pi , pj must
have the same properties as that with the particle i having the position and
momentum rj , pj and the particle j having the position and momentum ri , pi .
Therefore, the physical quantities cannot change if we exchange two particles,
say, i and j. The same holds true for any other permutation of particles. We
expect that also in quantum mechanics any observable describing a measurable
property of a system of many identical particles must be invariant under any
permutation of particles. This can be stated formally by noting that if |Ψi
is an arbitrary state of the system of identical particles and |Φi = P |Ψi is
the state with particles reordered by a permutation P than the average value
of any observable A must be the same in the two states (because permuting
identical particles cannot have a physical effect). That is,
hΨ|A|Ψi = hΦ|A|Φi = hΨ|P † AP |Ψi.
As this must be true for any state |Ψi we can conclude that
A = P † AP for any P.
This can also be written in the form [P, A] = 0, which is a formal statement
of the fact that A must be symmetric with respect to any permutation.

4
 As a consequence of this fact, if |Ψi is an eigenstate of an observable A
then P |Ψi is an eigenstate as well, belonging to the same eigenvalue. Indeed, if
A|Ψi = a|Ψi then A(P |Ψi) = P A|Ψi = P a|Ψi = a(P |Ψi). If one allowed an
arbitrary vector |Ψi to be a valid representation of a physical state then P |Ψi
would, in general, be a different vector for each P ∈ SN and every eigenstate
of the observable A would be N ! times degenerate. This enormous degeneracy
(think of 1024 electrons in a piece of a metal) would be completely undetectable
because it affects every physical observable. It turns out, however, that this is
not the case.

1.4 Completely symmetric and antisymmetric states:

bosons and fermions
It is an experimentally established law of nature that out of the huge variety of
states transforming under permutations in a different way only two kinds repre-
sent actual physical states. The first kind are the completely symmetric states,
that is, such states |Ψi that P |Ψi = |Ψi for every permutation. The second
kind are the completely antisymmetric states, such that P |Ψi = (−1)P |Ψi.
The states of some particles, caled bosons, can only be symmetric, while
states of the others (fermions) can only be asymmetric.
Symmetric and antisymmetric states can be obtained from any product
configuration using the symmetrizing and antisymmetrizing operators,
|ΨS i = S+ |i1 . . . iα . . . iN i, |ΨA i = S− |i1 . . . iα . . . iN i.
Both these procedures produce a superposition of all the possible permuta-
tions of the original configuration. Note that if the product configuration
contains n particles in the same state (that is, a label, say k, appears n times
in |i1 . . . iα . . . iN i) than both the operators S± will produce each configuration
n! times (with the identical labels in all possible orderings). However, for the
antisymmetrization, half of these identical configurations will have the “+”
sign, while the other half will have the “-” sign, because half of the n! permu-
tations of these n identical labels are even and half are odd. Therefore, the
effect of S− on such a configuration will be zero. This is just the formal version
of the Pauli exclusion principle: there is no physical state of many fermions in
which two or more fermions occupy the same single-particle state.
It is now easy to see that if S− |i1 . . . iα . . . iN i is non-zero than it is nor-
malized. Indeed, since all the states iα are different, all the permutations
P |i1 . . . iα . . . iN i are mutually orthogonal. Sum of N ! orthogonal, √ normalized
states has the norm of N ! which is canceled by the pre-factor 1/ N ! in the
definition of S− . Note also that by antisymmetrizing another configuration in
which particles occupy the same states but in different order we would get the
same antisymmetric state up to a sign. In fact,
S− |i2 i1 . . . iN i = −S− |i1 i2 . . . iN i,
as the two original configurations differ by a single transposition. Therefore,
in order to fix the sign we have to fix the order of the single particle states

5
i1 , i2 , . . . and insist that the state labels i1 , i2,... appear in the configuration used
for antisymmetrization in the prescribed order. Once this sign ambiguity is
solved, a state is fully characterized by the number of particles in each single-
particle state i, denoted by ni . Therefore, the states of N fermions which are
obtained by antisymmerization of product configurtions can be denoted as
X
|n1 n2 . . . ni . . .i = S− |i1 . . . iα . . . iN i, ni = 0, 1, ni = N, (2)
i

where ni is the number of times i appears among the labels iα on the right-hand
side. This representation of a many-particle state is called the occupation
number representation or the second quantization representation.
For bosons, symmetrization yields ni ! copies of each different configuration
if the original configuration contained the state i ni times. This produces a
factor of ni ! in the norm. Therefore, in order to get a normalized completely
symmetric state one has to cancel these factors and the symmetrized state of
bosons in the occupation number representation is

1
|n1 n2 . . . ni . . .i = √ S+ |i1 . . . iα . . . iN i,
n1 !n2 ! . . . ni ! . . .
X
ni = 0, 1, . . . ni = N.
i

In both cases (fermions or bosons), the orthogonality relation is

hn01 n02 . . . n0i . . . |n1 n2 . . . ni . . .i = δn1 n01 δn2 n02 . . . δni n0i . . . , (3)

which follows from the definition of these states.

We have shown that completely symmetric and antisymmetric states can be
created by (anti)symmetrizing the product configurations forming the canon-
ical basis. It remains to be shown that this procedure yields a complete set
of states in the Hilbert space HN of an N -particle system, that is, any com-
pletely symmetric or antisymmetric state can be expressed as a combination
of these symmetrized basis states. We note that a completely (anti)symmetric
state X
|Ψi = ci1 ...iN |i1 i . . . |iN i
i1 ...iN

satisfies |Ψi = (±1)P |Ψi, hence

1 X 1 X √
S± |Ψi = √ (±1)P P |Ψi = √ (±1)P (±1)P |Ψi = N !|Ψi.
N! P N! P

On the other hand, by the linearity of the operators S± ,

X
S± |Ψi = ci1 ...iN S± |i1 i . . . |iN i.
i1 ...iN

6
Hence,
1 1 X
|Ψi = √ S± |Ψi = √ ci1 ...iN S± |i1 i . . . |iN i.
N! N ! i1 ...iN
This shows that, indeed, any completely (anti)symmetric state is a linear com-
bination of (anti)symmetrized product configurations, that is, these states form
a complete set in HN . In the occupation number representation, this com-
pleteness relation reads
X
|n1 n2 . . . ni . . .ihn1 n2 . . . ni . . . | = IN , (4)
n1 +n2 +...+ni +...=N

where IN is the unit operator in HN .

Finally, let us note that the state obtained by antisymmetrizing a product
configuration can be written in the form of a Slater determinant,
 
 |i1 i1 |i1 i2 |i1 iN 
1  
S− |i1 i2 . . . iN i = √  ... ..
.
..  .
. 
N ! 
|iN i1 |iN i2 . . . |iN iN 

2 States and operators in the Fock space

In the previous section, we have introduced the occupation number represen-
tation, in which the states are labeled by the numbers of particles occupying
each single-particle state [Eqs. (2) and (3)]. We have also shown that these
states span the whole Hilbert space HN of an N -particle system (that is, they
form a basis in this space). So far, our formalism is completely equivalent to
the standard approach in quantum mechanics and we have just introduced a
new representation of the states. Now we make an essential step ahead.
We consider a Hilbert space F (called a Fock space) containing all the
vectors representing a system of an arbitrary number of identical particles
and, in addition, the vector |0i representing the vacuum (zero particles). For-
mally, this space is easy to construct. It is simply spanned by all the states
|n1 . . . ni . . .i without any restriction on the total number of particles. Com-
paring with Eqs. (2) and (3), we simply relax the requirement that
X
ni = N (not any more).
i

Formally, the Fock space is a direct sum of the N -particle spaces for all N ,

F = H0 ⊕ H1 ⊕ . . . ⊕ HN ⊕ . . . .

We postulate that the states with different total number of particles are or-
thogonal. This means that Eq. (3) can be rewritten for the Fock space

hn01 n02 . . . n0i . . . |n1 n2 . . . ni . . .i = δn1 n01 δn2 n02 . . . δni n0i . . . (orthogonality).
(5)

7
By construction, the states |n1 n2 . . . ni . . .i span the Fock space, therefore the
relation generalizing Eq. (4) is
X
|n1 n2 . . . ni . . .ihn1 n2 . . . ni . . . | = I (completeness), (6)
n1 n2 ...ni ,...

where the restriction of the total number of particles is removed.

One might notice that all the “usual” operators, like position, momentum,
energy, etc. do not change the number of particles (that is, they operate only
within one subspace HN . Why do we need this new huge space F? The
reason is twofold. First, the number of particles is not conserved in nature.
For instance, an electron and a positron can annihilate, giving rise to a photon.
Even in the solid state physics, we like to describe processes in which a photon
is absorbed and an electron–hole pair is created (although in this case speaking
of electrons and holes as a pair of particles is only a formal trick). Second,
we will see that representing the “usual” operators in terms of operators that
change the number of particles leads to enormous simplification of the notation
and allows one to treat problems that would be otherwise intractable. Exactly
this second aspect is the subject of this section.

2.1 Creation and annihilation operators for bosons

By definition, a creation operator a†i increases the number of particles in
the state i by 1. Formally, the definition is
√
a†i |n1 n2 . . . ni . . .i = ni + 1|n1 n2 . . . ni + 1 . . .i. (7)

This means that any state can be constructed by acting on the vacuum state
with appropriate creation operators,
1  n1  n2
† †
 ni
|n1 n2 . . . ni . . .i = √ a1 a2 . . . a†i . . . |0i.
n1 !n2 ! . . .
The action of its adjoint (Hermitian conjugate), ai , is found by conjugating
the definition given in Eq. (7),
p
hn01 n02 . . . n0i . . . |ai = n0i + 1hn01 n02 . . . n0i + 1 . . . |,
and multiplying both sides from the right by |n1 n2 . . . ni . . .i,
p Y
hn01 n02 . . . n0i . . . |ai |n1 n2 . . . ni . . .i = n0i + 1δn0i +1,ni δn0j ,nj , (8)
j6=i

where we used the orthogonality relation (5). Next, one uses the completeness
relation, Eq. (6), to write
X
ai |n1 . . . ni . . .i = |n01 . . . n0i . . .ihn01 . . . n0i . . . |ai |n1 . . . ni . . .i
n01 ...n0i ...
X p Y
= |n01 . . . n0i . . .i n0i + 1δn0i +1,ni δn0j ,nj ,
n01 ...n0i ... j6=i

8
where we used Eq. (8). Hence, one finds for the action of the annihilation
operator,
 √
ni |n1 . . . ni − 1 . . .i, ni > 0,
ai |n1 . . . ni . . .i = (9)
0, ni = 0.

With the explicit definitions of the creation and annihilation operators

given by Eqs. (7) and (9), one can easily check that

aj ai |n1 , . . . , ni , . . .i = ai aj |n1 . . . ni . . .i

and
a†j a†i |n1 . . . ni . . .i = a†i a†j |n1 . . . ni . . .i
(note that the cases i = j and i 6= j must be considered separately). For a
subsequent action of a creation and annihilation operator one finds

a†j ai |n1 . . . ni . . .i = ai a†j |n1 . . . ni . . .i, for i 6= j.

However, for i = j we have

√
a†i ai |n1 . . . ni . . .i = ni a†i |n1 , . . . , ni − 1, . . .i = ni |n1 . . . ni . . .i (10)

and
√
ai a†i |n1 . . . ni . . .i = ni + 1ai |n1 , . . . , ni + 1, . . .i = (ni + 1)|n1 . . . ni . . .i

for an arbitrary state |n1 . . . ni . . .i.

The above considerations lead directly to the commutation relations for
bosonic creation and annihilation operators,
h i
[ai , aj ] = a†i , a†j = 0, [aj , a†j ] = δij . (bosons)

Moreover, Eq. (10) shows that the states |n1 . . . ni . . .i are eigenstates of the
operators a†i ai and the corresponding eigenvalue is the number of particles in
the state i. Therefore, n̂i = a†i ai is called the number operator for the state i.
The total particle number operator is therefore,
X X
N̂ = n̂i = a†i ai . (11)
i i

The states |n1 . . . ni . . .i are therefore particle number eigenstates, that is,
states with a well-defined number of particles in each state. They are com-
monly called number states or Fock states. Keep in mind that these states
form an orthonormal basis in the Fock space.

9
2.2 Creation and annihilation operators for fermions
We would like now to define the creation and annihilation operators for fermions
in the same way as we did for bosons. In analogy with bosons, we want to
define creation operators in such a way that

S− |i1 i2 . . . iN i = a†i1 a†i2 . . . a†iN |0i.

Note, however, that

S− |i2 i1 . . . iN i = −S− |i1 i2 . . . iN i,

as the two original configurations differ by a single transposition. Therefore,

we must admit that
a†i1 a†i2 . . . a†i3 = −a†i2 a†i1 . . . a†i3 .
That is, for fermions, the order of the creation operators matters: changing the
order produces a “−” sign. Moreover, as already discussed above, an attempt
to create two fermions in the same single-particle state cannot lead to a valid
many-particle state. This means that
 2
a†i = 0.

As already pointed out in Sec. 1.4, the basis states for the Hilbert space
HN of an N -fermion system must be
X
|n1 n2 . . . ni . . .i = S− |i1 i2 . . . iN i = a†i1 a†i2 . . . a†iN |0i, ni = N,
i

where the creation operators for different states are arranged according to the
fixed ordering of the states. Next, we combine the states for different N to
form the Fock space, as previously. Now, however, upon acting with a creation
operator a†i on an N -particle state we not necessarily get a correct (N + 1)-
particle basis state because the operators may not be ordered properly. Let
in < i < in+1 for a certain n. Then

a†i a†i1 . . . a†in a†in+1 . . . |0i = (−1) nj †

. . . a†in a†i a†in+1 . . . |0i,
P
j<i ai 1

where the phase factor originates from flipping the creation operators until
a†i gets onto its place. If the occupation number representation is introduced
according to Eq. (2) then the above identity can be written as

a†i |n1 , . . . , ni , . . .i = (1 − ni )(−1)

P
nj
j<i |n1 , . . . , ni + 1, . . .i.
(fermions)
(12)
Here, the factor (1 − ni ) assures that the result is 0 if there already is a particle
in the state i and the phase factor follows from the discussion above. It is
convenient to treat the above equation as the definition of the fermionic
creation operator and to derive the properties (including those deduced

10
earlier) from this definition. Note that, consistently with this definition, the
Fock state can be expressed in the form
 n1  n2  ni
† †
|n1 n2 . . . ni , . . .i = a1 a2 . . . a†i . . . |0i, ni = 0, 1.

The action of the annihilation operator on the Fock states is found in a

way similar to the bosonic case. By conjugating Eq. (12), changing ni → n0i
and multiplying from the right by |n1 , . . . , ni , . . .i one finds

hn01 . . . n0i , . . . |ai |n1 . . . ni . . .i = δni ,n0i +1 .

Next, using this fact and the completeness relation (6), we find
X
ai |n1 . . . ni . . .i = |n01 . . . n0i . . .ihn01 . . . n0i . . . |ai |n1 . . . ni . . .i
n01 ...n0i ,...
P
= ni (2 − ni )(−1) j<i nj |n1 . . . ni − 1 . . .i
P
= ni (−1) j<i nj |n1 . . . ni − 1 . . .i. (13)

For a subsequent action of two annihilation operators, one finds, assuming

i>j

aj ai |n1 , . . . , nj , . . . , ni , . . .i
P
= aj ni (−1) i0 <i ni0 |n1 . . . nj . . . ni − 1 . . .i
P P
= ni nj (−1) i0 <i ni0 (−1) j0 <j nj0 |n1 . . . nj − 1 . . . ni − 1 . . .i

but

ai aj |n1 . . . nj . . . ni . . .i
P
= ai nj (−1) j0 <j nj0 |n1 . . . nj − 1 . . . ni . . .i
P P
= ni nj (−1) i0 <i ni0 −1 (−1) j0 <j nj0 |n1 . . . nj − 1 . . . ni − 1 . . .i,

where the difference in the second phase factor results from the fact that the
particle from the state j, preceding the state i, has already been removed by
the first operator. It is clear that the two results have an opposite sign. Hence,
we conclude that ai aj = −aj ai . Note that this is also true for i = j as ai ai = 0
By conjugation, we get a similar relation for the creation operators. It can be
checked in the same way that ai a†j = −a†j ai for i 6= j. Finally, we have

ai a†i |n1 . . . ni . . .i = ai (1 − ni )(−1) i0 <i ni0 |n1 . . . ni + 1 . . .i

P

P P
= (1 + ni )(1 − ni )(−1) i0 <i ni0 (−1) i0 <i ni0 |n1 . . . ni . . .i
= (1 − n2i )|n1 . . . ni . . .i

and

a†i ai |n1 . . . ni . . .i = a†i ni (−1) i0 <i ni0 |n1 . . . , ni − 1 . . .i

P

P P
= n2i (−1) i0 <i ni0 (−1) i0 <i ni0 |n1 . . . ni . . .i. (14)

11
This means that ai a†i + a†i ai = 1. Altogether, we find that the fermionic
operators satisfy the anticommutation relations
n o n o
{ai , aj } = a†i , a†j = 0, ai , a†j = δij , (fermions)

where {A, B} = AB + BA is the anticommutator of the two operators.

Moreover, we see from Eq. (14) that n̂i = a†i ai is the number operator
for the state i, so that the total particle number operator is given again by
Eq. (11).

2.3 Operators in the second quantization

First, let us consider a system of noninteracting identical particles (fermions
or bosons). We will now assume (for a while) that the single particle states
|ii introduced at the very beginning of the lecture are energy eigenstates,
corresponding to the single-particle energy eigenvalues Ei 4 . In the absence of
interactions, the total energy of a number state is just the sum of single-particle
energies of all the particles. If there are ni particles in the state i then their
total energy is ni i . Hence, the total energy of the system in a number state is
X
E= i ni .
i

As ni are the eigenvalues of the number operator and the number states form
a basis in the Fock space, the energy operator (Hamiltonian) for a system of
non-interacting bosons must have the form
X X
H= i n̂i = i a†i ai . (non-interacting particles)
i i

Next, we are going to represent an arbitrary single- or two-particle operator

conserving the number of particles in terms of the creation and annihilation
operators. A single-particle operator describes a property of a single particle,
like the position, momentum, angular momentum, energy etc. If such an
operator for a particle α is denoted by Gα than the operator for N particles is
X
G= Gα .
α

Each of the operators Gα can be expressed in the basis of single-particle states

X
Gα = hi|G|ji|iiα hj|α .
ij

4
This specific choice is not important for the general formalism. However, choosing the
basis states to be the eigensattes of a single-particle Hamiltonian is very common in practical
applications of the formalism.

12
In the matrix element the index α has been suppressed because all the particles
are identical, hence all the operators Gα are identical. The single-particle
operator G can now be written in the form
X X
G= hi|G|ji |iiα hj|α .
ij α
P
Let us focus on the operator α |iiα hj|α . It is convenient to treat the fermion
and boson cases separately.
For bosons, one has
X 1 X
|iiα hj|α | . . . ni . . . nj . . .i = √ |iiα hj|α S+ |i1 . . . iN i
α
n 1 !n 2 ! . . . α
1 X
= √ S+ |iiα hj|α |i1 . . . iN i,
n1 !n2 ! . . . α
P
where we used the fact thatPS+ commutes with the symmetric operator α |iiα hj|α .
Now, consider the object α |iiα hj|α |i1 . . . iN i. Each time iα = j for a certain
α, it is replaced by i. After symmetrization, each such configuration yields the
vector q
. . . (ni + 1)! . . . (nj − 1)! . . . | . . . ni + 1 . . . nj − 1 . . .i.
The same result will appear for each occurrence of j, that is, nj times. Hence,
X
S+ |iiα hj|α |i1 . . . iN i
α
q
= nj . . . (ni + 1)! . . . (nj − 1)! . . . | . . . ni + 1 . . . , nj − 1 . . .i

and
X √ √
|iiα hj|α | . . . ni . . . nj . . .i = nj ni + 1| . . . ni + 1 . . . nj − 1 . . .i
α

= a†i aj | . . . ni . . . nj . . .i,

where the final equality follows from Eqs. (7) and (9).
For fermions, the derivation is quite similar (there are no additional nor-
malization factors but one has to be careful with the sign). We have
X X X
|iiα hj|α | . . . ni . . . nj . . .i = |iiα hj|α S− |i1 . . . iN i = S− |iiα hj|α |i1 . . . iN i.
α α α

As we deal with fermions, the state j can appear at most once among iα . If
it does appear it will be changed into i. A non-zero result will be obtained
from antisimmetrization only if i was not present initially. Thus, if ni = 0 and
nj = 1 we obtain the initial configuration with j replaced by i, which we write
in the form,
X
S− |iiα hj|α |i1 . . . iα . . . iN i = nj (1 − ni )S− |i1 . . . iN ij→i .
α

13
Note, however, that by changing j into i we get the state labels in the wrong
order. The correct order is restored by moving the label i to the beginning
of the list
Pand thanPback to the place, where it shouldP be. If i < j then this
requires k<j nk + k<i nk flips. P If i > j than we need k<j nk transpositions
to move to the beginning and k<i nk −1 to move to the end, where “-1” stands
for the label j which is counted in the sum but shouldn’t be as it has just been
moved from its place. As a result, one finds
X
|iiα hj|α | . . . ni . . . nj . . .i
α
P P
nj (1 − ni )(−1)Pk<j nk +Pk<i nk | . . . ni + 1 . . . nj − 1 . . .i,

i < j,
= n + n −1
nj (1 − ni )(−1) k<j k k<i k
| . . . nj − 1 . . . ni + 1 . . .i, i > j.

Using the definitions of a†i and aj it is easy to verify that exactly the same
result is produced by a†i aj | . . . ni . . . nj . . .i. It can be checked that the two
results are also identical for i = j. Thus, we get
X
|iiα hj|α | . . . ni . . . nj . . .i = a†i aj | . . . ni . . . nj . . .i,
α

which is the same equation as for bosons.

In this way, we have found the second quantization representation of a
single-particle operator,
X X
G= Gα = hi|G|jia†i aj .
α ij

Two-particle operators typically represent interactions between particles.

Written in the basis of single particle states they have the form
1X 1 XX
F = Fαβ = hij|F|kli|iiα |jiβ hk|α hl|β
2 α6=β 2 α6=β ijkl
1X X
= hij|f |kli |iiα |jiβ hk|α hl|β ,
2 ijkl α6=β

where fαβ is an operator acting on the states of the particles α and β (for
instance, a typical interaction energy depends only on the coordinates of the
two particles). Again, the form of this term cannot depend on which pair
of particles we choose. The condition that α 6= β corresponds to the fact
that a particle does not interact with itself. The factor 1/2 assures that each
pair is taken into account only once. In order to find the second quantization
representation of such an operator we note that
X X X
|iiα |jiβ hk|α hl|β = |iiα hk|α |jiβ hl|β − |iiα hk|α |jiα hl|α
α6=β α,β α
X X X
= |iiα hk|α |jiβ hl|β − |iiα hl|α δkj .
α β α

14
In the first term, we can use the representation in terms of creation and anni-
hilation operators found above,
X X
|iiα hk|α = a†i ak , |jiα hl|α = a†j al .
α α

In the second term we write

X
|iiαhl|α δkj = a†i al δkj = a†i δkj al
α

and use the (anti)commutation relations δkj = ak a†j ± a†j ak , where the upper
sign is for fermions and the lower one is for bosons. As a result we get
X
|iiα |jiβhk|α hl|β = a†i ak a†j al − a†i ak a†j al ± a†i a†j ak al = a†i a†j al ak ,
α6=β

where we changed the order of operators using the commutation or anticom-

mutation relation. In this way, the obtained result is the same for fermions
and for bosons. Hence, the two-particle operators are represented as
1X
F = hij|F|klia†i a†j al ak .
2 ijkl

3 Field operators and field equations

So far we have defined the creation and annihilation operators related to a
certain basis of single-particle eigenstates which is typically chosen as the en-
ergy eigenbasis. However, there are other choices, the most important of which
are the basis of position eigenstates and of momentum eigenstates. These two
special cases will be discussed in this section and in the next one.
In order to perform a change of basis for the creation and annihilation
operators from a basis P {|ii} to another basis {|λi} one notes that |ii = a†i |0i,
†
|λi = aλ |0i, and |λi = i hi|λi|ii, hence
X X
a†λ = hi|λia†i and aλ = hλ|iiai ,
i i

where the second equality has been obtained by conjugation.

Let us now discuss the special case when |ii are the single-particle eigen-
states used so far and |λi are position eigenstates. Recall that the amplitude
hr|ii is, by definition, the wave function ψi (r) corresponding to the state |ii.

3.1 Field operators

Based on the above preliminary remarks, we define the field operators as the
creation and annihilation operators corresponding to position eigenstates,
X X
Ψ† (r) = ψi∗ (r)a†i , Ψ(r) = ψi (r)ai . (15)
i i

15
Their physical interpretation is that they create or annihilate a particle at a
point r.
Since Ψ(r) is a linear combination of annihilation operators which commute
or anticommute (for bosons and fermions, respectively) they obviously also
commute or anticommute. For two operators of different kind, one finds
X h i X
† 0 ∗ 0 †
ψi (r)ψj∗ (r 0 )δij
 
Ψ(r), Ψ (r ) ± = ψi (r)ψj (r ) ai , aj =
±
ij ij
X
= ψi (r)ψi∗ (r 0 ),
i

where [, ]± is another notation for the anticommutator and commutator (that

is, []+ ≡ {, } and []− ≡ [, ]) and the two cases refer to fermions and bosons,
respectively. Using P the completeness relation for the complete set of wave
∗ 0 0
functions we get i ψi (r)ψi (r ) = δ(r − r ). Hence, the commutation
relations for the field operators are

[Ψ(r), Ψ(r 0 )]± = 0,

 †
Ψ (r), Ψ† (r 0 ) ± = 0, Ψ(r), Ψ† (r 0 ) ± = δ(r − r 0 ).
  

Single-particle and two-particle operators are expressed by the field opera-

tors in an easy way. For a single-particle operator G one has
X
G= hi|G|jia†i aj . (16)
ij

Expressing the matrix element in the position representation,

Z
†
hi|G|jiai aj = d3 rψi∗ (r)Gψj (r)a†i aj .

Using the definition (15) one finds

Z
G = d3 rΨ† (r)GΨ(r).

In particular, for the kinetic energy one has

~2 ~2
Z Z
†
T =− 3
d rΨ (r)∆Ψ(r) = d3 r∇Ψ† (r)∇Ψ(r). (17)
2m 2m
and the single-particle potential energy U has the form
Z
U = d3 rΨ† (r)U (r)Ψ(r) (18)

A two-particle operator F is expressed as

1X XZ
F = hij|F|kli d3 rd3 r 0 ψi∗ (r)ψj∗ (r 0 )Fψl (r 0 )ψk (r)a†i a†j al ak
2 ijkl ijkl
Z
1
= d3 rd3 r 0 Ψ† (r)Ψ† (r 0 )FΨ(r 0 )Ψ(r)
2

16
An important example is the two-particle interaction energy
Z
1
V = d3 rd3 r 0 Ψ† (r)Ψ† (r 0 )V (r, r 0 )Ψ(r 0 )Ψ(r). (19)
2

An important operator, which has a convenient representation in terms

of the field operators, is the particle density at a point r0 . In the position
representation, X
ρ(r0 ) = δ(rα − r0 ).
α

The above operator has a structure of a single particle operator. According to

Eq. (16) we can write
Z
ρ(r0 ) = d3 rΨ† (r)δ(r − r0 )Ψ(r) = Ψ† (r0 )Ψ(r0 ). (20)

The operator of the total number of particles therefore is

Z Z
N̂ = d rρ(r) = d3 rΨ† (r)Ψ(r).
3

3.2 Field equations

By field equations one understands the equations of motion for the field opera-
tors in the Heisenberg picture, generated by the Hamiltonian H = T + U + V .
Let us recall that an operator A in the Heisenberg picture is defined as

A(t) = eiHt/~ Ae−iHt/~ (21)

and satisfies the equation of motion

i~Ȧ(t) = −[H, A(t)] = −[H(t), A(t)],

where we used the fact that

H(t) = eiHt/~ He−iHt/~ = H.

In order to derive the equations of motion we need to be able to calculate the

commutators and anticommutators of operators in the Heisenberg picture at
an arbitrary time t. However, if [A, B]± = C then

[A(t), B(t)]± = eiHt/~ Ae−iHt/~ eiHt/~ Be−iHt/~ ± eiHt/~ Be−iHt/~ eiHt/~ Ae−iHt/~
= eiHt/~ ABe−iHt/~ ± eiHt/~ BAe−iHt/~ = eiHt/~ [A, B]± e−iHt/~
= eiHt/~ Ce−iHt/~ = C(t).

Thus, the evolution of operators in the Heisenberg picture conserves the com-
mutation and anticommutation relations. We will also make use of the relation

[AB, C] = [A, C]B + A[B, C].

17
 Let us find the evolution equation for the fermionic field operator Ψ(r)
(the calculation for bosonic operators is very similar). We start the calculation
of the commutator appearing in the Eq. (21) with the single-particle energy
contribution. We will not write the time dependence explicitly. One finds
Z 
3 0 † 0 0 0
−[U, Ψ(r)] = − d r Ψ (r )U (r )Ψ(r ), Ψ(r)
Z
= − d3 r 0 U (r 0 ) Ψ† (r 0 )Ψ(r 0 )Ψ(r) − Ψ(r)Ψ† (r 0 )Ψ(r 0 )
 

Z
= − d3 r 0 U (r 0 ) −Ψ† (r 0 )Ψ(r)Ψ(r 0 )


− δ(r − r 0 ) − Ψ† (r 0 )Ψ(r) Ψ(r 0 )

 

= U (r)Ψ(r).

The term involving the kinetic energy is

~2
Z
d3 r 0 Ψ† (r 0 )∆0 Ψ(r 0 )Ψ(r) − Ψ(r)Ψ† (r 0 )∆0 Ψ(r 0 )

−[T, Ψ(r)] =
2m
~2
Z
d3 r 0 −Ψ† (r 0 )Ψ(r)∆0 Ψ(r 0 )

=
2m
− δ(r − r 0 ) − Ψ† (r 0 )Ψ(r) ∆0 Ψ(r 0 )
 

~2
= − ∆Ψ(r),
2m
where ∆0 denotes the Laplasian with respect to the coordinate r 0 and we used
the fact that {∆Ψ(r 0 ), Ψ(r)} = 0 since Ψ(r 0 ) anticommutes with Ψ(r) for any
r0 .
Finally, for the interaction term, we have
Z Z
1 3 0
d3 r 00 V (r 0 , r 00 ) Ψ† (r 0 )Ψ† (r 00 )Ψ(r)

−[V, Ψ(r)] = − dr
2
−Ψ(r)Ψ† (r 0 )Ψ† (r 00 ) Ψ(r 00 )Ψ(r 0 )

Z Z
1 3 0
d3 r 00 V (r 0 , r 00 ) Ψ† (r 0 ) δ(r − r 00 ) − Ψ(r)Ψ† (r 00 )
  
= − dr
2
− δ(r − r 0 ) − Ψ† (r 0 )Ψ(r) Ψ† (r 00 ) Ψ(r 00 )Ψ(r 0 )
 
Z
1
= − d3 r 0 Ψ† (r 0 )V (r 0 , r)Ψ(r)Ψ(r 0 )
2
Z
1
− d3 r 00 Ψ† (r 00 )V (r, r 00 )Ψ(r 00 )Ψ(r)
2
Z
= d3 r 0 Ψ† (r 0 )V (r 0 , r)Ψ(r 0 )Ψ(r),

where we use the fact that V (r, r 0 ) = V (r 0 , r) and changed the order of anni-
hilation operators in the last step.

18
 Hence, the field equation, that is, the evolution equation for the operator
Ψ(r) is
~2
i~Ψ̇(r, t) = − ∆Ψ(r, t) + U (r)Ψ(r, t)
2m
Z
+ d3 r 0 Ψ† (r 0 , t)V (r 0 , r)Ψ(r, t)Ψ(r 0 , t).

The evolution equation for Ψ† (r) is obtained by conjugation. The equation

for the bosonic field operator has exactly the same form. Note the similarity
to the Schrödinger equation, except for the last term.
Using the above equations one can find the evolution equation for the
density in the Heisenberg picture.
d
ρ(r, t) = Ψ† (r, t)Ψ̇(r, t) + Ψ̇† (r, t)Ψ(r, t)
dt
~2  †
 
1
Ψ (r, t)∆Ψ(r, t) − ∆Ψ† (r, t) Ψ(r, t)


= −
i~ 2m
 
~  † †


= −∇ · Ψ (r, t)∇Ψ(r, t) − ∇Ψ (r, t) Ψ(r, t) .
2im

As the density should satisfy the continuity equation ρ̇(r, t) = −∇ · j(r, t),
where j is the current density, we identify the operator
~  †
Ψ (r, t)∇Ψ(r, t) − ∇Ψ† (r, t) Ψ(r, t)


j(r, t) =
2im
as the current operator. In fact, it can be shown directly that the current
operator in the representation of field operators has exactly this form.

4 Momentum representation
The momentum representation is just the second quantization representation
in terms of the creation and annihilation operators, where the single particle
basis is chosen as the basis of free-particle states, labeled by the wave number
vector k and represented in the position representation by the plane waves,
1
ψk (r) = √ eik·r ,
V
where we assumed that the system volume is V and that the periodic boundary
conditions are imposed. According to Eq. (15), the field operators may be
represented as
1 X −ik·r † 1 X ik·r
Ψ† (r) = √ e ak , Ψ(r) = √ e ak . (22)
V k V k

The Hamiltonian in the momentum representation can be found using the

expressions in terms of the field operators. Substituting Eqs. (22) into Eq. (17),

19
performing the derivation and using the orthonormality condition
Z
1 0
d3 rei(k−k )·r = δk,k0
V
one finds easily for the kinetic energy
X ~2 k 2 †
T = ak ak .
k
2m

From Eq. (18), one finds for the potential energy

1X
U= Uk0 −k a†k0 ak ,
V k,k0

where Z
Uq = d3 re−iq·r U (r).

For the two-particle interaction which depends only on the relative position of
the two particles, r − r 0 (as is usually the case), we have
Z Z
1 X 0 0 0 0
V = 2
d 3
r d3 0
r e−iq·r e−iq ·r V (r − r 0 )eik ·r eik·r a†q a†q0 ak0 ak
2V kk0 qq 0
Z Z
1 X
i(k+k0 −q−q 0 )·r 0 0 00
= 2
dr3
e d3 r00 V (r 00 )e−i(k −q )·r a†q a†q0 ak0 ak
2V kk0 qq 0
1 X 1 X
= δk−q,q0 −k Vk0 −q0 a†q a†q0 ak0 ak = Vq−k a†q a†k+k0 −q ak0 ak
2V kk0 qq0 2V kk0 q
1 X
= Vp a†k+p a†k0 −p ak0 ak .
2V pkk0

In this derivation, we first changed the variables according to r 00 = r − r 0 and

then performed the integration over r and r 00 using the orthogonality relation
and the definition of the Fourier component of the interaction,
Z
Vq = d3 re−iq·r V (r).

Finally, we substituted q = k + p. As a result, the Hamiltonian for a system

of interacting identical particles in the momentum representation is
X ~2 † 1X 1 X
H= ak ak + Uk0 −k a†k0 ak + Vp a†k+p a†k0 −p ak0 ak . (23)
k
2m V k,k 0
2V pkk 0

Another useful relation is the expression for the Fourier transform of the
particle density,
Z X
ρ(q) = d3 rρ(r)e−iq·r = a†p ap+q ,
p

which is easy to get by substituting the field operators from Eq. (22) into
Eq. (20).

20
5 Spin
In the formalism of creation and annihilation operators, spin is just another
quantum number that has to be included to label the state. For instance,
in the momentum representation, the creation and annihilation operators are
akσ , a†kσ , with the commutation relations
h i
[akσ , ak σ ]± = akσ , ak0 σ0 = 0, [akσ , a†k0 σ0 ]± = δkk0 δσσ0 .
0 0
† †
±

The Hamiltonian of a system of interacting particles in the case of a spin-

independent single-particle potential and a spin-independent two-particle in-
teraction has the form
X ~2 † 1 X 1 X
H= akσ akσ + Uk0 −k a†k0 σ akσ + Vq a†k+q,σ a†k0 −q,σ0 ak0 σ0 akσ .
kσ
2m V kσ,k0
2V pkk0 σσ 0
(24)
The field operators now create or annihilate a particle at a given point and
with a given spin. Usually, they are written as Ψσ (r), Ψ†σ (r). Their relation
to the creation and annihilation operators is
X
Ψσ (r) = ψi (r, σ)ai ,
i

where ψi (r, σ) is the σ-component of the wave function (spinor) representing

the state i. The commutation relations for the field operators are
h i h i
0 † † 0 †
[Ψσ (r), Ψσ (r )]± = Ψσ (r), Ψσ0 (r ) = 0,
0 Ψσ (r), Ψσ0 (r ) = δσσ0 δ(r−r 0 ).
0
± ±

The Hamiltonian of the system of identical particles with spin in a spin-

independent one-particle potential and with spin-independent interaction in
the representation of field operators is
XZ  2 
3 ~ † †
H = dr ∇Ψσ (r)∇Ψσ (r) + Ψσ (r)U (r)Ψσ (r)
σ
2m
Z Z
1X
+ d r d3 r0 Ψ†σ (r)Ψ†σ0 (r 0 )V (r, r 0 )Ψσ0 (r 0 )Ψσ (r).
3
2 σσ0

6 Example: Ground state of the electron gas

(an elementary theory)
As an example we will calculate the lowest order Coulomb (exchange) correc-
tion to the energy of the electron gas at zero temperature in the presence of a
positive background that guarantees charge neutrality of the system. We con-
sider N electrons in a volume V. At the end, we will take the thermodynamical
limit, V → ∞, keeping the electron density n = N/V constant.

21
 From the introductory physics course or from the solid state course one
(hopefully) remembers that in the ground state, noninteracting fermions fill
the states up to the value of momentum
~kF = ~(3π 2 n)1/3 , (25)
called the Fermi momentum, where n is the particle density. This value of
momentum corresponds to the Fermi energy
~2 kF2 ~2 (3π 2 n)2/3
F = = .
2m 2m
The average kinetic energy per particle, k = Ek /N must be proportional to
the Fermi energy (as confirmed by the calculation presented below). Note
that the average distance r between particles in a system with a density n is
proportional to 1/n1/3 , hence
1
 ∼ F ∼ .
r2
On the other hand, the Coulomb energy is inversely proportional to the inter-
particle distance,
1
C ∼ .
r
Hence, if the density is high (r is small) the kinetic energy dominates and the
Coulomb energy can be treated as a correction. This is the limit we are going
to study in this example.
The Hamiltonian of the system is given by Eq. (23) without the single
particle potential U and with
e2 1 0
V (r − r 0 ) = 0
e−µ|r−r | ,
4πε0 |r − r |
where we introduce an exponential cut-off factor in order to deal with various
divergences that will appear in the calculation. At the end, after taking the
thermodynamic limit V → ∞, we will put µ = 0. Using the relation
Z
1 4π
d3 re−iq·r e−µr = 2
r q + µ2
one finds the Fourier transform of the Coulomb potential with the cut-off
e2 1
Vq = .
ε0 q + µ 2
2

Clearly, the q = 0 term diverges in the limit µ → 0. This term corresponds to

the Coulomb energy of the electrons. One expects that in a globally neutral
system this term will be exactly canceled by the Coulomb energy of the posi-
tive background and by the interaction energy between the electrons and the
background. Indeed, for the background energy one has
e2 1 n2 e2 1 N 2
Z Z
−µ|r−r 0 |
EB = d r d3 r 0
3
e =
4πε0 2 |r − r 0 | 2ε0 µ2 V

22
where we used the fact that the density of the background charge must be
equal to that of the electrons. The calculation is done by first shifting the
variables according to r 00 = r − r 0 and then performing the trivial integrals in
the spherical coordinates.
The interaction between the N electrons at the points ri and the uniform
background is given by

e2 X e2 1 N 2
Z
n
Ee−B = − d3 r e−µ|r−ri | = − ,
4πε0 i |r − ri | ε0 µ 2 V

where we again shifted the integration variable to obtain a trivial integral.

Finally, the q = 0 term of the Coulomb interaction is

e2 1 X † †
Vq=0 = a a 0 0 ak0 σ0 akσ .
2Vε0 µ2 kk0 σσ0 kσ k σ

Using the anticommutation relations and the definition of the number operator
we get
a†kσ a†k0 σ0 ak0 σ0 akσ = n̂kσ (n̂k0 σ0 − δkk0 δσσ0 ) .
We can then perform the sums using the definition of the total particle number
operator to get
e2 1
VC,q=0 = (N̂ 2 − N̂ ).
2Vε0 µ2
Since we work with states with a fixed total number of particles we can replace
the total particle number operator N̂ by its eigenvalue N . Thus, the q = 0
term reduces to a number, corresponding to the (direct) Coulomb energy of
the electron gas,
e2 1
Ee = (N 2 − N ).
2V ε0 µ2
Note that the second term yields the contribution to the total energy propor-
tional to N/V = n. The corresponding energy per particle is proportional to
n/N = 1/V and vanishes in the limit V → ∞, n = const.
As we can see, neglecting the vanishing term in the last expression the
three Coulomb energies cancel, as expected.
Hence, our task is to calculate the approximate ground state energy of the
system described by the Hamiltonian

H = H0 + V,

where
X ~2 q 2
H0 = a†q aq
q
2m
is the Hamiltonian of non-interacting fermions and

1 e2 X 0 1 †
V = a 0 a† ak 0 ak
2V ε0 qkk0 q 2 k +q k−q

23
is the Coulomb energy which will be treated as a perturbation. Here the prime
means that the q = 0 term should be ommitted in the summation. We will
calculate the leading order correction in the first order of the perturbation
theory.
The unperturbed ground state of the Hamiltonian H0 is formed by creating
the electrons in the states with k ≤ kF ,
Y †
|ψ0 i = ak,σ |0i.
k≤kF ,σ

The unperturbed eigenvalue is just the kinetic energy of all these particles,
X ~2 k 2 Z kF
(0) ~2 V 4 3~2 kF2 3
E = = 2 3
4π k = N = N F ,
k≤k σ
2m 2m (2π) 0 10m 5
F

where we used Eq. (25) to write V = 3π 2 N/kF3 and changed the summation
into integration according to the usual rule
V
X Z
... → 3
d3 k . . . .
k
(2π)
One defines the radius r0 such that a sphere of radius r0 has the volume
equal to the volume per particle in the system,
1 4
= πr03 ,
n 3
and the dimensionless quantity rs = r0 /a0 , where
4πε0 ~2
a0 =
me2
is the Bohr radius. Expressed in terms of the new parameters, the kinetic
energy is
 2/3
(0) e2 1 3 9π e2 2.21
E = N ≈ N.
8πε0 a0 rs2 5 4 8πε0 a0 rs2
The first order correction resulting from the Coulomb term is
e2 X 0 1
E (1) = 2
hψ0 |a†k+q,σ a†k0 −q,σ0 ak0 ,σ0 ak,σ |ψ0 i.
2ε0 V kk0 qσσ0 q

a non-zero term appears in the sum only if two electrons in the same state are
annihilated and created. As the term q = 0 is excluded, the only such term is
δσσ0 δk0 ,k+q a†k+q,σ a†k,σ ak+q,σ ak,σ = −δσσ0 δk0 ,k+q n̂k+q,σ n̂k,σ .
Since the state |ψ0 i is an eigenstate we can replace the number operators by
their values and the energy correction becomes
e2 X0 1
E (1) = − nk+q,σ nk,σ
2ε0 V kqσ q 2
e2 V2
Z Z
1
= − 2 6
d k d3 q 2 Θ(kF − |q + k|)Θ(kF − k).
3
2ε0 V (2π) q

24
Upon performing the integration and expressing the result by a0 and rs one
gets
 1/3
(1) e2 9π 3N e2 0.916
E =− ≈− N.
8πa0 ε0 rs 4 2π 8πa0 ε0 rs
Hence the leading terms in the energy of the system are

e2
 
2.21 0.916
E= N − .
8πa0 ε0 rs2 rs ...

This expansion is valid for rs  1. The omitted terms are called correlation
corrections and physically correspond to the fact that the probability of finding
an electron very close to another one is low because of Coulomb repulsion (that
is, the electron positions are correlated). Calculation of these terms requires
more advanced techniques of the many-body theory.

Acknowledgment
I am grateful to Dominik Wójt for pointing out many mistakes in the previous
version of this text and the other parts of these lecture notes.

25