Lagrangian quantum field theory

in momentum picture
I. Free scalar fields
Bozhidar Z. Iliev

Short title: QFT in momentum picture: I

arXiv:hep-th/0402006v1 1 Feb 2004

Basic ideas :
Began :
Ended :
Initial typeset :
Last update :
Produced :

June, 2001
June 2, 2001
July 31, 2001
June 8, 2001August 4, 2001
February 1, 2004
February 1, 2008

http://www.arXiv.org e-Print archive No. : hep-th/0402006

r TM

BO/ HO
Subject Classes:
Quantum field theory

2000 MSC numbers:

81Q99, 81T99

2003 PACS numbers:

03.70.+k, 11.10.Ef, 11.10.-z,
11.90.+t, 12.90.+b

Key-Words:
Quantum field theory, Pictures of motion
Pictures of motion in quantum field theory, Momentum picture
Free neutral (uncharged, Hermitian, real) scalar field
Free charged (non-Hermitian, complex) scalar field,
Equations of motion for free scalar field
Klein-Gordon equation, Klein-Gordon equation in momentum picture
Commutation relations for free scalar field, State vectors of free scalar field

Laboratory of Mathematical Modeling in Physics, Institute for Nuclear Research and Nuclear
Energy, Bulgarian Academy of Sciences, Boul. Tzarigradsko chaussee 72, 1784 Sofia, Bulgaria

E-mail address: bozho@inrne.bas.bg

URL: http://theo.inrne.bas.bg/bozho/

Contents
1 Introduction

2 The momentum picture

Free neutral scalar field

3 Description of free neutral scalar field in momentum picture

4 Analysis of the Klein-Gordon equation

5 Frequency decompositions and their physical meaning

11

6 The field equations in terms of creation and annihilation operators

12

7 The commutation relations

15

8 The vacuum and normal ordering

19

9 State vectors and transitions between them

21

Free arbitrary scalar field

24

10 Description of free scalar field

24

11 Analysis of the field equations

30

12 Frequency decompositions and their physical meaning

31

13 The field equations in terms of creation and annihilation operators

33

14 The charge and orbital angular momentum operators

36

15 The commutation relations

40

16 The vacuum and normal ordering

46

17 State vectors

49

18 Conclusion

51

References
This article ends at page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51
53

Abstract
The work contains a detailed investigation of free neutral (Hermitian) or charged (non-Hermitian) scalar fields and the describing them (system of) Klein-Gordon equation(s) in momentum picture of motion. A form of the field equation(s) in terms of creation and annihilation
operators is derived. An analysis of the (anti-)commutation relations on its base is presented.
The concept of the vacuum and the evolution of state vectors are discussed.

1.

Introduction

This paper pursuits a twofold goal. On one hand, it gives a detailed illustration of the
methods of Lagrangian quantum field theory in momentum picture, introduced in [1, 2], on
the simplest examples of free Hermitian (neutral, real) or non-Hermitian (charged, complex)
scalar field. On another hand, it contains an in-depth analysis of the (system of) Klein-Gordon equation(s) in momentum picture describing such fields. Most of the known fundamental
results are derived in a new way (and in a slightly modified form), but the work contains and
new ones.
We have to mention, in this paper it is considered only the Lagrangian (canonical) quantum field theory in which the quantum fields are represented as operators, called field operators, acting on some Hilbert space, which in general is unknown if interacting fields are
studied. These operators are supposed to satisfy some equations of motion, from them are
constructed conserved quantities satisfying conservation laws, etc. From the view-point of
present-day quantum field theory, this approach is only a preliminary stage for more or less
rigorous formulation of the theory in which the fields are represented as operator-valued distributions, a fact required even for description of free fields. Moreover, in non-perturbative
directions, like constructive and conformal field theories, the main objects are the vacuum
mean (expectation) values of the fields and from these are reconstructed the Hilbert space of
states and the acting in it fields. Regardless of these facts, the Lagrangian (canonical) quantum field theory is an inherent component of the most of the ways of presentation of quantum
field theory adopted explicitly or implicitly in books like [310]. Besides, the Lagrangian approach is a source of many ideas for other directions of research, like the axiomatic quantum
field theory [5, 9, 10].
The basic moments of the method, we will follow in this work, are the next ones:
(i) In Heisenberg picture is fixed a (second) non-quantized and non-normally ordered operator-valued Lagrangian, which is supposed to be polynomial (or convergent power series)
in the field operators and their first partial derivatives;
(ii) As conditions additional to the Lagrangian formalism are postulated the commutativity between the components of the momentum operator (see (2.6) below) and the Heisenberg
relations between the field operators and momentum operator (see (2.7) below);
(iii) Following the Lagrangian formalism in momentum picture, the creation and annihilation operators are introduced and the dynamical variables and field equations are written
in their terms;
(iv) From the last equations, by imposing some additional restrictions on the creation
and annihilation operators, the (anti)commutation relations for these operators are derived;
(v) At last, the vacuum and normal ordering procedure are defined, by means of which
the theory can be developed to a more or less complete form.
The main difference of the above scheme from the standard one is that we postulate the
below-written relations (2.6) and (2.7) and, then, we look for compatible with them and
the field equations (anti)commutation relations. (Recall, ordinary the (anti)commutation
relations are postulated at first and the validity of the equations (2.6) and (2.7) is explored
after that [11].)
In Sect. 2 are reviewed the basic moments of the momentum picture of motion in quantum
field theory.
The rest of the work is divided into two parts.
Part A, involving sections 39, deals with the case of neutral (Hermitian, real) free scalar
field. The contents of its sections is as follows:
In Sect. 3, the material of Sect. 2 is specialized to the case of free Hermitian (neutral,
real) scalar field; in particular, the Klein-Gordon equation in momentum picture is derived.
Sect. 4 is devoted to analysis of the Klein-Gordon equation (in momentum picture) in

terms of operators similar to (and, in fact, up to a phase factor and normalization constant,
identical with) the Fourier images of field operators in Heisenberg picture. From them, in
Sect. 5, are constructed the creation and annihilation operators which turn to be identical, up
to a phase factor and, possibly, normalization constant, with the ones known from Heisenberg
picture. Their physical meaning is discussed (or recalled). In Sect. 6, the 3-dimensional
creation and annihilation operators (depending on the 3-momentum) are introduced and
the field equation is written in their terms. It happens to be a tri-linear equation relative
to them. This new form of the field equation is utilized in Sect. 7 for a detailed analysis
of the (additional) conditions leading to the known commutation relations. In particular,
it is proved that, excluding the vanishing field case, the quantization of a free Hermitian
scalar field by anti -commutators is rejected by the field equation without an appeal to the
spin-statistics theorem (or other equivalent to it additional assertion).
Sect. 8 is devoted to the introduction of the concept of vacuum (state) which requires a
modification of the developed theory by a normal ordering of products of creation and/or
annihilation operators. The vacua of Heisenberg and momentum pictures happen to coincide.
The problem of state vectors, representing in momentum picture a free Hermitian scalar field,
is considered in Sect. 9. It turns to be rather trivial due to the absence of any interaction.
However, the construction of Fock base is recalled and the basic ideas of scattering theory
are illustrated in this almost trivial case.
The second part B of our work is devoted to the general case of charged or neural free
scalar field. Regardless of some overlap with part A, it concentrates mainly on the case of
non-Hermitian field. Most of the proofs in it refer or are based, at least partially, on similar
ones in the Hermitian case, investigated in part A. The problems of the choice of the initial
Lagrangian and the right definitions of the energy-momentum and charge operators are
partially discussed. The layout of part B, involving sections 1017, is similar to the one of
part A.
The description of arbitrary free scalar field is presented in Sect. 10.
An analysis of the system of Klein-Gordon equations describing free scalar fields is presented in Sect. 11. A feature of the momentum picture is that, in a case of non-Hermitian
field, the two equations of this system are not separate equations for the field and its Hermitian conjugate; they are mixed via the momentum operator. The creation and annihilation
operators are introduced in Sect. 12 and the field equations are written in their terms in
Sect. 13. They turn to be trilinear equations similar to the ones appearing the parafield
theory. The commutation relations are extracted from them in Sect. 15. These reveal the
non-equivalence of the theories build from different initial Lagrangians. To agree the results,
one needs different additional hypotheses/conditions, depending on the concrete Lagrangian
utilized. The best Lagrangian (of three considered) is pointed out. The charge and orbital
angular momentum operators are considered in Sect. 14. The vacuum is defined in Sect. 16,
where the normal ordering of the dynamical variables is described too. We point that the
last operation is the final step which leads to identical theories build from different initial
Lagrangians. Some problems concerning the state vectors of free scalar fields are discussed
in Sect. 17.
Sect. 18 closes the paper by pointing to its basic results.
The books [35] will be used as standard reference works on quantum field theory. Of
course, this is more or less a random selection between the great number of papers on the
theme.1
Throughout this paper ~ denotes the Plancks constant (divided by 2), c is the velocity
of light in vacuum, and i stands for the imaginary unit.
1

The reader is referred for more details or other points of view, for instance, to [6, 12, 13] or the literature
cited in [36, 12, 13].

The Minkowski spacetime is denoted by M . The Greek indices run from 0 to dim M 1 =
3. All Greek indices will be raised and lowered by means of the standard 4-dimensional
Lorentz metric tensor and its inverse with signature (+ ). The Einsteins
summation convention over indices repeated on different levels is assumed over the whole
range of their values.
At the end, a technical remark is in order. The derivatives with respect to operator-valued
(non-commuting) arguments will be calculated according to the rules of the classical analysis
of commuting variables, which is an everywhere silently accepted practice [3, 11]. As it is
demonstrated in [14], this is not quite correct but does not lead to incorrect results (except
the non-uniqueness of the conserved quantities) when free scalar fields are concerned. We
shall pay attention on that item at the corresponding places in the text.

2.

The momentum picture

The momentum picture in quantum field theory was introduce in [1, 2]. Its essence is in the
following.
Let us consider a system of quantum fields, represented in Heisenberg picture of motion
by field operators i (x) : F F, i = 1, . . . , n N, in systems Hilbert space F of states and
depending on a point x in Minkowski spacetime M . Here and henceforth, all quantities in
Heisenberg picture will be marked by a tilde (wave) over their kernel symbols. Let P
denotes the systems (canonical) momentum vectorial operator, defined via the energy-momentum tensorial operator T of the system, viz.
Z
1

T0 (x) d3 x.
(2.1)
P :=
c
x0 =const

Since this operator is Hermitian, P = P , the operator

U(x, x0 ) = exp


1 X
(x x0 ) P ,
i~

(2.2)

where x0 M is arbitrarily fixed and x M ,2 is unitary, i.e. U (x0 , x) := ( U(x, x0 )) =

U 1 (x, x0 ) = ( U(x, x0 ))1 and, via the formulae
X 7 X (x) = U(x, x0 )( X )

A(x)
7 A(x) = U(x, x0 ) ( A(x))
U 1 (x, x0 ),

(2.3)
(2.4)

realizes the transition to the momentum picture. Here X is a state vector in systems Hilbert
: F F is (observable or not) operator-valued function of x M
space of states F and A(x)
which, in particular, can be polynomial or convergent power series in the field operators
i (x); respectively X (x) and A(x) are the corresponding quantities in momentum picture.
In particular, the field operators transform as
i (x) 7 i (x) = U(x, x0 ) i (x) U 1 (x, x0 ).

(2.5)

Notice, in (2.2) the multiplier (x x0 ) is regarded as a real parameter (in which P is

linear). Generally, X (x) and A(x) depend also on the point x0 and, to be quite correct, one
2

The notation x0 , for a fixed point in M , should not be confused with the zeroth covariant coordinate
0 x of x which, following the convention x := x , is denoted by the same symbol x0 . From the context,
it will always be clear whether x0 refers to a point in M or to the zeroth covariant coordinate of a point
x M.

should write X (x, x0 ) and A(x, x0 ) for X (x) and A(x), respectively. However, in the most
situations in the present work, this dependence is not essential or, in fact, is not presented
at all. For this reason, we shall not indicate it explicitly.
As it was said above, we consider quantum field theories in which the components P
of the momentum operator commute between themselves and satisfy the Heisenberg relations/equations with the field operators, i.e. we suppose that P and i (x) satisfy the
relations:
[ P , P ] = 0
[ i (x), P ] = i~ i (x).

(2.6)
(2.7)

Here [A, B] := A B B A, being the composition of mappings sign, is the commutator/anticommutator of operators (or matrices) A and B. The momentum operator P
commutes with the evolution operator U(x, x0 ) (see below (2.12)) and its inverse,
[ P , U(x, x0 )] = 0

[ P , U 1 (x, x0 )] = 0,

(2.8)

due to (2.6) and (2.2). So, the momentum operator remains unchanged in momentum picture,
viz. we have (see (2.4) and (2.8))
P = P .
(2.9)
Since from (2.2) and (2.6) follows
i~

U(x, x0 )
= P U(x, x0 )
x

U(x0 , x0 ) = idF ,

(2.10)

we see that, due to (2.3), a state vector X (x) in momentum picture is a solution of the
initial-value problem
i~

X (x)
= P ( X (x))
x

X (x)|x=x0 = X (x0 ) = X

(2.11)

which is a 4-dimensional analogue of a similar problem for the Schr

odinger equation in
quantum mechanics [1517].
By virtue of (2.2), or in view of the independence of P of x, the solution of (2.11) is
1

X (x) = U(x, x0 )( X (x0 )) = e i~ (x

x ) P

( X (x0 )).

(2.12)

Thus, if X (x0 ) = X is an eigenvector of P (= P ) with eigenvalues p ,

P ( X (x0 )) = p X (x0 ) (= p X = P ( X )),

(2.13)

we have the following explicit form of the state vectors

1

X (x) = e i~ (x

x )p
0

( X (x0 )).

(2.14)

It should clearly be understood, this is the general form of all state vectors as they are eigenvectors of all (commuting) observables [5, p. 59], in particular, of the momentum operator.
In momentum picture, all of the field operators happen to be constant in spacetime, i.e.
i (x) = U(x, x0 ) i (x) U 1 (x, x0 ) = i (x0 ) = i (x0 ) =: (0) i .

(2.15)

Evidently, a similar result is valid for any (observable or not such) function of the field
operators which is polynomial or convergent power series in them and/or their first partial

derivatives. However, if A(x)

is an arbitrary operator or depends on the field operators in a
different way, then the corresponding to it operator A(x) according to (2.4) is, generally, not

spacetime-constant and depends on the both points x and x0 . As a rules, if A(x) = A(x, x0 )
is independent of x, we, usually, write A for A(x, x0 ), omitting all arguments.
It should be noted, the Heisenberg relations (2.7) in momentum picture transform into the
identities i = 0 meaning that the field operators i in momentum picture are spacetime
constant operators (see (2.15)). So, in momentum picture, the Heisenberg relations (2.7) are
incorporated in the constancy of the field operators.
Let L be the systems Lagrangian (in Heisenberg picture). It is supposed to be polynomial or convergent power series in the field operators and their first partial derivatives, i.e.
i (x), i (x)) with denoting the partial derivative operator relative to the th
L = L(
coordinate x . In momentum picture, it transforms into
i (x), yj )
L = L(

yj =

1
[j , P ] ,
i~

(2.16)

i.e. in momentum picture one has simply to replace the field operators in Heisenberg picture
with their values at a fixed point x0 and the partial derivatives j (x) in Heisenberg picture
with the above-defined quantities yj . The (constant) field operators i satisfy the following
algebraic Euler-Lagrange equations in momentum picture:3
i o
n L(
j , yl )
j , yl )
1 h L(

= 0.

, P

1
i
i~
yi
[j , P ]
yj = i~

(2.17)

Since L is supposed to be polynomial or convergent power series in its arguments, the

equations (2.17) are algebraic, not differential, ones. This result is a natural one in view
of (2.15).
By virtue of (2.4) and (2.9), the momentum operator (2.1) can be written as
1
P = P =
c

U 1 (x, x0 ) T0 U(x, x0 ) d3 x.

(2.18)

x0 =x00

This expression for P will be employed essentially in the present paper.

Part A
Free neutral scalar field
The purpose of this part of the present work is a detailed exploration of a free neutral
(Hermitian, real, uncharged) scalar field in momentum picture.4
After fixing the notation and terminology, we write the Klein-Gordon equation in momentum picture and derive its version in terms of creation and annihilation operators. The
3

In (2.17) and similar expressions appearing further, the derivatives of functions of operators with respect to operator arguments are calculated in the same way as if the operators were ordinary (classical)
fields/functions, only the order of the arguments should not be changed. This is a silently accepted practice in the literature [4, 5]. In the most cases such a procedure is harmless, but it leads to the problem of
non-unique definitions of the quantum analogues of the classical conserved quantities, like the energy-momentum and charge operators. For some details on this range of problems in quantum field theory, see [14]; in
particular, the loc. cit. contains an example of a Lagrangian whose field equations are not the Euler-Lagrange
equations (2.17) obtained as just described.
4
A classical real field, after quantization, becomes a Hermitian operator acting on the systems (fields)
Hilbert space of states. That is why the quantum analogue of a classical real scalar field is called Hermitian
scalar field. However, it is an accepted common practice such a field to be called (also) a real scalar field.
In this sense, the terms real and Hermitian scalar field are equivalent and, hence, interchangeable. Besides,
since a real (classical or quantum) scalar does not carry any charge, it is called neutral or uncharged scalar
field too.

famous commutation relations are extracted from it. After defining the vacuum for a free
Hermitian scalar field, some problems concerning the state vectors of such a field are investigated.

3.

Description of free neutral scalar field

in momentum picture

Consider a free neutral (Hermitian) scalar field with mass m. The corresponding to such a
field operator will be denoted by .
It is Hermitian, i.e.
(x) = (x),

(3.1)

where the dagger denotes Hermitian conjugation relative to the scalar product h|i of fields
Hilbert space F, and it is described in the Heisenberg picture by the Lagrangian
1
1
,
( ),

L = L(
)
= m2 c4 + c2 ~2 ( )
2
2

(3.2)

which in momentum picture transforms into (see (2.16))


1
1
y )
= m2 c4 0 0 c2 [0 , P ] [0 , P ] .
L = L(,
1
y = i~
[0 , P ]
2
2

(3.3)

Here 0 is the constant value of the field operator in momentum picture i.e. (see (2.15))5
(x) = U(x, x0 ) (x)

U 1 (x, x0 ) = (x0 ) = (x
0 ) =: 0 .

(3.4)

Since the operator U(x, x0 ) is unitary (see (2.2) and use the Hermiticity of the momentum
operator), the (momentum) field operator 0 is also Hermitian, i.e.
0 = 0 .
So, we have

(3.5)

:=

0 , y ) 
L(
= i~c2 [0 , P ]

1
y
[0 , P ]
y = i~
L
L
=
= m2 c4 0 .
0
0

(3.6)
(3.7)

Consequently, equation (2.17), in this particular situation, yields

m2 c2 0 [[0 , P ] , P ] = 0.

(3.8)

This is the Klein-Gordon equation in momentum picture. It replaces the usual Klein-Gordon
equation


2 2
e + m c idF (x)


= 0,
(3.9)
~2
5

The index 0 in 0 indicates the dependence on x0 , according to (3.4).

As pointed at the end of the Introduction, the calculation of the derivatives in (3.6) and (3.7) below
is not quite correct mathematically. However, the field equation (3.8) is correct; for its rigorous derivation,
see [14].
6

e := , in Heisenberg picture.7
where 
The energy-momentum tensorial operator T has two (non-equivalent) forms for free
Hermitian scalar field, viz.

T =
L = c2 ~2 ( )
( )
L
(3.10a)


1
1

+
) L = c2 ~2 ( )
( )
+ ( )
( )
L,
= (
2
2
(3.10b)

which in momentum picture read respectively:

1
T = c2 [0 , P ] [0 , P ] + c2 m2 c2 0 0 + [0 , P ] [0 , P ]
2


1 2
T = c [0 , P ] [0 , P ] + [0 , P ] [0 , P ]
2


1
+ c2 m2 c2 0 0 + [0 , P ] [0 , P ] .
2

(3.11a)

(3.11b)

Let us say a few words on the two versions, (3.10a) and (3.10b), of the energy-momentum
operator. The former one is a direct analogue of the classical expression for the energy-momentum tensor, while the latter variant is obtained from (3.10a) by a Hermitian symmetriza
tion. The second expression is symmetric and Hermitian, i.e. T = T and T
= T ,
while the first one is such if and commute for all subscripts and . However, as
we shall see, both forms of T lead to one and the same (Hermitian) momentum operator.
In this sense, the both forms of T are equivalent in quantum field theory. More details on
this problem will be given in Sect. 10.
Before going on, let us make a technical remark. The derivatives in (3.6) and (3.7) are
calculated according to the rules of classical analysis of commuting variables, which is not
correct, as explained in [14]. However, the field equation (3.8) or (3.9) is completely correct
for the reasons given in loc. cit. Besides, the two forms (3.10) of the energy-momentum
tensor operator are also due to an incorrect applications of the rules mentioned to the region
of analysis of non-commuting variables; the correct rigorous expression turns to be (3.10b).
The reader is referred for more detail on that item to [14], in particular to section 5.1 in it.
The only reason why we use a non-rigorous formalism is our intention to stay closer to the
standard books on Lagrangian quantum field theory. This approach will turn to be harmless
for the range of problems considered in the present paper.
A free neutral or charged scalar field has a vanishing spin angular momentum and possesses an orbital angular momentum, which coincides with its total angular momentum. The
orbital angular momentum density operator is

L (x) = L (x) = x T (x) x T (x)

= x T

x T

(3.12)
(3.13)

in Heisenberg and momentum pictures, respectively. The angular momentum operator in

7

As a simple exercise, the reader may wish to prove that the DAlembert operator (on the space of
operator-valued functions) in momentum picture is () = ~12 [[, P ] , P ] . (Hint: from the relation
A(x)
e
i ) = ~12 [[
i , Pt ] Pt ] ;
i~ x = [ A(x), Pt ] with Pt = i~ x + p idF (see [18]), it follows that (
t
t
now prove that in momentum picture P = P + P .) Now, the equation (3.8) follows immediately from
here and the usual Klein-Gordon equations (3.9). By the way, the above representation for () remains
valid in any picture of motion as it is a polynomial relative to () and P ; in particular, (3.8) is equivalent to
m2 c2
0 [
0 , P ] , P ] = 0 which is equivalent to (3.9) due to the (Heisenberg) relation [ 0 , P ] = i~ .

Heisenberg and momentum picture, respectively, is

Z
Z
1
1
0
0
0
3

L :=
{x T (x) x T (x)} d3 x
L (x) d x =
c
c
x0 =const
x0 =const
Z
1
=
U 1 (x, x0 ) {x T 0 x T 0 } U(x, x0 ) d3 x
c

(3.14)

x0 =const

L (x, x0 ) = U(x, x0 ) L U 1 (x, x0 ).

(3.15)

Since the spin angular momentum of a free scalar field is zero, the equations
L = 0

d L
=0
dx0

L = 0

(3.16)

express equivalent forms of the conservation law of angular momentum. As the treatment
of the orbital angular momentum for a neutral and charged scalar fields is quite similar, we
shall present a unified consideration of the both cases in Sect. 14.
A simple, but important, conclusion from (3.8) is that the operator
M2 : 0 7

1
[[0 , P ] , P ]
c2

(3.17)

should be interpreted as square-of-mass operator of the field in momentum (and hence in any)
picture. This does not contradict to the accepted opinion that the square-of-mass operator
is equal to the Lorentz square of the (divided by c) momentum operator. The problem here
is in what is called a momentum operator and in what picture the considerations are done.
Indeed, in Heisenberg picture, we can write
1
(i~)2
1
2 ( )
P ] , P ] = 2 [i~ ,
P ] = 2 ( )( (x))

M
= 2 [[ ,
c
c
c


1
1
(~)2
(x)

= 2 PQM P QM ( (x))

= = 2 [[ ,
Pt ] , P t ] , (3.18)
= 2 
c
c
c

where [18] PQM = i~ and Pt = PQM + p idF , with p = const and idF being the
identity mapping of F, are respectively the quantum mechanical and translational momentum
2 with the square of (divided
operators. So, we see that the conventional identification of M
by c) momentum operator corresponds to the identification of the last operator with PQM
For details on the last item the reader is referred to [18]. It
(or of its square with ~2 ).
should be noted, the eigenvalues of the operator (3.17) on the solutions of (3.8) characterize
the field (or its particles), while the eigenvalues of c12 P P on the same solutions are
characteristics of the particular states these solutions represent.

4.

Analysis of the Klein-Gordon equation

Our next aim is to find, if possible the (general) explicit form of the (constant) field operator
0 . The Klein-Gordon equation (3.8) is not enough for the purpose due to the simple
fact that the (canonical) momentum operator P depends on 0 . To show this, recall
the definition (2.1) of P and the expression (2.18) for it through the energy-momentum
operators T and T , given in Heisenberg and momentum pictures, respectively, by (3.10)
and (3.11). Therefore (3.8), (2.18), (3.11), and the explicit relation (2.2) form a closed
algebraic-functional system of equations for determination of 0 (and P ).
At the beginning of our analysis of the equations defining , we notice the evident solution
[0 , P ] = 0

for m = 0

(4.1)

of (3.8) which, by virtue of the Heisenberg equations/relations

(x)

[ (x),

P ] = i~
x

(4.2)

leads to (x)

= (x
0 ) = const for m = 0. Consequently (in the zero mass case) our system
of equations admits a, generally non-vanishing, solution (see also (3.4))
(x)

= (x
0 ) = const = 0

P = P = 0

for m = 0.

(4.3)

Evidently, (4.1) implies

T = 0

L = 0,

(4.4)

due to (3.11) and (3.13), and, consequently, the dynamical variables for the solutions (4.1)
vanish, i.e.
P = 0
L = 0.
(4.5)
For the general structure of the solutions of (3.8) is valid the following result.
Proposition 4.1. Every solution 0 of (3.8) and (2.7) is of the form
Z





0 =
d3 k f+ (k)0 (k)
2 + f (k)0 (k)
2 2
k0 =+

m c +k

k0 =

m2 c2 +k2

which can, equivalently, be rewritten as

Z
0 =
d4 k(k2 m2 c2 )f (k)0 (k).

(4.6)

(4.7)

Here: k = (k0 , k1 , k2 , k3 ) is a 4-vector with dimension of 4-momentum, k2 = k k = k02 k12

k2 k32 = k02 k2 with k being the components of k and k := (k1 , k2 , k3 ) = (k1 , k2 , k3 ) being
the 3-dimensional part of k, () is the (1-dimensional) Dirac delta function, 0 (k) : F F
is a solution of the equation
[0 (k), P ] = k 0 (k),

(4.8)

f are complex-valued functions (resp. distributions (generalized functions)) of k for solutions

different from (4.3) (resp. for the solutions (4.3)), and f is a complex-valued function (resp.
distribution) of k for solutions
different from (4.3) (resp. for the solutions (4.3)). Besides
p
2
2

f (k)|
m c + k2 f (k) for solutions different from (4.3).
2 = 2
2 2
k0 =

m c +k

Remark 4.1. In fact, in (4.6) enter only those solutions of (4.8) for which
k2 := k k = k02 k2 = m2 c2 .

(4.9)

Besides, a non-vanishing solution of (4.8) is a solution of (3.8) iff the condition (4.9) holds.
(Proof: write (3.8) with 0 (k) for 0 and use (4.8) twice.)
Remark 4.2. Evidently, the l.h.s. of (4.8) vanishes for the solutions (4.3). Therefore, we have
0 (x, 0) = 0 (x0 , 0) = const = 0 (0)

P = P = 0

(4.10)

where
0 (x, k) := U 1 (x, x0 ) 0 (k) U(x, x0 ).

(4.11)

In terms of (4.6), this solution is described by m = 0 and, for example, f (k) = ( 12 a)3 (k),
a C or f (k) such that f (k)|k0 =|k| = (1 2a)|k|3 (k), a C.

Remark 4.3. Since 0 is Hermitian (see (3.5)), we have


f (k)0 (k)

k0 =

m2 c2 +k2

due to (4.6). Also the similar relation


= f (k)0 (k)

k0 =

m2 c2 +k2

(f (k)0 (k)) = f (k)0 (k),

(4.12)

(4.13)

holds as a corollary of (4.7).8

Proof. Since the proposition was proved for the degenerate solutions (4.3) in remark 4.2,
below we shall suppose that (m, k) 6= (0, 0) (and, hence P 6= 0).
The equality (4.7) is equivalent to (4.6) (for solutions different from (4.3)) due to (y 2
1
2
((y a) + (y + a)) for a > 0.
a ) = 2a
Let 0 be given by (4.7). Using (4.8), we get
Z
[0 , P ] = d4 kk (k2 m2 c2 )f (k)0 (k).
(4.14)
Inserting this into (3.8) and, again, applying (3.8), we see that (4.7) is a solution of (3.8).
So, it remains to be proved that any solution of (3.8) is of the form (4.7).
The equation (4.8) in Heisenberg picture reads (see (2.5))
[ (x,
k), P ] = k (x,
k).

(4.15)

Combining this with (4.2) (with (x,

k) for (x)),

we find
1

(x,
k) = e i~ (x

x )k
0

(x
0 , k) = e i~ (x

x )k
0

0 (k)

(4.16)

as
0 (k) = (x
0 , k).

(4.17)

Hence, due to (2.5), the equation (4.7) in Heisenberg picture reads

(x)

d4 k(k2 m2 c2 )e+i ~ k x e i~ k x0 f (k)0 (k).

(4.18)

At the end, recalling that any solution (different from (4.3)) of the Klein-Gordon equation (3.9), satisfying the Heisenberg relation (5.3), admits a Fourier expansion of the form [3
5]
Z
1

(4.19)
(x)

=
d4 k(k2 m2 c2 )ei ~ k x (k)
for some operator-valued function 0 (k) satisfying (4.8) under the condition (4.9), viz.9
P ] = k (k)

[ (k),

k2 = m2 c2 ,

(4.20)

The same result follows also from the below-written equations (4.19) and (4.21). Indeed, (4.19) and

= imply
(k) = (k), which, in view of (4.21), entails (4.13).
0
0
9
The first equality in (4.20) is the Fourier image of (2.7).

the proof is completed by the identification

1

(k)

0 , k)
= e i~ k x0 f (k)0 (k) = e i~ k x0 f (k) (x

(4.21)

and subsequent return from Heisenberg to momentum picture.

Meanwhile, we have proved two quite important results. On one hand, by virtue of (4.21)
and the homogeneous character of (4.8), any solution of (3.8) can be written as
Z
(4.22)
0 = (k2 m2 c2 )0 (k) d4 k
where 0 (k) are appropriately normalized (scaled) solutions of (4.8) which solutions can be
1
identified, up to the phase factor e i~ x0 k , with the Fourier coefficients of (x),

viz. (see (4.19)

and (4.21))
1
= e i~ x0 k 0 (k).
(4.23)
(k)

Therefore, by virtue of (4.13) and (4.21), the operators 0 (k) appearing in (4.22) satisfy the
relation (cf. (4.13))
(0 (k)) = 0 (k).
(4.24)
On the other hand, this implies that the operator 0 (k), entering in (4.22), is (up to a constant) identical with the usual momentum representation of a scaler field (x)

in Heisenberg
with k2 = m2 c2 , representing a free Herpicture [35]. Consequently, the operators (k)

mitian scalar field in momentum representation of Heisenberg picture of motion in quantum

field theory, are nothing else but (suitably normalized) solutions of (4.8) under the condition (4.9), i.e. of the Klein-Gordon equation (3.8), which form a basis in the operator space
of all solutions of (3.8). In short, the momentum representation of a scalar field in Heisenberg
picture is a suitably chosen base for the solutions of the Klein-Gordon equation in momentum
picture. This result is quite important from two aspects. On one hand, it reveals the real
meaning of the momentum representation in Heisenberg picture of quantum field theory. On
another hand, it allows us to apply freely in momentum picture all of the already established
results concerning the Fourier images of the field operators, observables and other operators
in Heisenberg picture. In particular, this concerns the frequency decompositions and creation
and annihilation operators theory.

5.

Frequency decompositions and their physical meaning

Since the decomposition (4.6) (or (4.7) if the solutions (4.3) are excluded) is similar to the one
leading to the frequency decompositions (in Heisenberg picture for free fields) in quantum
field theory, we shall introduce similar (in fact identical) notation. Defining10
(
f (k)0 (k) for k0 0
,
(5.1)

0 (k) =
0
for k0 < 0
we see that

0 =

0 = +
0 + 0

d3 k
0 (k)|

[
0 (k), P ] = k 0 (k)
10

k0 =

(5.2)
m2 c2 +k2

k0 =

m2 c2 + k2

Notice, we do not exclude the case k0 = 0 as it is done in the literature.

(5.3)
(5.4)

due to (4.6) and (4.8). Notice, since 0 and 0 (k) satisfy (3.5), (4.13) and (4.12), we have

(
0 (k)) = 0 (k).

(5.5)

The equation (5.4) implies the interpretation of a free scalar field in terms of particles.
Indeed, if Xp is a state vector of a state with 4-momentum p, i.e.
P ( Xp ) = p Xp ,

(5.6)

then, applying (5.4), we obtain

P (
0 (k)( Xp )) = (p k )0 (k)( Xp )

k0 =

m2 c2 + k2

(5.7)

So, +
0 (k) (resp. 0 (k)) can be interpreted as an operator creating (resp.
p annihilating) a
particle (quant of thepfield) with mass m and 4-momentum k with k0 = m2 c2 + k2 , i.e. a
particle with energy m2 c2 + k2 and 3-momentum k. In this situation, the vacuum for a
scalar field should be a state X0 with vanishing 4-momentum and such that
p

k0 = m2 c2 + k2 .
(5.8)
0 (k)( X0 ) = 0

The action of the operators +

0 (k) on X0 produces one- or multiple-particle states. More
details concerning the definition of a vacuum and construction of state vectors from it will
be given in sections 8 and 9.
For a comparison with existing literature (see, e.g., [35, 12]), we notice that the creation/annihilation operators , introduced in momentum representation of Heisenberg picture of motion, are defined by
(
for k0 0
(k)

(5.9)
(k) =
0
for k0 < 0

where (k)

is the Fourier image of (x)

(see (4.19)). Therefore, by virtue of (4.21) and

proposition 4.1, we have the following connection between the creation/annihilation operators
in Heisenberg and momentum picture:
p
1
(k) = e i~ x0 k
m2 c2 + k2 .
(5.10)
(k)
k
=
0
0
We leave to the reader to prove as an exercise that the general formula (2.5) implies
p
1
1
(5.11)
(k) = e i~ x k U(x, x0 )
(x, x0 )
k0 = m2 c2 + k2 ,
0 (k) U

which, in view of (2.7) and (5.4), is equivalent to (5.10). (Hint: apply the Fourier decomposition from Sect. 4, then show that x (k) = 0 and, at last, set x = x0 in (5.11).)

6.

The field equations in terms of creation and

annihilation operators

Let us return now to our main problem: to be found, if possible, the explicit form of the field
operator 0 (and momentum operator P ).
At first, we shall express P in terms of the creation and annihilation operators
0.
Regardless of the fact that the result is known (see, e.g., [3, eq. (3.26)] or [4, eq. (12.11)]),
we shall reestablish it in a completely new way.

Since (5.2)(5.4) entail

[0 , P ] =

d3 k k (+
0 (k) + 0 (k)) |

k0 =

m2 c2 +k2

from (3.11), after some algebra (involving (5.2) and (5.3)), we get the energy-momentum
operator as
2

T = c

1
= c2
2


1

d3 k d3 k (k k + k k )(+
0 (k) + 0 (k))
2

1

+


2 2 +

(6.1a)
(+
0 (k ) + 0 (k )) + m c (0 (k) + 0 (k))(0 (k ) + 0 (k ))
2

d3 k d3 k (k k k k + k k )(+
0 (k) + 0 (k))

2 2 +
(+
0 (k ) + 0 (k )) + m c (0 (k) + 0 (k))(0 (k ) + 0 (k )) , (6.1b)
p
p
2
where k0 = m2 c2 + k2 , k0 = m2 c2 + k and the two expressions for T correspond to
the two its versions in (3.11).

The idea is now the last result to be inserted into (2.18) and to commute
0 (k) and 0 (k )

with U(x, x0 ) in order to move 0 (k) and 0 (k ) to the right of U(x, x0 ). Rewriting (4.8)
as 0 (k) P = ( P k idF ) 0 (k), by induction, we derive
0 (k) ( P1 Pa ) = {( P1 k1 idF ) ( Pa ka idF )} 0 (k)

(6.2)

for any a N, which, in view of the expansion of the r.h.s. of (2.2) into a power series, implies
1

0 (k) U(x, x0 ) = e i~ (x

x )k
0

U(x, x0 ) 0 (k).

(6.3)

So, we have (see (5.1))

1

x )(k )

x )(k k )

i~ (x

0 (k) U(x, x0 ) = e

i~ (x
+
0 (k) 0 (k ) U(x, x0 ) = e

U(x, x0 )
0 (k)

U(x, x0 )+
0 (k) 0 (k )

(x x
i~
0 )(k k )

0 (k) 0 (k ) U(x, x0 ) = e

(6.4)

U(x, x0 )
0 (k) 0 (k ).

At the end, substituting (6.1) into (2.18), applying the derived commutation rules (6.4),
performing the integration over x, which yields -functions like (k k ), and the integration
over k , we finally get after, a simple, but lengthy and tedious, calculation, the following
result
Z
1

+
3
(6.5)
P =
k | 2 2 2 {+
0 (k) 0 (k) + 0 (k) 0 (k)} d k,
k0 = m c +k
2
where we have introduced the 3-dimensional renormalized creation and annihilation operators


3
1/2

0 (k) 
0 (k) := (2c(2~) k0 )

k0 =

m2 c2 +k2

(6.6)

Notice, the result (6.5) is independent of from what form of T , (3.11a) or (3.11b), we have
started. The only difference of our derivation of (6.5) from similar one in the literature (see,
e.g., [3, sec. 3.2]) is that we have not exclude the massless case and the degenerate solu
tion (4.3) from our considerations. The operator +
0 (k) (resp. 0 (k)) is called the creation
(resp. annihilation) operator (of the field).

We would like to emphasize on the relations

(
0 (k)) = 0 (k)

(6.7)

which follow from (6.6) and (5.5). Actually these equalities are equivalent to the supposition
that 0 is Hermitian field operator.
As a result of (5.10), the operators (k), corresponding to (6.5) in (the momentum
representation of) Heisenberg picture of motion, are [3, Sec. 3.2]


1/2

(k) : = c(2~)3
(2k0 )1/2 (k)  2 2 2
k0 = m c +k
(6.8)

1
(k).
= e i~ x0 k  2 2 2
0
k0 =

m c +k

Similarly, we have also the connection (see (5.11))


1
(k) = e i~ x k  2 2 2 U 1 (x, x0 )
0 (k) U(x, x0 ).
k0 =

m c +k

(6.9)

Consequently, the integrand in (6.5) and similar ones which will be met further in this work,
look identically in terms of
(k).
0 (k) and
Summarizing the above results, we are ready to state and analyze the following problem.
Problem 6.1. Let
0 =

+
0

+
0

(+
0 (k) + 0 (k))|

k0 =

m2 c2 +k2

d3 k

Z 
1/2
p

3
(+
2c(2~)3 m2 c2 + k2
=
0 (k) + 0 (k)) d k

(6.10)

be a solution of the Klein-Gordon equation (3.8). Find the general explicit form of the
operators
0 (k) which are solutions of the equations (see (5.4))
p

(k),
P
]
[
=
k

(k)
k
=
m2 c2 + k2
(6.11)

0
0
0

where P is given by (6.5).

Inserting the representation (6.5)

pinto (6.11) and writing the expression k 0 (k) as

3
3
q 0 (k) (k q) d q with q0 = m2 c2 + q 2 , we, after a simple algebraic manipulation,
obtain
Z

+

+
q |q =m2 c2 +q2 [
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]
0
3
3
2
0 (k) (k q) d q = 0 (6.12)

Consequently
0 (k) must be solutions of

[
0 (k), [0 (q), 0 (q)]+ ] 20 (k) (k q) = f (k, q)

(6.13)

where [A, B]+ := A B + B A is the anticommutator of operators (or matrices) A and B

and f (k, q) are operator-valued (generalized) functions such that
Z
(6.14)
q |q =m2 c2 +q2 f (k, q) d3 q = 0.
0

Looking over the derivation of (6.13), we see its equivalence with the initial system of
Klein-Gordon equation (3.8) and Heisenberg relations (2.7) in momentum picture. Consequently, (6.13) is the system of field equations in terms of creation and annihilation operators
in momentum picture.
As a simple test of our calculations, one can prove that the commutativity of the components P of the momentum operator, expressed by (2.6) (see also (2.9)), is a consequence
of (6.5), (6.13), and (6.14).

7.

The commutation relations

The equations (6.13) are the corner stone of the famous (anti)commutation relations in quantum field theory for the considered here free Hermitian scalar field.11 The equations (6.13)
form a system of two homogeneous algebraic equations of third order relative to the functions

0 (k). Generally, it has infinitely may solutions, but, at present, only a selected class of
them has a suitable physical meaning and interpretation. This class will be described a little
below.
Since for the physics is essential only the restriction of
0 (k) on the physically realizable
states, not on the whole systems Hilbert space F, we shall analyze (6.13) in this case,
following the leading idea in similar cases in, e.g., [3, subsec. 10.1], or [4, 70], or [5, p. 65].
It consists in admitting that the commutator or anticommutator of the creation and/or
annihilation operators (for free fields) is a c-number, i.e. it is proportional to the identity
mapping idF of the systems Hilbert space F of states. In particular, in our case, this
(additional hypothesis) results in

[
0 (k), 0 (q)] = (k, q) idF

[
0 (k), 0 (q)] = (k, q) idF

(7.1)

where = , [A, B] := AB B A, and

and are complex-valued (generalized) functions. It should clearly be understood, the conditions (7.1) are additional to the Lagrangian
formalism and do not follow from it. In fact, they or the below written commutation relations (7.28) (which under some conditions follow from (7.1)) have to be postulated.
Defining

[A, B] = A B + B A

(7.2)

and applying the identity

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

= 1

(7.3)

for = 1, viz.
[A, B C] = [A, B] C B [A, C] ,

(7.4)

due to = 1 and [A, B]1 = [A, B] , we rewrite (6.13) as

+

[
0 (k), 0 (q)] 0 (q) 0 (q)[0 (k), 0 (q)]

+
+[
0 (k), 0 (q)] 0 (q) 0 (q)[0 (k), 0 (q)]
3
2
0 (k) (k

(7.5)

q) = f (k, q).

Now, substituting the additional conditions (7.1) into (7.5), we get

+
(1 )
0 (q) [0 (k), 0 (q)]

+ (1 )+
0 (q) [0 (k), 0 (q)] 20 (k) (k q) = f (k, q). (7.6)

For = +1, this equality reduces to 2

0 (k) (k q) = f (k, q) which, when inserted
into (6.14), entails

k |

k0 =

11

m2 c2 +k2

0 (k) = 0

(7.7)

The idea for arbitrary (Hermitian/real or non-Hermitian/complex, free or interacting) fields remains the
same: one should derive the Euler-Lagrange equations in momentum picture and to solve them relative to
the field operators by using the explicit expression of the (canonical) momentum operator through the field
operators.

for any = 0, 1, 2, 3 and k. Consequently, the choice = +1, i.e.

[
0 (k), 0 (q)]+ = + (k, q) idF

[
0 (k), 0 (q)]+ = + (k, q) idF ,

(7.8)

leads, in view of (6.5) and (7.7), to

P = 0

(7.9)

which, by virtue of Klein-Gordon equation (3.8), implies

m2 c2 0 = 0.

(7.10)

So, if m = 0, we get the solution (4.3) and, if m 6= 0, we derive

0 = 0

for m 6= 0

(7.11)

and, consequently,

0 (k) 0

for m 6= 0

(7.12)

We interpret the obtained solution (7.11) (or (7.12)) of the Klein-Gordon equation (3.8) as
a complete absence of the physical scalar field.12
Consider now (7.6) with = 1. Writing it explicitly for the upper, +, and lower,
, signs, we see that (7.6) is equivalent to

0 (q) [0 (k), 0 (q)] + [0 (k), 0 (q)] 0 (q) 0 (q) (k q) =

1
f (k, q)
2

(7.13)

where, in accord with (7.1), the commutators are

[
0 (k), 0 (q)] = (k, q) idF

[
0 (k), 0 (q)] = (k, q) idF .

(7.14)

Here, for brevity, we have omit the subscript , i.e. we write and for
and
respectively.
The properties of and can be specified as follows.13
Let Xp be a state vector with fixed 4-momentum p (see (5.6)). Defining

Xk,q
:= (
0 (k) 0 (q))( Xp )

Yk,q
:= (
0 (k) 0 (q))( Xp ),

(7.15)

.
) = (p k q ) Yk,q
P ( Yk,q

(7.16)

we, applying (5.7), get

) = (p k q ) Xk,q
P ( Xk,q

Noticing that

Xk,q
+ Xq,k
= [
0 (k), 0 (q)] ( Xp )

Yk,q
+ Yq,k
= [
0 (k), 0 (q)] ( Xp )

for = 1, we see that

12

P [
0 (k), 0 (q)] ( Xp ) = (p k q )[0 (k), 0 (q)] ( Xp )

P [
0 (k), 0 (q)] ( Xp ) = (p k q )[0 (k), 0 (q)] ( Xp ).

(7.17)

Since, for a scalar field, the solutions (4.3) do not lead to any physically predictable results, they, in the
massless case, should also be interpreted as absence of the field.
13
For the initial idea, see [3, subsec. 10.1], where the authors (premeditated or not?) make a number of
implicit assumptions which reduce the generality of the possible (anti)commutation relations.

It is worth to mention, in the derivation of (7.17) no additional hypothesis, like (7.1), have
been used.
Applying (7.17) for = 1 and imposing the additional conditions (7.14), we, due to (5.6),
find
(k + q) (k, q) = 0
(k q) (k, q) = 0
p
p
)


m2 c2 + k2 + m2 c2 + q 2 (k, q) = 0
p
p


m2 c2 + k2 m2 c2 + q 2 (k, q) = 0.

(7.18)
(7.19)

Here, with necessity, and must be regarded as distributions (generalized functions)

as otherwise the equations (7.18) have only the trivial solutions
(k, q) = 0 for k + q 6= 0

(k, q) = 0

for k q 6= 0

(7.20)

which, as it can easily be seen, reproduce the trivial solution (7.11) of the Klein-Gordon
equation. Rewriting (7.18) as q (k, q) = k (k, q) and q (k, q) = k (k, q), we see
that
f (q) (k, q) = f (k) (k, q)
f (q) (k, q) = f (k) (k, q)
(7.21)
for a function f (k) which is supposed to be polynomial or convergent power series.14 As a
result of (7.21), the second equation in (7.19) is equivalent to the identity 0 = 0, while the
first one reduces to
p
m2 c2 + k2 (k, q) = 0.
(7.22)
Inserting (7.14) into (7.13), we find

3
f (k, q) =
0 (q) (k, q) + 0 (q)( (k, q) (k q)).
2

(7.23)

At the end, besides (7.22), the condition (6.14) is the

p only one remaining to be satisfied.
Substituting (7.23) into it and using (7.21) for f (q) = m2 c2 + q 2 in the case = 0 and for
f (q) = qa in the one with = a = 1, 2, 3, we obtain
Z


3

3
(k, q)
0 (q) + ( (k, q) (k q))0 (q) d q
Z
p

3

3
2
2
2
+ (k, q)
0= m c +k
0 (q) + ( (k, q) (k q))0 (q) d q,
0 = ka

(7.24a)
(7.24b)

where, in the second equation, the term containing vanishes due to (7.22). Since these
equations must be valid for arbitrary k, the integrals in them should vanish if k 6= 0. Forming
the sum and difference of these integrals in that case, we get
Z
3
(k, q)
(7.25a)
0 (q) d q = 0
Z
3
( (k, q) 3 (k q))
(7.25b)
0 (q) d q = 0.
If (m, k) 6= (0, 0), the standard (Bose-Einstein) commutation relations are extracted from
equations (7.24) (or (7.25) if k 6= 0) by imposing a second, after (7.1), additional condition.
14

We cannot write, e.g., (k, q) = const 3 (k+q) as the equation yg(y) = 0, y R, has a solution g(y) =
2
(y)
+2 y 2 d dy
const(y), but this is not its general solution; e.g., its solutions are g(y) = 0 (y)+1 y d(y)
2 +
dy
with 0 , 1 , . . . being constant numbers.

If we demand (7.24) to be valid for arbitrary

0 (q) whose commutators are c-numbers,
then (7.24) entail
(k, q) = 0
(k, q) = 3 (k q)
(7.26)
which convert (7.22) into identity.
Let us say also a few words on the special case when m = 0 and k = 0. In it one
cannot extract any information from (7.24) and (7.22). Recall, if m = 0, the operators

0 (0) describe creation/annihilation of massless particles with zero 4-momentum. Since,

in the Hermitian scalar case, the 4-momentum and mass are the only characteristics of the
fields quanta, the particles corresponding to

0 (k)

with m = 0 and k = 0

(7.27)

are unphysical as they cannot lead to some observable consequences. The operators (7.27)
are special kind of the solutions (4.3) of the Klein-Gordon equation and, consequently, can
be interpreted as absence of the field under consideration. Besides, as we proved above under
the hypothesis (7.14), the only restrictions to which
0 (0) with m = 0 should be subjected
are the conditions (7.14) with k = 0, arbitrary q, and any (generalized) functions and
. Thus, to ensure a continuous limit when (m, k) (0, 0), we, by convention, choose
(0, q) and (0, q) to be given by (7.26) with k = 0.15
So, we have obtained the next solution of the problem 6.1. In momentum picture, the
Klein-Gordon equation (3.8) admits a solution (6.10), i.e.
Z 
1/2
p

3
(+
2c(2~)3 m2 c2 + k2
0 =
0 (k) + 0 (k)) d k,
in which the creation/annihilation operators
0 (k) satisfy only the commutation relations

[
0 (k), 0 (q)] = 0

3
[
0 (k), 0 (q)] = (k q) idF .

(7.28)

Now, it is a trivial exercise to be verified that, in view of (6.5) and (7.28), the equality (6.11) is identically valid.
We would like to remark, in the above considerations the massless case, i.e. m = 0, is
not neglected. However, obviously, the commutation relations (7.28) exclude the degenerate
solution (7.11) or, equivalently, (7.20) or (7.12).
More generally, one can look for solutions of the tri-linear (para?)commutation relation (6.13), under the condition (6.14) which do not satisfy the additional conditions (7.1).
But this is out of the aims of this work.16
As we have noted several times above, the concepts of a distribution (generalized function) and operator-valued distribution appear during the derivation of the commutation
relations (7.28). We first met them in the tri-linear relations (6.13). In particular, the

canonical commutation relations (7.28) have a sense iff [

0 (k), 0 (k)] is an operator-valued
distribution (proportional to idF ), which is not the case if the field 0 (or 0 ) is an ordinary
operator acting on F. These facts point to inherent contradiction of quantum field theory if
the field variables are considered as operators acting on a Hilbert space. The rigorous mathematical setting requires the fields variables to be regarded as operator-valued distributions.
15

In this way we exclude from the theory a special kind of absent (unphysical) field described by m = 0
and 0 = 0 or
0 (k) = 0.
16
Tri-linear relations, like (6.13), are known as paracommutation relations and were discovered in [19] (See
also [20, 21]). However, it seems that at present are not indications that solutions of (6.13), which do not
satisfy (7.28), may describe actually existing physical objects or phenomena [10, 22, 23]. This is one of the
reasons the quantum field theory to deal with (7.28) instead of (6.13) (or the equivalent to it Klein-Gordon
equation (3.8) (in momentum picture) or (3.9) (in Heisenberg picture). Elsewhere we shall demonstrate how
the parabose-commutation relations for a free scalar field can be derived from (6.13).

However, such a setting is out of the scope of the present work and the reader is referred to
books like [9,10,24,25] for more details and realization of that program. In what follows, the
distribution character of the quantum fields will be encoded in the Diracs delta function,
which will appear in relations like (6.13) and (7.28).
Ending the discussion of the commutation relations for a free Hermitian scalar field, we
would like to note that the commutation, not anticommutation, relations for it were derived
directly from the Klein-Gordon equation (3.8) without involving the spin-statistics theorem,
as it is done everywhere in the literature [3, 5]. In fact, this theorem is practically derived
here in the special case under consideration. Besides, we saw that the commutation relations
can be regarded as additional restrictions, postulated for the field operators as stated, e.g.,
in [5, pp. 5960], which must be compatible with the equations of motion. In fact, as we
proved, these relations are, under some assumptions, equivalent to the equations of motion,
i.e. to the Klein-Gordon equation in our case. Said differently, the commutation relations
convert the field equation(s) into identity (identities) and, in this sense are their solutions. An
alternative viewpoint is the commutation relations (7.28) to be considered as field equations
(under the conditions specified above) with respect to the creation and annihilation operators
as field operators (variables).
To close this section, we have to make the general remark that the tri-linear relations (6.13) (together with (6.14)) are equivalent to the initial Klein-Gordon equation (in
terms of creation/annihilation operators) and all efforts for the establishment of the commutation relations (7.28) reflect, first of all, the fact of extraction of physically essential solutions
of these equations.17 In this sense, we can say that the commutation relations (7.28) are a
reduction of the initial Klein-Gordon equation (3.8) (in momentum picture) or (3.9) (in
Heisenberg picture), under the conditions (7.1) and the assumption that (7.24) hold for any

0 (q).

8.

The vacuum and normal ordering

The vacuum of a free Hermitian scalar field 0 is a particular its state, described by a state
vector X0 which carries no energy-momentum and, correspondingly, it is characterized by a
constant (in spacetime) state vector, i.e.
P ( X0 (x)) = 0
X0 (x) = X0 (x0 ) = X0

(8.1)
(8.2)

where X0 and X0 are the vacua in Heisenberg and momentum pictures respectively. Equation (8.1) can be taken as a macroscopic definition of the vacuum state vector X0 . Since,
microscopically, the field is considered as a collection of particles (see Sect. 5), the vacuum
should be considered as a vector characterizing a state with no particles in it. Recalling the

interpretation of creation, +
0 (k), and annihilation, 0 (k), operators from Sect. 5, we can
make the definition (8.1) more precise by demanding

0 (k)( X0 ) = 0

X0 6= 0.

(8.3)

This is the everywhere accepted definition of a vacuum for a free Hermitian scalar field.
However, it does not agree with the expression (6.5) for the momentum operator and the

commutation relations (7.28). In fact, commuting +

0 (k) and 0 (k), according to (7.28), in
the second term in the integrand in (6.5) and using (8.3), we get
(
Z
4
for = 0
1
3
3
3
. (8.4)
k (0) d k X0 = X0 (0)
P ( X0 ) =
2
2
2
2
( ) for = 1, 2, 3
17

One may recognize in (6.13) a kind of paracommutation relations which are typical for the so-called
parastatistics [1921, 26].

Of course, this is a nonsense which must be corrected. The problem can be solved by repairing the r.h.s. of (6.5), by replacing the commutation relations (7.28) with other relations
(compatible with (6.13) and (6.14)), or by some combination of these possibilities. At this
point, we agree with the established procedure for removing (8.4) from the theory. If one
accepts not to change the logical structure of the theory, the only possibility is a change
in the Lagrangian from which all follows. Since in (8.4) the infinities come from the term
+

0 (k) 0 (k) in (6.5), it should be eliminated somehow. The known and, it seems, well
working procedure for doing this, which we accept, is the following one. At first the Lagrangian and the dynamical variables, obtained from it and containing the field 0 , should

be written, by means of (5.1)(5.3) in terms of creation, +

0 (k), and annihilation, 0 (k) operators. Then, any composition (product) of these operators, possibly appearing under some
integral sign, must be change so that all creation operators to stand to the left relative to all
annihilation operators.18 The described procedure for transforming compositions of creation
and annihilation operators is known as normal ordering and the result of it is called normal
product of the corresponding operators. This item is discussed at length in the literature to
which the reader is referred for details [35, 27].
The mapping assigning to a composition of creation/annihilation operators their normal
product will be denoted by N and called normal ordering operator. The action of N
on polynomials or convergent power series of creation/annihilation operators is defined by
extending it by linearity; the resulting mapping
being denoted by N too.


+

(k)

(k)
=

(k)

Since, evidently, N
0 (k), the representation (6.5) of the
0
0
0
momentum operator changes, after normal ordering, into
Z

P =
ds kk | 2 2 2 +
(8.5)
0 (k) 0 (k).
k0 =

m c +k

(Notice, after normal ordering, we retain the notation P for the object resulting from (6.5).
This is an everywhere accepted system of notation in the literature and it is applied to all
similar situations, e.g. for the Lagrangian L or energy-momentum operator T . After some
experience with such a system of doubling the meaning of the symbols, one finds it useful
and harmless.) Now the equality (8.1) is a trivial corollary of (8.5) and (8.3).
As X0 6= 0, we shall assume that the vacuum X0 can be normalized to unity, viz.
h X0 | X0 i = 1,

(8.6)

where h|i : F F C is the Hermitian scalar product of the Hilbert space F of states.
In fact, the value h X0 | X0 i is insignificant and its choice as the number 1 is of technical
character. In this way, in many calculations, disappears the coefficient h X0 | X0 i. Prima facie
one can loosen (8.6) by demanding X0 to have a finite norm, but this only adds to the theory
the insignificant constant h X0 | X0 i which can be eliminated by a rescaling of X0 .19
Let us summarize the above discussion. The vacuum of a free scalar field is its physical
state which does not contains any particles and has zero energy and 3-momentum. It is
described by a state vector, denoted by X0 (in momentum picture) and also called the
18

The so-formulated rule is valid only for integer spin particles/fields. By virtue of (7.28), the order of the
different creation/annihilation operators relative to each other is insignificant, i.e. it does not influence the
result of described procedure.
19
If X0 has an infinite norm, so is the situation with any other state vector obtained from X0 via action
with creation operators, which makes the theory almost useless.

vacuum of the field, such that:

X0 6= 0
X0 = X0

(8.7b)

h X0 | X0 i = 1

(8.7d)

0 (k)( X0 )

(8.7a)

=0

(8.7c)

for any 3-momentum k. Since the existence of the vacuum X0 in the Heisenberg picture is a
known theorem, the condition (8.7b), expressing the coincidence of Heisenberg vacuum and
momentum vacuum, ensures the existence of the vacuum X0 in momentum picture. Besides,
to make the theory sensible, we have assumed normal ordering of the creation/annihilation
operators in the Lagrangian and all observables derived from it.
The normal ordering of products changes the field equations (6.13) into
1

3
[
0 (k), 0 (q) 0 (q)] 0 (k) (k q) = f (k, q)
2

(8.8)

as the anticommutator in (6.13) originates from the expression (6.5) for the momentum
operator (before normal ordering). The conditions (6.14) remains unchanged. However, by
means of (7.4) with = 1 and the commutation relations (7.28), one can verify that (8.8)
and (6.14) are identically valid. This means that, in fact, the commutation relations (7.28)
play a role of field equations under the hypotheses made.

9.

State vectors and transitions between them

According to the general theory of Sect. 2, the general form of a state vector X (x) of a free
Hermitian scalar field in momentum picture is
1

X (x) = U(x, x0 )( X (x0 )) = e i~ (x

x ) P

( X (x0 )),

(9.1)

where x0 M is an arbitrarily fixed point in the Minkowski spacetime M , P is given

via (8.5), and the initial value X (x0 ) of X at x0 is identical with the (constant) state vector
X representing the same state as X (x) but in Heisenberg picture, i.e.
X (x0 ) = X .

(9.2)

In particular, if Xp is a state vector at x0 with fixed 4-momentum p , i.e.

P ( Xp ) = p Xp ,

(9.3)

then its general form at a point x is

1

Xp (x) = e i~ (x

x )p
0

Xp

(9.4)

and P ( Xp (x)) = p Xp (x)

From the vacuum X0 can be constructed a base for the systems Hilbert space F, called
the Fock base. A general s-particle state containing s particles with momenta k1 . . . , ks (some
of these vectors may be identical) of a scalar field 0 has the form


+
X (k1 , . . . , ks ) = fs (k1 , . . . , ks ) +
(9.5)
0 (k1 ) 0 (ks ) ( X0 )

for some function fs characterizing the distribution of the particles.20 Notice, by virtue (4.9),
the 4-momenta k1 , . . . , ks are subjected to the conditions ka2 = m2 c2 , a = 1, . . . , s, with
m being the mass of the quanta of the field 0 . An arbitrary state, described via the
Klein-Gordon equation (3.8) or its version expressed by (4.8) and (4.9), can be presented as
a superposition of all possible states like (9.5), viz.
Z
XZ


+
X =
dk1 . . .
dks fs (k1 , . . . , ks ) +
(9.6)
0 (k1 ) 0 (ks ) ( X0 ).
s0

The above results, concerning free Hermitian scalar field, are identical with similar ones in
the momentum representation in ordinary quantum field theory (in Heisenberg picture); the
difference being that now 0 (k) is the field operator in the momentum picture, which, as we
proved, are identical with the Fourier images of the field operators in Heisenberg picture. So,
the Fock base goes without changes from Heisenberg into momentum picture.
One of the main problems in quantum field theory is to find the amplitude for a transition
from some initial state Xi (xi ) into a final state Xf (xf ), i.e. the quantities
Sf i (xf , xi ) := h Xf (xf )| Xi (xi )i

(9.7)

called elements of the so-called S-matrix (scattering matrix). Ordinary one considers the
limits of (9.7) with x0f + and x0i or, more generally, xf + and xi .
These cases are important in the scattering theory but not for the general theory, described
here, and, respectively, will not be discussed in our work. If we know Xf and Xi at some
(0)
(0)
points xf and xi , respectively, then, combining (9.1) and (9.7), we get
(0)

(0)

(0)

(0)

Sf i (xf , xi ) = h Xf (xf )|e i~ (xi xf ) P ( Xi (xi ))i = h Xf (xf )| U(xi , xf )( Xi (xi ))i. (9.8)
Consequently (cf. [11, 107]), the operator ( U(xf , xi )) = U 1 (xf , xi ) = U(xi , xf ), where
means (Hermitian) conjugation, has to be identified with the S-operator (often called also
S-matrix). To continue the analogy with the S-matrix theory, we can expand the exponent
in (9.8) into a power series. This yields (see (8.5))
U(xi , xf ) = idF +

n=1

U (n) (xi , xf )

(9.9)

1 1
U (n) (xi , xf ) :=
(x1 xf 1 ) . . . (xi n xf n )
n! (i~)n i
Z
(1)
+ (n)
(1)
(n)
+
)
) (9.10)
d3 k(1) . . . d3 k(n) k(1)
k(n)
0 (k ) 0 (k ) 0 (k
0 (k
n
1

q
(a)
where k0 = m2 c2 + (k(a) )2 , a = 1, . . . , n. This expression is extremely useful in scattering
theory when one deals with states having a fixed number of particles.21
The first thing one notices, is that the vacuum, defined via (8.7), cannot be changed, viz.
U(xi , xf )( X0 ) X0 ,
20

(9.11)

In (9.5) we have omit a spacetime dependent factor which the reader may recover, using (9.4); the
vector (9.5) corresponds to the vector Xp in (9.4) in a case of s-particle state.
21
Since we are dealing with a free field, there is no interaction between its quanta (particles) and, hence,
there is no real scattering. However, the method, we present below, is of quite general nature and can be
applied in real scattering problems. This will be illustrated in a separate paper.

and if X1 (x) is a state vector of a state containing at least one particle, then
h X1 (x)| X0 i 0

(9.12)

This simple result means that the only non-forbidden transition from the vacuum is into
itself, i.e.
h X0 | X0 i = 1 6= 0.

(9.13)

The results just obtained are known as the stability of the vacuum.
In accord with (9.4), (8.5) and (7.28), the vector
1

X (x, p) = e i~ (x

x )p
0

+
0 (p)( X0 )

(9.14)

describes a particle at a point x M with 4-momentum p. Similarly, an n-particle state,

n 1, in which the ith particle is at a point xi with 4-momentum pi , i = 1, . . . , n, is
represented by22
o
n1 X
1
+
(xi x0 )(pi ) (+
X (x1 , p1 ; . . . ; xn , pn ) = exp
0 (p1 ) 0 (pn ))( X0 ). (9.15)
i~
n!
i=1
n

After some simple algebra with creation and annihilation operators, which uses (7.28), (6.7)
and (8.7c), one finds the following transition amplitude from m-particle state into n-particle
state, m, n N:
o
n1 X
1
mn exp
(xi yi )(pi )
n!
i~
n

h X (y1 , q1 ; . . . ; yn , qn )| X (x1 , p1 ; . . . ; xm , pm )i =

(i1 ,...,in )

i=1

(pn q i1 ) (pn1 qi2 ) . . . 3 (p1 qin ), (9.16)

where mn is the Kronecker -symbol, i.e. mn = 1 for m = n and mn = 0 for m 6= n, and the
summation is over all permutations (i1 , . . . , in ) of (1, . . . , n). The presence of mn in (9.16)
means that an n-particle state can be transformed only into an n-particle state; all other
transitions are forbidden. Besides, the -functions in (9.16) say that if the 4-momentum of a
particle changes, the transition is also forbidden. So, the only change an n-particle state can
experience is the change in the coordinates of the particles it contains. All these results are
quite understandable (and trivial too) since we are dealing with a free field whose quanta
move completely independent of each other, without any interactions between them.
Notice, if in (9.16)
set m = n and integrate
over all momenta, we get a pure phase
o
n we

1 Pn
factor, equal to exp i~ i=1 (xi yi )(pi ) . Since, in this case, the module of the square
of (9.16) is interpreted as a probability for the transition between the corresponding states,
this means that the transition between two n-particle states is completely sure, i.e. with
100% probability.
At the end of this section, we remark that the states (9.15) are normalized to unity,
Z
d3 q 1 . . . d3 q n h X (x1 , q1 ; . . . ; xn , qn )| X (x1 , p1 ; . . . ; xm , pm )i = 1,
(9.17)
due to (8.7d). However, the norm h X (x1 , p1 ; . . . ; xn , pn )| X (x1 , p1 ; . . . ; xm , pm )i is infinity as
it is proportional to (3 (0))n , due to (9.16). If one works with a vacuum not normalized to
22

Since the vacuum and creation/annihilation operators in Heisenberg and momentum pictures coincide,
we use the usual Fock base to expand the state vectors. The spacetime depending factor comes from (9.4).

unity, in (9.15) the factor h X0 | X0 i1/2 will appear. This will change (9.16) with the factor
h X0 | X0 i1 .

Part B
Free arbitrary scalar field
Until now the case of free Hermitian (neutral) scalar field was explored. In the present, second, part of the present investigation, the results obtained for such a field will be transferred
to the general case of free arbitrary, Hermitian (real, neutral, uncharged) or non-Hermitian
(complex, charged), scalar field.23
As we shall see, there are two essential peculiarities in the non-Hermitian case. On
one hand, the field operator and its Hermitian conjugate are so mixed in the momentum
operator that one cannot write (in momentum picture) separate field equations for them.
On the other hand, a non-Hermitian field carries a charge. These facts will later be reflected
in the corresponding commutation relations.

10.

Description of free scalar field

In Heisenberg picture a free arbitrary, Hermitian or non-Hermitian, scalar field is described

via a field operator (x),

which may be Hermitian,

(x) = (x)

(10.1a)

(x) 6= (x).

(10.1b)

or non-Hermitian,

The properties of a free non-Hermitian scalar field are, usually [35, 28], encoded in the
Lagrangian
L := m2 c4 (x) (x)

+ c2 ~2 ( (x)) ( (x)),

(10.2)

which in momentum picture, according to the general rules of Sect. 2 (see equation (2.16)),
reads
L = m2 c4 0 0 c2 [0 , P ] [0 , P ] ,

(10.3)

where (see (2.15) and cf. (3.4))

(x) = U(x, x0 ) (x)

U 1 (x, x0 ) = (x0 ) = (x
0 ) =: 0

(x) = U(x, x0 ) (x) U 1 (x, x0 ) = (x0 ) = (x0 ) =: 0

(10.4a)
(10.4b)

are the corresponding to (x)

and (x), respectively, constant field operators in momentum

picture. In general, the operator 0 can be Hermitian,
0 (x) = 0
23

(10.5a)

A classical complex field, after quantization, becomes a non-Hermitian operator acting on the systems
(fields) Hilbert space of states. That is why the quantum analogue of a classical complex scalar field is
called non-Hermitian scalar field. However, it is an accepted common practice such a field to be called
(also) a complex scalar field. In this sense, the terms complex scalar field and non-Hermitian scalar field are
equivalent and, hence, interchangeable. Besides, since a complex (classical or quantum) scalar field carries a
charge, it is called also charged scalar field. Cf. footnote 4.

or non-Hermitian,
0 6= 0 .

as a result of (10.1) and the unitarity of U(x, x0 ).

Defining
(
1 for = (Hermitian (neutral) field)
,
( )
:=
0 for 6= (non-Hermitian (charged) field)
we can unify the Lagrangians (10.2) and (3.2) by writing
1
1

L :=
m2 c4 (x) (x)

+
c2 ~2 ( (x)) ( (x)).

1 + ( )

1 + ( )

As an alternative to (10.7), one may also put

1
1

L :=
m2 c4 (x)

(x) +
c2 ~2 ( (x))

( (x))
1 + ( )

1 + ( )

(10.5b)

(10.6)

(10.7)

(10.8)

There is also one more candidate for a Lagrangian for a free arbitrary scalar field. Since
such a field is equivalent to two independent free Hermitian scalar fields
1 = 1 and
2 = 2 with masses equal to the one of and such that = 1 + i 2 , i being the

imaginary unit, we can set L = L0 ( 1 ) + L0 ( 2 ) with L0 defined by (3.2). Taking into

account that = 1 i 2 , one can transform the last Lagrangian into the form


1

m2 c4 (x) (x)

+ (x)

(x)
L :=
2(1 + ( ))

(10.9)


1
c2 ~2 ( (x)) ( (x))

+ ( (x))

( (x)) .
+
2(1 + ( ))

This Lagrangian, which is the half of the sum of (10.7) and (10.8), also reduces to (3.2)
in the Hermitian case = (or 2 = 0 in terms of the Hermitian fields 1 and 2 ).
Evidently, the Lagrangian (10.9) is a symmetrization of the r.h.s. of (10.7) or (10.8) relative
to and with coefficient 21 . The advantage of (10.9) is that in it the field and its
Hermitian conjugate enter in a symmetric way, which cannot be said relative to (10.2),
(10.7) and (10.8).
L , L,
and L can
Going some steps ahead, the consequences of the Lagrangians L,
be summarized as follows: (i) All of these Lagrangians lead to identical (Klein-Gordon)
field equations for and ; (ii) The energy-momentum, momentum and charge operators
generated by these Lagrangians are, generally, different; (iii) After the establishment of the
commutation relations and a normal ordering of products (compositions), the momentum

and L become identical and equal to one half of

and charge operators generated by L , L,
the ones induced by L. Therefore, in view of these assertions, all of the Lagrangians given
above can be considered as equivalent.24
In momentum picture, by virtue of (2.16), the Lagrangians (10.7), (10.8) and (10.9) are:
1
1

(10.10)
L =
m2 c4 0 0
c2 [0 , P ] [0 , P ]
1 + ( )
1 + ( )
1
1

L =
m2 c4 0 0
c2 [0 , P ] [0 , P ]
(10.11)
1 + ( )
1 + ( )


1

L =
m2 c4 0 0 + 0 0
2(1 + ( ))
(10.12)


1

2
c [0 , P ] [0 , P ] + [0 , P ] [0 , P ] ,

2(1 + ( ))
24

However, in Sect. 15, we shall see that the Lagrangian (10.9) carries more information than (10.7), (10.8)
and (10.2). In this sense, it is the best one.

where (cf. (10.6))

(
1
(0 ) :=
0

for 0 = 0 (Hermitian (neutral) field)

= ( ).

for 0 =
6 0 (non-Hermitian (charged) field)

(10.13)

From here, we derive:

aL
L
=
= m2 c4 0
0
0
:=

aL
= i~c2 [0 , P ]
y

aL

= m2 c4 0
0
aL
:=
= i~c2 [0 , P ]
y

(10.14)
(10.15)

1
1
where a = , , , y := i~
[0 , P ] , y := i~
[0 , P ] and we have followed the differentiation rules of classical analysis of commuting variables. At this point, all remarks, made in
Sect. 3 in a similar situation, are completely valid too. For a rigorous derivation of the field
equations (10.16) and energy-momentum tensors (10.21) below, the reader is referred to [14].
By means of the above equalities, from (2.17), we get the field equations for 0 and 0 :

m2 c2 0 [[0 , P ] , P ] = 0

m2 c2 0

[[0 ,

P ] , P ] = 0.

(10.16a)
(10.16b)

So, regardless of the Lagrangians one starts, the fields 0 and 0 satisfy one and the same
Klein-Gordon equation. However, in contrast to the Heisenberg picture, in momentum picture these equations are not independent as the momentum operator P , appearing in (10.16)
and given via (2.1) or (2.18), also depends on 0 and 0 through the energy-momentum operator T . Hence, to determine 0 , 0 and P , we need an explicit expression for T as a
function of 0 and 0 .
If was a classical complex/real field, we would have
T =

1
{
( )
+
( )} L
1 + ( )

where the means complex conjugation. The straightforward transferring of this expression
in the quantum case results in
(1)
T
=

{
( )
+
( )} L.
1 + ( )

(10.17)

However, if
and and
and do not commute, this T is non-Hermitian,

T 6= T . This situation can be corrected by a Hermitian symmetrization of the first

two terms in (10.17), which gives
(2)
T
=


1
(10.18)

( )
+ ( )
+
( ) + ( )

L.
2(1 + ( ))

Evidently, if = ,
and the Lagrangians (10.7) and (10.8) are employed, the equations (10.17) and (10.18) reduce to (3.10a) and (3.10b), respectively. But these are not
the only possibilities for the energy-momentum operator. Often (see, e.g., [3, eq. (3.34)],
or [5, eq. (2-151)], or [28, eq. (6)]), one writes it in the form
(3)
T
=

{
( )
+
( )}
L.
1 + ( )

(10.19)

As an alternative, we may also choose

(4)
T
=

{
( ) +
( )} L.
1 + ( )

(10.20)

The last two expressions are suitable as

and
are proportional to and
respectively.
Partial discussion of the problem how should be defined the quantum energy-momentum
tensorial operator T and a similar one for the current (see below), the reader can find
in [28, 2], where a reference to an early work of B. W. Gordon in Z. Phys. (vol. 40, p. 117.
1926) is given.

From the view-point of symmetry, T = T , and Hermiticity T

= T , the energy-momentum operators (10.18)(10.20) are indistinguishable (if one and the same Lagrangian is
used in them).
So, if we want to explore all possibilities, we have to look for the consequences of the four
energy-momentum operators (10.17)(10.20) for any one of the three Lagrangians (10.7)
(10.9). However, instead of investigating these 12 cases, we shall study only 3 of them. The
reason is that in [14] we have proved that for a given Lagrangian, from the Schwingers
action principle [5,29], follow, between other things, also unique expressions for all conserved
quantities in terms of the field operators. In particular, it implies unique energy-momentum
and current operators. These operators, for the Lagrangians (10.7)(10.9) will be pointed
below. Since the discussion of the way of their selection is out of the range of the present
work, the reader may: (i) take this by faith; (ii) consider it as a lucky choice; (iii) look on it
as an additional hypothesis/postulate; (iv) explore the consequences of the other 8 cases to
see that they lead to contradictions in the theory;25 (v) apply the results obtained in [14].
The correct energy-momentum operators for the Lagrangians (10.7)(10.9) are respectively (10.19), (10.20) and (10.18).26 Explicitly, in view of (10.15), we have
1

c2 ~2 {( ) ( )
+ ( ) ( )}
L
1 + ( )

(4)
c2 ~2 {( )
( ) + ( )
( )} L
T =
1 + ( )

1
(2)
T =
c2 ~2 {( ) ( )
+ ( ) ( )

2(1 + ( ))

(3)
T =

+ ( )
( ) + ( )
( )}
25

(10.21a)
(10.21b)

(10.21c)

L.

For instance, for them, generally, there are not one particle states with fixed energy, i.e. there are not
one particle eigenvectors of the zeroth component of the momentum operator.
26
Since (10.2) and (10.7) are proportional, all results for (10.7) can trivially be formulated for (10.2).

In momentum picture, these operators read (see (10.10)(10.15)):

1
c2 {[0 , P ] [0 , P ] + [0 , P ] [0 , P ] }
1 + (0 )
1
+
c2 {m2 c2 0 0 + [0 , P ] [0 , P ] }
1 + (0 )
1
(4)

T =
c2 {[0 , P ] [0 , P ] + [0 , P ] [0 , P ] }
1 + (0 )
1
+
c2 {m2 c2 0 0 + [0 , P ] [0 , P ] }
1 + (0 )
1
(2)
T =
c2 {[0 , P ] [0 , P ] + [0 , P ] [0 , P ]
2(1 + (0 ))

(3)

T =

+ [0 , P ] [0 , P ] + [0 , P ] [0 , P ] }
1
c2 {m2 c2 0 0 + m2 c2 0 0
+
2(1 + (0 ))

(10.22a)

(10.22b)

(10.22c)

+ [0 , P ] [0 , P ] + [0 , P ] [0 , P ] }.
Any one of the equations (10.22), together with (10.16), (2.18) and (2.2) form a complete
system of equations for explicit determination of , and P . It will be analyzed in the
subsequent sections.
Since the Lagrangians of a free general scalar field are invariant under (constant) phase
transformations, such a field carries a, possibly vanishing, charge (see, e.g., [3,5]. The (total)
is defined by
charge operator Q
Z

:= 1
Q
c

J0 (x) d3 x

(10.23)

x0 =const

where J (x) is a Hermitian operator,

J (x) = J (x),

(10.24)

and J are conserved quandescribing the fields current considered a little below. Since Q
tities, viz. they satisfy the equivalent conservation laws

dQ
=0
dx0

J = 0,

(10.25)

0 for a = 1, 2, 3, due to (10.23), we can write

and a Q
= 0.
Q

(10.26)

as a generator of (constant) phase transformations [3, 5] leads to the

The consideration of Q
following (Heisenberg) equations/relations27
= q
[ ,
Q]

= q
[ , Q]

(10.27)

where q is a constant, equal to the opposite charge of the particles (quanta) of (see below
Sect. 12), such that
q=0
for = .

(10.28)
27

The equations (10.27) follow from the transformation properties of and

too.

 is Hermitian, i.e.
The charge operator Q
= Q

(10.29)

and commutes with the momentum operator P ,

P ] = 0.
[ Q,

(10.30)

As a consequence of (10.30) and (2.2), the charge operator commutes with the (evolution)
operator U(x, x0 ) responsible to the transition from Heisenberg picture to momentum one,
U(x, x0 )] = 0.
[ Q,

(10.31)

=: Q,
Q(x) = Q

(10.32)

Combining this with (2.4), we get

that is the charge operators in Heisenberg and momentum pictures coincide.28

In momentum picture, the equations (10.27) and (10.29), evidently, read:
[ , Q] = q

[ , Q] = q

Q = Q.

(10.33)
(10.34)

In this picture the equality (10.23) is convenient to be rewritten as (see (2.4))

Q=

1
c

U 1 (x, x0 ) J0 (x) U(x, x0 ) d3 x.

(10.35)

x0 =x00

The only thing, we need for a complete determination of Q, is the explicit definition of
the (quantum) current J . If was a free classical arbitrary, real or complex, scalar field,
we would have
q
(x) (x)

(x) (x)).
(10.36)
J (x) = (
i~
The straightforward transferring of this result into the quantum case gives
q
J(1) (x) = {
(x) (x)

(x) (x)}.
i~

(10.37a)

But, since the current operator must satisfy (10.24), the quantities (10.37a) are not suitable
, ] 6= 0. Evidently,
for components of a current operator if [
, ]
6= 0 and/or [
here the situation is quite similar to the one with the definition of the energy-momentum
operator considered above. So, without going into details, we shall write here is a list of
three admissible candidates for a current operator:29


q

J(2) =
2i~


q
(3)

J =

i~


q
(4)

J =


.
i~

(10.37b)
(10.37c)
(10.37d)

commuting with the momentum operator.

Obviously, (10.31) and (10.32) are valid for any operator Q
For a partial discussion of the problem, see [28, 2]. The expression (10.37c) is the one most often used
in the literature [3, 11, 28]; however, the definition (10.37a) is utilized too, for instance, in [5, p. 99].
28

29

Similarly to the case of energy-momentum operator, to any one of the Lagrangians (10.7)
(10.9), there corresponds a unique current operator. These operators are as follows (see [14]):
(3)
J
(4)
J
(2)
J

q

i~


q



=
i~
q
=

+

2i~
=

(10.38a)
(10.38b)

(10.38c)

As a consequence of (10.15), these current operators in momentum picture read respectively:

(3)
J
(4)
J
(2)
J

1
= qc2 0 [0 , P ] [0 , P ] 0
2


1
= qc2 [0 , P ] 0 0 [0 , P ]
2


1
= qc2 0 [0 , P ] + [0 , P ] 0 [0 , P ] 0 0 [0 , P ] .
4

(10.39a)
(10.39b)
(10.39c)

A free scalar field has no spin angular momentum and possesses a, generally, non-vanishing
orbital angular momentum, as described in Sect. 3 (in particular, see equations (3.12)(3.16)).
It will be explored in Sect. 14 directly in terms of creation and annihilation operators.
According to the Klein-Gordon equations (10.16), the field operators 0 and 0 are
eigen-operators for the mapping (3.17) with eigenvalues equal to the square m2 of the mass
(parameter) m of the field (more precisely, of its quanta). Therefore the interpretation of
the operator (3.17) as a square-of-mass operator of the field is preserved also in the case of
free arbitrary, Hermitian of non-Hermitian, scalar field. At the same time, the square of the
momentum operator, c12 P P , has an interpretation of a square of mass operator for the
solutions of the field equations, i.e. for the fields states.

11.

Analysis of the field equations

As we know, the field operator 0 and its Hermitian conjugate 0 satisfy the Klein-Gordon
equations (10.16) which are mixed through the momentum operator P , due to the simultaneous presentation of 0 and 0 in the energy-momentum operator(s) (10.22). However,
for 0 and 0 are completely valid all of the results of Sect. 4 as in it is used only the
Klein-Gordon equation (in momentum picture for 0 ) and it does not utilize any hypotheses
about the concrete form of the momentum operator P . Let us formulate the main of them.
Proposition 11.1. The solutions 0 and 0 of (10.16) can be written as
Z





+ f (k)0 (k)
0 =
d3 k f+ (k)0 (k)
k0 = m2 c2 +k2
k0 =+ m2 c2 +k2
Z





0 =
d3 k f+ (k)0 (k)
2 + f (k)0 (k)
2 ,
2 2
2 2
k0 =+

m c +k

k0 =

m c +k

(11.1a)
(11.1b)

where 0 (k), 0 (k) : F F are solutions of

[0 (k), P ] = k 0 (k)

[0 (k), P ] = k 0 (k)

(11.2)

and, for solutions different from the degenerate solutions

[0 , P ] = 0

[0 , P ] = 0

for m = 0

(11.3)

or, in Heisenberg picture,

(x)

= (x
0 ) = 0

(x) = (x0 ) = 0

P = P = 0

for m = 0,

(11.3 )

the symbols f and f denote complex-valued functions of k and for the solutions (11.3) they
stand for some distributions of k.
p
Notice, as a result of the restriction k0 = m2 c2 + k2 in (11.1), only the solutions
of (11.2) for which
k2 = k02 k2 = m2 c2
(11.4)
are significant.
It should be emphasized, the operator 0 (k) in (11.1) is not the Hermitian conjugate of
0 (k). In fact, the reader can verify that (11.1) imply the equalities (cf. (4.12))



= f (k)0 (k)
f (k)0 (k)
2
2
2
k0 = m2 c2 +k2
k0 = m c +k
(11.5)




=
f
(k)
(k)
.
f (k)0 (k)

0
2
2
2 2
2 2
k0 =

m c +k

k0 =

m c +k

However, in the Hermitian case, i.e. for 0 = 0 , the equations (11.1a) and (11.1b) must be
identical and, consequently, we have
f (k) = f (k) 0 (k) = 0 (k)

for 0 = 0 .

(11.6)

In this case, the equations (11.5) reduce to (4.12).

Proposition 11.2. The solutions of (10.16) have the representations
Z
Z

4
2
2 2
0 = (k2 m2 c2 )0 (k) d4 k
0 = (k m c )0 (k) d k

(11.7)

where 0 (k) and 0 (k) are suitably normalized solutions of (11.2) which, up to a phase factor
1

equal to e i~ x0 k , coincide with the Fourier coefficients of 0 (x) and 0 (x) (in Heisenberg
picture for solutions different from (11.3)).
It should be emphasized, the solutions (11.3) are completely unphysical as they have
zero (energy-)momentum operator (see (10.22) and (2.18)), zero total charge (see (10.39))
and zero orbital angular momentum (see (3.14)) and, consequently, they cannot lead to some
physically predictable consequences.

12.

Frequency decompositions and their physical meaning

The presented in Sect. 5 frequency decompositions of a free Hermitian scalar field are based
on the Klein-Gordon equation, or, more precisely, on (4.6) and (4.8), and do not rely on a
concrete representation of the energy-momentum operator. Hence they can mutatis mutandis
be transferred in the general case of Hermitian or non-Hermitian scalar field. The basic
moments of that procedure are as follows.
Let us put
(
f (k)0 (k) for k0 0

0 (k) :=
0
for k0 < 0
(
(12.1)

f ((k)0 (k) for k0 0

.
0 (k) :=
0
for k0 < 0

As an evident consequence of (11.5), we have the equalities

= 0 (k)
0 (k)

0 (k) =
0 (k)

(12.2)

which mean that the operators 0 (k) are not the Hermitian conjugate of
0 (k). In the

Hermitian case, 0 = 0 , (12.1) reduce to (5.1) due to (11.6).

In view of (11.1) and (11.2), we have:

0 =

0 = +
0 + 0

d3 k
0 (k)|

k0 =

[
0 (k), P ] = k 0 (k)

m2 c2 +k2

0 = 0 + + 0
Z

0 =
d3 k0 (k)|

k0 =

[0 (k), P ] = k 0 (k)

(12.3)

k0 =

m2 c2 +k2

m2 c2 + k2

[
0 (k), Q] = q0 (k) [0 (k), Q] = q0 (k)

[
0 , Q] = q0

[0 , Q] = q0 .

(12.4)
(12.5)
(12.6)

If Xp is a state vector characterizing a state with 4-momentum p (see equation (5.6)),

then (12.5) entail
p

P (
(k)(
X
))
=
(p

k
)
(k)(
X
)
k
=
m2 c2 + k2
p

0
p
0
0
(12.7)
p
P (0 (k)( Xp )) = (p k )0 (k)( Xp )
k0 = m2 c2 + k2 .

So, +
create particles with 4-momentum k, while
annihilate such
0 and 0
0 and 0

particles. If = , the operators 0 and 0 coincide, while for 6= they are different.
In the last case the difference comes from the existence of non-zero charge operator (10.23) for
which the (Heisenberg equations/)relations (10.27) hold. If Xe is a state vector corresponding
to a state with total charge e, i.e.

Q( Xe ) = e Xe ,

(12.8)

then, from (10.33) and (12.6), we get:30

Q(0 ( Xe )) = (e q)0 ( Xe )

Q(
0 ( Xe )) = (e q)0 ( Xe )

Q(0 ( Xe )) = (e + q)0 ( Xe )

Q(0 ( Xe )) = (e + q)0 ( Xe ).

(12.9)

Q(
0 (k)( Xe )) = (e q)0 (k)( Xe ) Q(0 (k)( Xe )) = (e + q)0 (k)( Xe ).

Therefore 0 ,
and 0 (k)
0 and 0 (k) decrease the fields charge by q, while 0 , 0
+
+
increase it by the
p same quantity. So, in a summary, 0 (k) and 0 (k) create particles with

4-momentum ( m2 c2 + k2 , k) and charges (q)

p and (+q), respectively, while 0 (k) and
0 (k) annihilate particles with 4-momentum ( m2 c2 + k2 , k) and charges (+q) and (q),
respectively.31 , 32
30

These considerations do not use concrete forms, like (10.39), of the current operator J .

By convention, the particles created by +

0 or annihilated by 0 (k) are called particles, while the ones
+

created by 0 (k) or annihilated by 0 (k) are called antiparticles.

32
Since (12.9) originates from the equation (10.27), which is external to the Lagrangian formalism, one

should accept the given interpretation of

0 (k) and 0 (k) by some reserve. However, this interpretation
is confirmed in the later development of the theory on the base of a notion of n-particle, n N, states (see
Sect. 17).
31

13.

The field equations in terms of creation and

annihilation operators

If one wants to obtain from (10.16) a system of equations for the momentum and field
operators, an explicit expression for P , as a function (functional) of the field operators 0
and 0 , is required. To find it, we shall proceed as in Sect. 6, when the Hermitian case,
0 = 0 , was investigated.
Since (12.3)(12.5) imply
Z

[0 , P ] = {k (+
0 (k) + 0 (k))}|k0 = m2 c2 +k2 d k
Z
(13.1)
+

[0 , P ] = {k (0 (k) + 0 (k))}|
2 d k,
2 2
k0 =

m c +k

the energy-momentum operators (10.22), in view of (12.4), can be written as:

Z
1
2
(3)
c
d3 k d3 k {(k k k k + k k )
T =
1 + (0 )

(4)

(2)

(0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))

(13.2a)

(+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))

(13.2b)

+ m2 c2 (0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))}
Z
1
c2
=
d3 k d3 k {(k k k k + k k )
1 + (0 )

+ m2 c2 (+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))}
Z
1
c2
d3 k d3 k {(k k k k + k k )
=
2(1 + (0 ))

(0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))

+ (k k k k + k k )

+
+

(13.2c)

(+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))

m2 c2 (0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))
+

m2 c2 (+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))}

p
2
where k0 = m2 c2 + k2 and k0 = m2 c2 + k .
Performing with these expressions the same manipulations as the ones leading from (6.1)
to (6.5), we derive the following expressions for the momentum operator:33, 34
Z
1

+
3
(3)
P =
k | 2 2 2 {0 + (k)
(13.3a)
0 (k) + 0 (k) 0 (k)} d k
k0 = m c +k
1 + (0 )
Z
1

+
(4)

3
k | 2 2 2 {+
P =
(13.3b)
0 (k) 0 (k) + 0 (k) 0 (k)} d k
k0 = m c +k
1 + (0 )
Z
1

+
(2)
k | 2 2 2 {0 + (k)
P =
0 (k) + 0 (k) 0 (k)
k0 = m c +k
2(1 + (0 ))
(13.3c)

+
+

3
+ 0 (k) 0 (k) + 0 (k) 0 (k)} d k.
p

33

Notice, the equalities (6.4) remain valid in the general case. Besides, these equations hold if in them

some or all of the operators

0 (k) and 0 (k ) are replaced with 0 (k) and 0 (k ), respectively.
34
For the other 8 combinations of the Lagrangians (10.7)(10.9) and energy-momentum operators (10.17)

(10.20), in the integrands in the r.h.s. of (13.3) terms proportional to

0 (k) 0 (k0 , k) and 0 (k)

0 (k0 , k) will appear. They are responsible for the contradictions mentioned in Sect. 10 (see, in particular,
footnote 25).

Here (0 ) is defined by (10.13) and the following shortcuts are introduced:




0 (k) k0 = m2 c2 +k2

1/2 

0 (k)  2 2 2
0 (k) := 2c(2~)3 k0


0 (k) :=

2c(2~)3 k0

1/2

k0 =

m c +k

(13.4)
.

The operators +
0 (k) and 0 (k) (resp. 0 (k) and 0 (k)) are called the creation (resp.
annihilation) operators (of the field (fields particles)).
Obviously, for a Hermitian field, 0 = 0 , all of the three expressions in (13.3) reduce
to the right Hermitian result (6.5); besides, the 3-dimensional creation/annihilation operators (13.4) reduce to (6.6), as one should expect. In the non-Hermitian case, 0 6= 0 ,
the operator (13.3b), as a function of the creation/annihilation operators (13.4), formally
coincides with the momentum operator obtained from the Lagrangian (10.2) in the literature [3, eq. (3.39)]. However, it should be remarked, our creation and annihilation operators
for the Lagrangian (10.2) with 0 6= 0 differ by a phase factor from the ones in Heisenberg
picture used in the literature (see below equations (13.5)). Generally, the three momentum
operators (13.3) are different, but, after normal ordering, they result into one and the same
momentum operator (see Sect. 16, equation (16.2)).
For a comparison with expressions in (the momentum representation of) Heisenberg picture, it is worth to be noticed that, due to proposition 11.2, the creation/annihilation operators (k) and (k) in (the momentum representation of) Heisenberg picture are
(cf. (6.8))


1
(k) = e i~ x0 k 

k0 =

m2 c2 +k2


1
(k) = e i~ x0 k 

k0 =

m2 c2 +k2

0 .

(13.5)

(Relations like (6.9) are, of course, also valid; (13.5) is a consequence from them for x = x0 .)
Therefore, quadratic expressions, like the ones in the integrands in (13.3), look in one and
the same way in momentum and Heisenberg pictures.
Now we are ready to obtain the field equations in terms of creation and annihilation operators. Since (12.3)(12.5) are equivalent to (10.16), the equations, we want to derive, are (12.5)
with P given via (13.3). At this stage of the development of the theory, we will get three,
generally different, systems of equations, corresponding to the three Lagrangians, (10.7),
(10.8) and (10.9), we started off. But, as after normal ordering the three momentum operators (13.3) become identical, these systems of equations will turn to be identical after normal
ordering.
In terms of the operators (13.4), the equations (12.3)(12.5) can equivalently be rewritten
as:
Z

1/2 +
3
(0 (k) +
2c(2~)3 k0
0 =
0 (k)) d k
Z
(13.6)

1/2 +

3
3
(0 (k) + 0 (k)) d k
2c(2~) k0
0 =
p

=
k

(k)
[
(k),
P
]
=
k

(k)
k
=
m2 c2 + k2 . (13.7)
[
(k),
P
]
0

0
0

0
0
p
p
Inserting the equalities (with k0 = m2 c2 + k2 and q0 = m2 c2 + q 2 )
Z
Z

3
3
k 0 (k) = q 0 (k) (q k) d q
k 0 (k) = q 0 (k)3 (q k) d3 q

and (13.3) into (13.7), we, after some algebra, obtain the next variants of the systems of field

equations for an arbitrary scalar field in terms of creation and annihilation operators:
Z

+

+
q |q =m2 c2 +q2 [
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]
0
(13.8a)
3

3
(1 + (0 ))0 (k) (q k) d q = 0
Z


+
q |q =m2 c2 +q2 [0 (k), 0 + (q)
0 (q) + 0 (q) 0 (q)]
0
(13.8b)

(1 + (0 ))0 (k)3 (q k) d3 q = 0
Z

+
+

q |q =m2 c2 +q2 [
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]
0
3
3
(1 + (0 ))
0 (k) (q k) d q = 0
Z


q |q =m2 c2 +q2 [0 (k), +

0 (q) 0 (q) + 0 (q) 0 (q)]
0

(1 + (0 ))0 (k)3 (q k) d3 q = 0


+

+
q |q =m2 c2 +q2 [
0 (k), [0 (q), 0 (q)]+ + [0 (q), 0 (q)]+ ]
0
3
3
2(1 + (0 ))
0 (k) (q k) d q = 0
Z


+
q |q =m2 c2 +q2 [0 (k), [0 + (q),
0 (q)]+ + [0 (q), 0 (q)]+ ]
0

2(1 + (0 ))0 (k)3 (q k) d3 q = 0.

(13.9a)

(13.9b)

(13.10a)

(13.10b)

Consequently, to the Lagrangians (10.7)(10.9) correspond respectively the following

three systems of equations for

0 and 0 :
+

+
[
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]

(1 + (0 ))
0 (k) (q k) = f (k, q)

+
[0 (k), 0 + (q)
0 (q) + 0 (q) 0 (q)]

(1 + (0 ))0 (k)3 (q k) = f

+
+

[
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]

(k, q)

(1 + (0 ))
0 (k) (q k) = f (k, q)

[0 (k), +
0 (q) 0 (q) + 0 (q) 0 (q)]

(1 + (0 ))0 (k)3 (q k) = f

(k, q)

+
[
0 (k), [0 (q), 0 (q)]+ + [0 (q), 0 (q)]+ ]

2(1 + (0 ))
0 (k) (q k) = f (k, q)

+
[0 (k), [0 + (q),
0 (q)]+ + [0 (q), 0 (q)]+ ]

2(1 + (0 ))0 (k)3 (q k) = f

(k, q),

(13.11a)
(13.11b)

(13.12a)
(13.12b)

(13.13a)
(13.13b)

where the operator-valued (generalized) functions af (k, q) and af (k, q) with a = , ,

must be such that
Z
Z
a
3

(13.14)
q |q = m2 c2 +q2 f (k, q) d q = q |q =m2 c2 +q2 af (k, q) d3 q = 0.
0

Generally the systems of equations (13.11), (13.12) and (13.13) are different. They will become equivalent after normal ordering, when the equations in them will turn to be identical.
From the derivation of (13.11)(13.14), it is clear that these systems of equations are equivalent to the initial system of Klein-Gordon equations (10.16). Thus, we can say that these
systems represent the field equations in terms of creation and annihilation operators.
As a verification of the self-consistence of the theory, one can prove the commutativity
between the components of any one of the momentum operators (13.3), i.e.
[ P , P ] = 0,
(3)

(4)

(13.15)

(2)

where P = P , P , P , is a consequence of (13.3), (13.11)(13.13) and the identity (7.4) with = 1.

14.

The charge and orbital angular momentum operators

In Sect. 10, we introduced the charge operator (10.23) (see also (10.35)) and pointed to
different possible definitions of the defining it current operator. In particular, to the Lagrangians (10.7)(10.9) correspond respectively the current operators (10.39a)(10.39c) in
momentum picture. Below we shall express the charge operator Q through the creation and
annihilation operators (12.1).
Substituting (12.3), (12.4) and (13.1) into (10.39), we get:35
Z

1 2

(3)
d3 k d3 k k (0 + (k) + 0 (k)) (+
J = qc
0 (k ) + 0 (k ))
2

k (0 + (k) + 0 (k)) (+
(14.1a)
0 (k ) + 0 (k ))
Z

1
+

(4)
J = qc2 d3 k d3 k k (+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))
2
+

k (+
(14.1b)
0 (k) + 0 (k)) (0 (k ) + 0 (k ))
Z

1
(2)

J = qc2 d3 k d3 k k (0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))
4
+

+ k (+
0 (k) + 0 (k)) (0 (k ) + 0 (k ))

k (0 + (k) + 0 (k)) (+
0 (k ) + 0 (k ))
+

k (+
0 (k) + 0 (k)) (0 (k ) + 0 (k )) .

(14.1c)

Now we have to perform the following three steps: (i) insert these expressions with = 0

into (10.23); (ii) apply (6.4), possibly with 0 (k) and/or 0 (k ) for
0 (k) and/or 0 (k );

3
3
(iii) integrate over x, which gives (2~) (k + k ) for products of equal-frequency operators
and (2~)3 3 (k k ) for products of different-frequency operators. As a result of these steps,
we obtain:36
Z


(3)

(14.2a)
Q = q d3 k 0 + (k)
0 (k) 0 (k) 0 (k)
Z


(4)
= q d3 k + (k) (k) (k) + (k)
Q
(14.2b)
0
0
0
0

35
Since we apply only results based on the general properties of the momentum operator, the below obtained
equalities for the current and charge operators are independent of the concrete choice of energy-momentum
operator, like (10.17)(10.20), and, consequently, of the particular form of the momentum operator, like (13.3).
However, they depend on the Lagrangians (10.7)(10.9) from which the theory is derived.
36
When deriving (14.2), one gets the equalities in it with 1+q(0 ) for q with (0 ) defined by (10.13). Since

q = 0 for 0 = 0 and (0 ) = 0 for 0 6= 0 , we have

q
1+ (0 )

q in all cases.

 (2)

1
= q
2

+
d3 k 0 + (k)
0 (k) 0 (k) 0 (k)

+

+
+
0 (k) 0 (k) 0 (k) 0 (k) .

(14.2c)

Recall, here the 3-dimensional creation/annihilation operators are defined via (13.4). Notice, if 0 = 0 , the charges (14.2a) and (14.2b), as well as the defining them respective current operators (10.39a) and (10.39b), vanish as q = 0 in this case, while the charge (14.2c)
and the defining it current operator (10.39c) vanish for two reasons, if 0 = 0 : due to
q = 0 and due to the vanishment of the integrand in (14.2c) or the expression in parentheses
in (10.39c).
Generally, the three charge operators (14.2a)(14.2c), originating respectively from the
Lagrangians (10.7)(10.9), are different, but they become identical after normal ordering (see
Sect. 16).
Let us turn our attention now to the orbital angular momentum operator of free scalar
field. In Heisenberg picture, it is given via (3.14) To obtain an explicit expression for the
orbital angular momentum operator L (in Heisenberg picture) through the creation and

annihilation operators
0 (k) and 0 (k) (see (13.4)), we have to do the following: substitute
each of the equalities in (13.2) into the last equality in (3.14), to apply (6.4) to all terms in
the obtained equation, then to integrate over x, which results in -function terms, and, at
last, to perform the integration over k by means of the arising -functions.37 This procedure
results in:
(3)
L

37

(3)
x0 P

(3)
x0 P

Z
n


i~
+
3
d k 0 (k) k k
+
0 (k)
2(1 + (0 ))
k
k


o



0 (k) k k +
(k)
(14.5a)

0
k
k
k0 = m2 c2 +k2

The integrals, one has to calculate, are of the type

J=

, =+,

d3 x

d3 k d3 k xa U 1 (x, x0 ) {0 (k)A
(k, k ) 0 (k )} U(x, x0 )

, =+,

d3 x

i~ (x
d3 k d3 k {0 (k)A
(k, k ) s (k )}xa e

x
0 )(k + k )

(14.3)

p
p
2

where a = 1, 2, 3, A
k0 = m2 c2 + k2 , and k0 = m2 c2 + k .
(k, k ), = 1, 2, . .. , are some functions,

3

3

The integration over x results in (2~) i~ (ka + k a ) (k + k ). Simple manipulations with the re-

(y)
maining terms, by invoking the equality f (y) (y)
= fy
(y) in the form
y

3 (y z)
d y d zf (y, z)
=
(y a z a )
3

gives the following result:

J = (2~)3

, =+,

(x0 x00 )

d3 y d3 z

3 (y z)
1
f (y, z) f (z, y)
2
(y a z a )
Z


1
d3 y d3 z 3 (y z)

=
f (y, z),
2
y a
z a
1

d3 k d3 k 3 (k + k )e i~ (x

x0
0 )k0 (1+ 1)

k 
1  ka
o
1 

+ a + x0a + i~ a + a
0 (k)A
(k, k ) 0 (k ) .
2 k0
k0
2
k
k

(14.4)

The particular form of A

(k, k ) is clear from (3.14) and (10.22). So, applying several times (14.4), calculating
the appearing derivatives, and, at last, performing the trivial integration over k or k by means of 3 (k+ k ),
one can derive (14.5) after simple, but lengthy and tedious algebraic manipulations.

 (4)
L

(2)
L

= x0 P x0 P

(4)
x0 P

(2)

(4)
x0 P

Z
n


i~
+
3
+
d k 0 (k) k k 0 (k)
2(1 + (0 ))
k
k


o



0 (k) k k 0 + (k)  2 2 2 (14.5b)

k
k
k0 = m c +k

(2)

Z
n

i~

+
3
+
d k 0 (k) k k
0 (k)
4(1 + (0 ))
k
k






+
+ 0 (k) k k 0 (k) 0 (k) k k +
0 (k)
k
k
k
k

o




0 (k) k k 0 + (k)  2 2 2 . (14.5c)

k
k
k0 = m c +k

Here the derivatives with respect to k0 , like

k 0 0 (k),

must be set equal to zero, and38


 A(k) 

B(k) 
A(k)k B(k) := k

B(k)
+
A(k)

k
k
k
= k

A(k) B(k)
(14.6)
k

for operators A(k) and B(k) having C 1 dependence on k. If the operators A(k) and B(k)
tend to zero sufficiently fast at spacial infinity, then, by integration by parts, one can prove
the equality
Z

o
n



d k A(k) k k B(k)  2 2 2
k
k
k0 = m c +k
Z
n

o



=2
d3 k A(k) k k B(k)  2 2 2
k
k
k0 = m c +k
Z


o
n


3
= 2 d k
k k A(k) B(k)  2 2 2 . (14.7)
k
k
k0 = m c +k
3

By means of these equations, one can reduce (two times) the number of terms in (14.5), but
we prefer to retain the more (anti)symmetric form of the results by invoking the operation
introduced via (14.6).
Since for a neutral scalar field 0 = 0 and (0 ) = 1, in this case the three operators (14.5) reduce to
0

38

i~
= x0 P x0 P +
4



d k
k k
0 (k)
k
k

o




(k)
(14.8)
0 (k) k k +

0
k
k
k0 = m2 c2 +k2
3

+
0 (k)

More generally, if : {F F} {F F} is a mapping on the operator space over the systems Hilbert

space, we put A B := (A) B + A (B) for any A, B : F F. Usually [4, 12], this notation is used
for = .

with P given by (6.5). The reader may verify that the equality (14.8) holds for any one of
the energy-momentum operators (3.11) (or (3.10)) for the Lagrangian (3.3) (or (3.2)).
It is a trivial calculation to be shown that in terms of the operators (13.5), representing
the creation/annihilation operators in (momentum representation of) Heisenberg picture, in
the equations (14.5) and (14.8) the first two terms, proportional to the momentum operator,
should be deleted and tildes over all creation and annihilation operators must be added.
Using the explicit formulae (13.3), (14.2) and (14.5), by means of the identity (7.4), with
= 1, and the field equations (13.11)(13.13), one can verify the equations:
P ] = 0
[ Q,
[ L , P ] = i~( P P ),

(14.9)
(14.10)

denotes any one of the orbital momentum operators in (14.5) (resp.

where L (resp. Q)
charge operators in (15.21)). The first of these equalities confirms the validity of the (external
to the Lagrangian formalism) equation (10.30). We emphasize on the sign before i~ in the
r.h.s. of (14.10), which is opposite to the one usually assumed in the literature, for instance
in [5, p. 77, eq. (2-87)] or in [6, eq. (2.187)].39
From (14.9) and (2.2), we get
U(x, x0 )] = 0
[ Q,

(14.11)

Q = Q,

(14.12)

and, hence,

due to (2.4). Therefore the charge operator is one and the same in momentum and Heisenberg
pictures.
Applying (14.10), we get the orbital angular momentum of a scalar field in momentum
picture as
L = U(x, x0 ) L U 1 (x, x0 ) = L + [ U(x, x0 ), L ] U 1 (x, x0 )
= L + (x x0 ) P (x x0 ) P ,
(14.13)
due to the equality
[ L , U(x, x0 )] = {(x x0 ) P (x x0 ) P } U(x, x0 ).
which is a consequence of (14.10) and (2.2).40
tions (14.5) read:
(3)
L

(3)
x P

(3)
x P

(14.14)

Thus, in momentum picture, the equa-

Z
n


i~
+
3
d k 0 (k) k k
+
0 (k)
2(1 + (0 ))
k
k


o



0 (k) k k +
(k)
(14.15a)

0
k
k
k0 = m2 c2 +k2

39
Equation (14.10) with +i~ for i~ in its r.h.s. is part of the commutation relations for the Lie algebra of
the Poincare group see, e.g., [9, pp.143147] or [10, sec. 7.1]. However, such a change of the sign in the r.h.s.
of (14.10) contradicts to the results and the physical interpretation of the creation and annihilation operators
vide infra.
P
40

To derive equation (14.14), notice that (14.10) implies [ L , P1 Pn ] = n

i=1 i P

i P P1 Pi1 Pi+1 Pn , due to [A, B C] = [A, B] C + B [A, C] , and expand the

exponent in (2.2) into a power series. More generally, if [A(x), P ] = B (x) with [B (x), P ] = 0, then
1
(x x0 )B (x) U(x, x0 ); in particular, [A(x), P ] = 0 implies [A(x), U(x, x0 )] = 0.
[A(x), U(x, x0 )] = i~
Notice, we consider (x x0 ) as a real parameter by which the corresponding operators are multiplied and
which operators are supposed to be linear in it.

 (4)
L

(2)
L

(4)
x P

(4)
x P

Z
n


i~
3
+
+
d k 0 (k) k k 0 (k)
2(1 + (0 ))
k
k


o



0 (k) k k 0 + (k)  2 2 2 (14.15b)

k
k
k0 = m c +k

Z
n


i~
+
3
=

+
d k 0 (k) k k
0 (k)
4(1 + (0 ))
k
k






+
+ 0 (k) k k 0 (k) 0 (k) k k +
0 (k)
k
k
k
k

o




0 (k) k k 0 + (k)  2 2 2 . (14.15c)

k
k
k0 = m c +k
(2)
x P

(2)
x P

These three angular momentum operators are different but, after normal ordering, they will
be mapped into one and the same operator (see Sect. 16).

15.

The commutation relations

The trilinear systems of equations (13.11)(13.13) are similar to the (system of) Klein-Gordon
equation(s) (6.13) and, correspondingly, will be treated in an analogous way.
Since the equations (13.11b), (13.12b), and (13.13b) can be obtained from (13.11a),

a
a
(13.12a), and (13.13a) by replacing
(k, q), re0 (k) with 0 (k) and f (k, q) with f
spectively, all of the next intermediate considerations will be done only for the former set
of equations; only the more essential and final results will be doubled, i.e. written for
0 (k)

and 0 (k).
First of all, applying the identity (7.4) several times, we rewrite (13.11a), (13.12a),
and (13.13a) respectively as (recall, = 1)
+
+

[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+
+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

(15.1)

(1 + (0 ))
0 (k) (k q) = f (k, q)

+
+
[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+
+

+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

(15.2)

(1 + (0 ))
0 (k) (k q) = f (k, q)

+
+

[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+
+

+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+
+
+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+[
0 (k), 0 (q)]

+
0 (q)

2(1 +

(15.3)

+
0 (q) [
0 (k), 0 (q)]
3

(0 ))
0 (k) (k q) = f (k, q).

Now, following the know argumentation [35], we shall impose an additional condition
saying that the commutators, = 1, or anticommutators, = +1, of all combinations of
creation and/or annihilation operators should be proportional to the identity operator idF
of the Hilbert space F of the considered free arbitrary scalar field.

It is easily seen, the equations (15.1) and (15.2) (and similar ones obtained from them
with 0 (k) for
0 (k)) do not make any difference between the choices = 1 and =
+1. But, for equation (15.3), the situation is completely different. Indeed, for = +1,
which corresponds to quantization of a scalar field by anticommutators, equation (15.3)

3
reduces to 2(1 + (0 ))
0 (k q) = f (k, q) which, when inserted in (13.14), entails

k | 2 2 2 (k) = 0 for any k; a similar result follows from (13.13b), i.e. we have
k0 =

m c +k

k |

k0 =

m2 c2 +k2

0 (k) = 0

k |

k0 =

m2 c2 +k2

0 (k) = 0

for = +1

(15.4)

which, by (13.3c), implies

(2)
P

= 0.

(15.5)

Consequently, since (15.5) and the Klein-Gordon equations (10.16) imply

m2 c2 0 = 0,

m2 c2 0 = 0

(15.6)

the choice = +1 for (15.3) is possible only for the degenerate (unphysical) solutions (11.3)
(or (11.3 ) in Heisenberg picture) and for the solution
0 = 0

0 = 0

for m 6= 0,

(15.7)

the last of which is equivalent to

0 (k) = 0 0 (k) = 0

for m 6= 0.

(15.8)

According to equations (10.37) and (15.5), the degenerate solutions (11.3) and (15.7) carry
no 4-momentum and charge and, hence, cannot be detected. So, these solutions should be
interpreted as an absence of the scalar field and, if one starts from the Lagrangian (10.9),
they are the only ones that can be quantized by anticommutators.41
Let us return to the consideration of equations (15.1)(15.3) and similar ones with 0 (k)
for
0 (k) having in mind that = 1 for (15.3). Writing explicitly them for the upper, +,
and lower, , signs, we see that they can equivalently be represented respectively in the
forms:

[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

(1 +

3
(0 ))
0 (k) (k

(15.9)

q) = f (k, q)

[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

+[
0 (k), 0 (q)] 0 (q) 0 (q) [0 (k), 0 (q)]

(15.10)

(1 + (0 ))
0 (k) (k q) = f (k, q)

0 (q) [
0 (k), 0 (q)] + 0 (q) [0 (k), 0 (q)]

0 (q) [
0 (k), 0 (q)] + 0 (q) [0 (k), 0 (q)]

(15.11)

1
3
f (k, q).
(1 + (0 ))
0 (k) (k q) =
2

41
In fact, the last assertion completes the proof of spin-statistics theorem for free arbitrary scalar field.
Notice, in this proof we have not used any additional hypotheses, like charge conjugation/symmetry or
positivity of the Hilbert space metric (cf. [3, sec. 10.2]).

Let us write explicitly the above-stated additional condition concerning the (anti)commutators of creation and annihilation operators. We have (cf. equations (7.1)):

[
0 (k), 0 (q)] = a (k, q) idF

[0 (k), 0 (q)] = a (k, q) idF

[
0 (k), 0 (q)] = b (k, q) idF

[0 (k), 0 (q)] = b (k, q) idF

[
0 (k), 0 (q)] = d (k, q) idF

[0 (k),
0 (q)] = d (q, k) idF

[
0 (k), 0 (q)] = e (k, q) idF

[0 (k),
0 (q)] = e (q, k) idF

(15.12)

where = 1 for (15.1) and (15.2), = 1 for (15.3), and a

, a , . . . , e are some
complex-valued (generalized) functions, which we have to determine. These last functions
are subjected to a number of restrictions which can be derived in the same way as (7.18)
and (7.19) in the Hermitian case.42 One can easily verify that these restrictions are:

(k + q)a
(k, q) = 0 (k + q)a (k, q) = 0 (k + q)d (k, q) = 0

(15.13a)

(k q)b
(k, q) = 0 (k q)b (k, q) = 0 (k q)e (k, q) = 0
p
p



m2 c2 + k2 + m2 c2 + q 2 (k, q) = 0
for = a
, a , d
p
p



m2 c2 + k2 m2 c2 + q2 (k, q) = 0
for = b
, b , e .

(15.13b)
(15.14a)
(15.14b)

Regarding a
, a , . . . ,e as distributions, from (15.13), we derive (cf. (7.21)):

f (q)(k, q) = f (k)(k, q)

for = a
, a , d

(15.15a)

f (q)(k, q) = f (+k)(k, q)

b
, b , e

(15.15b)

for =

for any function f which is polynomial or convergent power series. In view of (15.15), the
equalities (15.14b) are identically satisfied, while (15.14a) are equivalent to the equations
p

m2 c2 + k2 (k, q) = 0
for = a
(15.16)
, a , d .

Substituting the equalities (15.12) into equations (15.9)(15.11) and similar ones with
for
0 (k), we see that the restrictions (13.14), in view of (15.15), give the next

systems of equations for the unknown (generalized) functions a

, a , . . . , e :
Z


ka
d3 q 0 (q)a
(k, q) 0 (q)d (k, q) 0 (q)b (k, q)
(15.17a)

3
+0 (q) e (k, q) (1 + (0 )) (k q) = 0
Z
p


m2 c2 + k2
d3 q 0 (q)a
(k, q) + 0 (q)d (k, q)
(15.17b)

3
0 (q)b (k, q) + 0 (q) e (k, q) (1 + (0 )) (k q) = 0
Z


d3 q 0 (q)d
ka
(q, k) 0 (q)a (k, q) + 0 (q)b (k, q)
(15.17c)

3
+0 (q) e (q, k) (1 + (0 )) (k q) = 0
Z
p


2
2
2
m c +k
d3 q 0 (q)d
(q, k) + 0 (q)a (k, q)
(15.17d)

3
+ 0 (q)b (k, q) + 0 (q) e (q, k) (1 + (0 )) (k q) = 0.
0 (k)

Notice, (7.15)(7.17) remain valid if we replace in them

0 (k) and/or 0 (q) with 0 (k) and/or 0 (q)
respectively; see (12.7).
42

Here: a = 1, 2, 3, = for (15.10) and = 1 for (15.9) and (15.11), = 1 for (15.9)
and (15.10), and = 1 for (15.11) (vide supra). Notice, (15.17a) and (15.17b) correspond
to (15.9)(15.11), while (15.17c) and (15.17d) correspond to the same equations with 0 (k)
for
0 (k).
If we impose a second, after (15.12), additional condition, namely that equation (15.17)

to be valid for arbitrary

0 (q) and 0 (q), we see that, if (m, k) 6= (0, 0), the only solution

of (15.17) relative to a
, a , . . . , e is:

a
(0 ) (k, q) = a(0 ) (k, q) = d(0 ) (k, q) = 0

(15.18a)

3
b
(0 ) (k, q) = b(0 ) (k, q) = (0 ) (k q)

(15.18b)

3
e
(0 ) (k, q) = (0 ) (k q),

where (0 ) is defined via (10.13) and:

1 for equation (15.11) with any 0

(0 ) =
and for equations (15.9) and (15.10) with 0 = 0

1 for equations (15.9) and (15.10) with 0 6= 0

(
(0 ) for equation (15.10)
.
(0 ) =
+1
for equations (15.9) and (15.11)

(15.18c)

(15.19)

Evidently, (15.18a) converts (15.16) into identity and, consequently, under the hypotheses
made, (15.18) is the general solution of our problem.
It should be emphasized on the fact that the function (0 ) in (15.18) takes care of what
is the field 0 , Hermitian or non-Hermitian, while the functions (0 ) and (0 ) take care
of from what Lagrangian, (10.7)(10.9), we have started off.
Before commenting on the solutions (15.18), we want to say some words on the case
m = 0 and k = 0 for which the equations (15.17) and (15.16) take the form of the identity
0 = 0 and, consequently, no information can be extracted from them. The above analysis
reveals that, under the additional conditions (15.12), the field equations do not impose some
restrictions on the operators

0 (k) and 0 (k)

for m = 0 and k = 0,

(15.20)

i.e. these operators must satisfy (15.12) with m = 0 and k = 0 and arbitrary a
(0, q),
(0,
q)
(with

=
1
for
(15.1)
and
(15.2)
and

=
1
for
(15.3)).
To
a (0, q), . . . , e

ensure a continuous limit (m, k) (0, 0), we shall assume by convention that a (0, q),
a (0, q), . . . , e
(0, q) are given via (15.18) with k = 0 (and m = 0). From physical
point of view (see Sect. 12), the operators (15.20) describe creation/annihilation of massless
particles with vanishing 4-momentum and charge q, which is zero for a Hermitian (neutral)
filed and non-zero for a non-Hermitian (charged) one. Consequently, in the non-Hermitian
case, the theory admits existence of free, charged, massless scalar particles with vanishing
4-momentum, which are quanta of free, charged, massless scalar field. As far as the author of
these lines knows, such particles/fields have not been observed until now. In the Hermitian
case, as we pointed in Sect. 7, the operators (15.20) reduce to (7.27) and describe unphysical
particles/fields which are experimentally unobservable.
Since the (anti)commutation relations (15.12) are extremely important for quantum field
theory, we shall write them explicitly for the obtained solutions (15.18). As the three versions
of some equations, like (15.17), (15.1)(15.3), (13.11)(13.13), (13.3), etc., originate from the
Lagrangians (10.7)(10.9) we have started off, we shall associate the found (anti)commutation

relations with the initial Lagrangians rather than with the particular equations utilized in
their derivation.
Since equations (15.11), for which (0 ) = 1 and (0 ) = +1, originate from the Lagrangian (10.9), we can assert that the Lagrangian (10.9) implies the following commutation
relations:

[
0 (k), 0 (q)] = 0

[0 (k), 0 (q)] = 0

3
[
0 (k), 0 (q)] = (0 ) (k q) idF

[0 (k), 0 (q)] = (0 )3 (k q) idF

3
[
0 (k), 0 (q)] = (k q) idF

3
[0 (k),
0 (q)] = (k q) idF

[
0 (k), 0 (q)] = 0

[0 (k),
0 (q)] = 0

(15.21)

where 0 denotes the zero operator on F and (0 ) takes care of is the field neutral (0 = 0 ,
(0 ) = 1) or charged (0 6= 0 , (0 ) = 0) and ensures a correct commutation relations in
the Hermitian case (see (7.28)).
Since the equations (15.9) are consequences of the Lagrangian (10.7), we can assert that
the next (anti)commutation relations follow from the Lagrangian (10.7):

[
0 (k), 0 (q)] = 0

[0 (k), 0 (q)] = 0

3
[
0 (k), 0 (q)] = (0 ) (k q) idF

[0 (k), 0 (q)] = (0 )3 (k q) idF

3
[
0 (k), 0 (q)] = (k q) idF

3
[0 (k),
0 (q)] = (k q) idF

[
0 (k), 0 (q)] = 0

[0 (k),
0 (q)] = 0

(15.22)

where = 1 (commutation relations) for a Hermitian filed, 0 = 0 , and = 1 (commutation or anticommutation relations) for a non-Hermitian filed, 0 6= 0 .
At last, since the Lagrangian (10.8) entails (15.10), the Lagrangian (10.8) implies the
following (anti)commutation relations:

[
0 (k), 0 (q)] = 0

[0 (k), 0 (q)] = 0

3
[
0 (k), 0 (q)] = (0 ) (k q) idF

[0 (k), 0 (q)] = (0 )3 (k q) idF

3
[
0 (k), 0 (q)] = (k q) idF

3
[0 (k),
0 (q)] = (k q) idF

[
0 (k), 0 (q)] = 0

[0 (k),
0 (q)] = 0

(15.23)

where = 1 (commutation relations) for a Hermitian filed, 0 = 0 , and = 1 (commutation or anticommutation relations) for a non-Hermitian filed, 0 6= 0 .
It should be emphasized, for a Hermitian (neutral, real) field, when = 1 in (15.22)
and (15.23), the commutation relations (15.21), (15.22), and (15.23) coincide and, due to
(0 ) = 1 in this case, are identical with (7.28); thus, they correctly reproduce the already
established results in Sect. 7. However, for a non-Hermitian (charged) field, for which (0 ) =
0, we have three independent sets of (anti)commutation relations:
(i) the commutation relations (15.21) correspond to the Lagrangians (10.7) and (10.8), with
the choice = 1 for the both ones, and the Lagrangian (10.9);
(ii) the anticommutation relations (15.22) with = +1 correspond to the Lagrangian (10.7)
with the choice = +1;
(iii) the anticommutation relations (15.23) with = +1 correspond to the Lagrangian (10.8)
with the choice = +1.

The relations (15.23) with = +1 differ from (15.22) with = +1 only in the sign
before the -function in the last row. This is quite understandable as the Lagrangian (10.8)
can be obtained from (10.7) by replacing with and with .
If we make the same

change in (15.22), i.e. 0 0 , we see that (15.22) transforms into (15.23). Since (15.22)
and (15.23) are identical (resp. different) for = 1 (resp. = +1), we conclude that
the theory is invariant (resp. non-invariant) under the change 0 0 or, equivalently,

0 0 , called charge conjugation [3, 5, 11], if and only if it is quantized via commutators
(resp. anticommutators) if one starts from any one of the Lagrangians (10.7) and (10.8). The
theory is always charge symmetric, i.e. invariant under charge conjugation, if one starts from
the Lagrangian (10.9).
Thus, for a free non-Hermitian scalar field, we see a principal difference between the
Lagrangian (10.9), on one hand, and the Lagrangians (10.7) and (10.8), on the other hand:
the first Lagrangian entails quantization with commutators, while the other two imply quantization either with commutators (identical with the one of the previous case) or with anticommutators and one needs a new additional condition/hypothesis to make a distinction
between these two cases. As it is well known, the correct quantization of a free scalar field is
via commutators, not by anticommutators [3, 5, 11]. Usually (see loc. cit.) this result is derived, for a charged field, from the Lagrangian (10.2) by invoking a new additional condition,
like charge symmetry, or positivity of the Hilbert space metric, or spin-statistics theorem.43
The above considerations show that these additional conditions are not required if one starts
from the Lagrangian (10.9); in fact, these conditions are corollaries from it, as we saw with
the charge symmetry and spin-statistics theorem (saying that a scalar field, which is a spin
zero field, must be quantized via commutators). This is not a surprising result, if we recall
that the Lagrangian (10.9) is a sum of the Lagrangians of two (independent) Hermitian scalar
fields (see Sect. 10) for which a similar result was established in Sect. 7. In conclusion, the
Lagrangian (10.9) is richer in consequences than the Lagrangians (10.7) and (10.8);44 the
cause for this is that and enter in (10.9) on equal footing, i.e. (10.9) is invariant under
the change , which cannot be said relative to (10.7) and (10.8).
Relying on the above discussion, the commutation relations (15.21) will be accepted from
now on in this paper. As we proved, under the hypotheses made, they are equivalent to
the initial system (10.16) of Klein-Gordon equations. If, by some reason one rejects these
hypotheses, the system (10.16) of Klein-Gordon equations will be equivalent to the trilinear
relations (13.11)(13.13) corresponding to the Lagrangians (10.7)(10.9). But, at present, it
seems that correct description of the real physical world is given by (15.21), not by the more
general trilinear equations mentioned, i.e. there are indications that the so-called parafields,
satisfying trilinear equations similar to (13.13), do not exist in the Nature [23].
Similarly to the said at the end of Sect. 7, the considerations in the present section
naturally lead to the operator-valued distribution character of the field variables and hence
of the creation/annihilation operators. However, such a rigorous treatment is out of the range
of this paper, in which it will be incorporated in the appearance of Diracs delta function in
some formulae.
Ending this section, we want to say that, due to the above considerations, the Lagrangian (10.9) is the best one for the correct description of arbitrary, neutral or charged,
free scalar field.
43

The particular additional conditions mentioned above are, in fact, equivalent to postulating quantization
via commutators in the case of free scalar field if one starts from someone of the Lagrangians (10.2), (10.7),
and (10.8).
44
And also the Lagrangian (10.2) which is two times (10.7) and is, usually, used in the literature.

16.

The vacuum and normal ordering

The arguments, leading to a correct definition of a vacuum and the need of normal ordering of compositions (products) of creation and/or annihilation operators, are practically the
same as in the Hermitian case, studied in Sect. 8. Without repeating them mutatis mutandis, we shall point only to the difference when the field 0 is non-Hermitian (charged).
There are two of them: (i) since in this case we have two types of annihilation operators,

viz.
0 (k) and 0 (k), the condition (8.3) should be doubled, i.e. to it one must add the
equality 0 (k)( X0 ) = 0, X0 being the vacuum (state vector); (ii) as now the field possesses a non-vanishing charge operator, the combinations of (14.2) with the commutation
relations (15.21) leads to infinities, like (8.4) for the momentum operator of a Hermitian
field.45
Thus, arguments, similar to the ones in Sect. 8, lead to the following definition of a
vacuum for a free arbitrary scalar field.
Definition 16.1. The vacuum of a free arbitrary scalar field 0 is its physical state that
contains no particles and possesses vanishing 4-momentum and (total) charge. It is described
by a state vector, denoted by X0 (in momentum picture) and called also the vacuum (of the
field), such that:
X0 6= 0
X0 = X0

0 (k)( X0 )

(16.1a)
(16.1b)
=

h X0 | X0 i = 1.

0 (k)( X0 )

=0

(16.1c)
(16.1d)

As we said above, the formulae (13.3) and (14.2), together with the commutation relations (15.21), imply senseless (infinity) values for the 4-momentum and charge of the vacuum.
They are removed by redefining the dynamical variables, like the Lagrangian, momentum operator and charge operator, by writing the compositions (products) of the field, and/or creation and/or annihilation operators in normal order, exactly in the same way as described in
Sect. 8. Besides, the definition of the normal ordering operator (mapping) N is also retained
the same as in Sect. 8, with the only remark that now it concerns all creation/annihilation

operators, i.e.
0 (k) and 0 (k).

(k) = +
(k) 0 (k) = N 0 (k) +
Since the evident equalities N +
0 (k) 0 (k)
0
0




+
+
+

and N
0 (k)0 (k) = N 0 (k)0 (k) = 0 (k)0 (k) hold, the three momentum
operators (13.3) transform, after normal ordering, into a single momentum operator, viz. into
the operator
Z
1

+
3
k | 2 2 2 {0 + (k)
P =
(16.2)
0 (k) + 0 (k) 0 (k)} d k .
k0 = m c +k
1 + (0 )
Similarly, the three charge operators (14.2) transform, after normal ordering, into a single
charge operator, viz. the operator
Q=q

+
0 + (k)
0 (k) 0 (k) 0 (k) d k .

(16.3)

45
Applying (13.3) and (15.21), the reader can easily obtain the versions of (8.4) for a non-Hermitian field.
The results will be senseless infinities, like the ones in (8.4), which are removed via normal ordering (vide
infra).

One can verify the equations

+
+
+

N
0 (k) A 0 (k) = N 0 (k) A 0 (k) = 0 (k) A as (k)

+
+
N 0 (k) A +
0 (k) = N 0 (k) A 0 (k) = as (k) A as (k),

with A = k k , as a result of which the three angular momentum operators (14.15) transform, after normal ordering, into a single orbital angular momentum operator given by

n


+
d k 0 (k) k k
0 (k)
k
k


o


+
+ 0 (k) k k 0 (k)  2 2 2 (16.4)
k
k
k0 = m c +k

i~
= x P x P +
2(1 + (0 ))

where P is given by (16.2). This equation, in Heisenberg picture and expressed via the
Heisenberg creation and annihilation operators (13.5), reads

L =

i~
2(1 + ( 0 ))



d k
k k
0 (k)
k
k


o


+
+ 0 (k) k k 0 (k)  2 2 2 . (16.5)
k
k
k0 = m c +k
3

0 + (k)

In a case of neutral (Hermitian) scalar field, when 0 = 0 and (0 ) = 1, the last expression
for the orbital angular momentum operator reproduces the one presented in [12, eq. (3.54)],
due to the first equality in (14.7).
Applying (16.4), (14.7), (13.6) and (15.21), one can verify the equations
[0 , L ] = x [0 , P ] x [0 , P ] ,

[0 , L ] = x [0 , P ] x [0 , P ]

(16.6)

which in Heisenberg picture respectively read

[ (x),

L ] = i~(x x ) (x)

[ (x), L ] = i~(x x ) (x).

(16.7)

These equations, together with (2.7), express the relativistic covariance of the theory considered [11].
Similarly, applying (16.3), (13.6) and (15.21), we obtain the equations
[0 , Q] = q0

[0 , Q] = q0 ,

(16.8)

which in Heisenberg picture take the form (10.27). Analogously, we have

[
0 (k), Q] = q0 (k)

[0 (k), Q] = q0 (k),

(16.9)

which, evidently, entail

[ P , Q] = 0 [ Q, Q] = 0 [ L , Q] = 0.

(16.10)

At last, we shall derive the commutation relations between the components of the orbital angular momentum operator (16.4), which coincides with the total angular momentum
operator. To simplify the proof and to safe some space, we shall work in Heisenberg picture and employ the Heisenberg creation and annihilation operators (13.5), which satisfy the

same commutation relations as their momentum picture

counterparts. At first, we notice
p
2
2
that (16.5) and (15.21) imply the equations (p0 := m c + p2 )
o
n
] = i~ p p

[
(p),
L
0 (p)
0
p
p
n

o
[ 0 (p), L ] = i~ p p 0 (p).
p
p

(16.11)

If ( ) denotes antisymmetrization with respect to the indices and , i.e. a subtraction

of the previous terms combined with the change and +(
0 (p)) means
0 (p)
that one has to add the previous terms by making the change
0 (p), then we
0 (p)
get by applying (16.5), (16.11) and (16.4):
Z

n + (k) (k)
0 + (k)
0
0
0 (k)
d3 k k
k

k
k

k
k
k
k
o

+ (
(k)

(k))

)
0
0
Z
n

i~

+
0 (k)
0 (k)

(k)k
= i~
d3 k 0 + (k)k
0

(1 + ( )

k
k
o
+ (
0 (k)) ( ) ( )
0 (k)




= i~ L ( ) ( ) = i~ L ( ) ( ) ,

[ L , L ] =

i~
2i~
2(1 + ( )

p
where k0 := m2 c2 + k2 in the integrands, the terms containing derivatives with respect to
k were integrated by parts and the antisymmetries relative to and and and were
taken into account. The explicit form of the result obtained is:


(16.12)
[ L , L ] = i~ L L L + L ,
which in momentum picture reads


[ L , L ] = i~ L L L + L .

(16.13)

It should be noted the minus sign in the multiplier i~ in the r.h.s. of (16.12) relative to a
similar one in the last equation in [12, eqs. (3.51)] for a neutral scalar field.
The equations (16.6)(16.13) are valid also before the normal ordering is performed, i.e.
if the orbital angular momentum, momentum and charge operators are replaced with any
one of the corresponding operators in (14.15), (13.3) and (14.2), respectively.
We would like to emphasize, equation (16.12) (or (16.13)) is a consequence of (16.5)
and (16.11), which is equivalent to (16.7), and this conclusion is independent of the validity
of the commutation relations (15.21) and/or the normal ordering (before normal ordering
equation (16.12) follows from (16.11) and (14.5)). Similar result concerns equations (16.8)
and (16.10).
So, we see that, at the very end of building of the theory of free scalar fields, all of the
three Lagrangians (10.7)(10.9) lead to one and the same final theory.46 This is a remarkable
fact which is far from evident at the beginning and all intermediate stages of the theory.
Acting with the operators (16.2) and (16.3) on the vacuum X0 , we get
46

P ( X0 ) = 0

Q( X0 ) = 0

L ( X0 ) = 0

(16.14)

Recall (see Sect. 15), the Lagrangians (10.7) and (10.8) require an additional hypothesis for the establishment of the commutation relations (15.21) and, in this sense the arising from them theory is not equivalent
to the one build from the Lagrangian (10.9).

which agrees with definition 16.1 and takes off the problem with the senseless expressions
for the 4-momentum, charge and orbital angular momentum of the vacuum before redefining
the dynamical variables via normal ordering.
As a result of the above uniqueness of the momentum operator after normal ordering,
the three systems of field equations (13.11)(13.13), together with the conditions (13.14),
transform after normal ordering into the next unique system of equations:47
+

+
[
0 (k), 0 (q) 0 (q) + 0 (q) 0 (q)]

(1 + (0 ))
0 (k) (q k) = f (k, q)

+
[0 (k), 0 + (q)
0 (q) + 0 (q) 0 (q)]

(1 + (0 ))0 (k)3 (q k) = f (k, q)

Z
Z

q |q = m2 c2 +q2 f (k, q) d q = q |q =m2 c2 +q2 f (k, q) d3 q = 0.

0

(16.15a)
(16.15b)
(16.15c)

Applying (7.4) with = 1, one can verify that (16.15) are identically valid due to the commutation relations (15.21). In this sense, we can say that the commutation relations (15.21)
play a role of field equations with respect to the creation and annihilation operators (under
the hypotheses made in their derivation).

17.

State vectors

A state vector of a free arbitrary scalar field is, of course, given via the general formula (9.1)
in which, now, the momentum operator P is given by (16.2). This means that the evolution
operator U(x, x0 ) is
n 1 x x Z
o

+
3
0
k | 2 2 2 {0 + (k)
U(x, x0 ) = exp
(k)
+

(k)

(k)}
d
k
.
0
0
0
k0 = m c +k
i~ 1 + (0 )
(17.1)
A state vector of a state with fixed 4-momentum is, of course, described by (9.4).
Similarly to the neutral field case, the amplitude, describing a transition from an initial
state Xi (xi ) to final state Xf (xf ), is (9.7) and admits the representation (9.8) through the
S-matrix U(xi , xf ). The expansion of the exponent in (17.1) into a power series results in
the following series for U(xi , xf ) (cf. (9.9) and (9.10))
U(xi , xf ) = idF +

n=1

U (n) (xi , xf )

(17.2)

1
1

(x1 xf 1 ) . . . (xi n xf n )
U (n) (xi , xf ) :=
n! i~(1 + (0 )) n i
Z
 + (1)
(1)
(1)
+ (1)
(n)

(k
)

(k
)
+

(k
)

d3 k(1) . . . d3 k(n) k(1)

0
0
0
0 (k )
n
1
 + (n)

(n)
(n)
0 (k )
) + +
) 0 (k(n) ) (17.3)
0 (k
0 (k

q
(a)
where k0 = m2 c2 + (k(a) )2 , a = 1, . . . , n.
According to (9.4) and the considerations in Sect. 12, a state vector of a state containing
n particles and n antiparticles, n , n 0, such that the i th particle has 4-momentum pi
47

The normal ordering must be applied only to the anticommutators in (13.11)(13.13) as these terms
originate from the corresponding momentum operators (13.3) before normal ordering.

and the i th antiparticle has 4-momentum pi , where i = 0, 1, . . . , n and i = 0, 1, . . . , n ,

is given by the equality
X (x; p1 ; . . . ; pn ; p1 ; . . . ; pn )

n
n
o
n1
X
X
1
1

(x x0 )
=
(pi ) + (x x0 )
(pi )
exp
i~
i~
n !n !
i =1
i =1


+
+
+
0 (p1 ) 0 (pn ) 0 (p1 ) 0 + (pn ) ( X0 ), (17.4)

where, in view of the commutation relations (15.21), the order of the creation operators is
inessential. If n = 0 (resp. n = 0), the particle (resp. antiparticle) creation operators and
the first (resp. second) sum in the exponent should be absent. In particular, the vacuum
corresponds to (17.4) with n = n = 0. The state vector (17.4) is an eigenvector of the
P
P
momentum operator (16.2) with eigenvalue (4-momentum) ni =1 pi + ni =1 pi and is also
an eigenvector of the charge operator (16.3) with eigenvalue (q)(n n ).48
The reader may verify, using (15.21) and (12.2), that the transition amplitude between
two states of a charged field, like (17.4), is:
h X (y; q1 ; . . . ; qn ; q1 ; . . . ; qn )| X (x; p1 ; . . . ; pm ; p1 ; . . . ; pm )i

n
n
n1
o
X
X
1
1

(x y )
(pi ) + (x y )
(pi )
= m n m n exp
n !n !
i~
i~
i =1
i =1
X
3 (pn q i )3 (pn 1 q i ) . . . 3 (p1 q i )

(i1 ,...,in )

(i
1 ,...,in )

3 (pn q i )3 (pn 1 q i ) . . . 3 (p1 q i ) (17.5)

1

where the summations are over all permutations (i1 , . . . , in ) of (1, . . . , n ) and (i1 , . . . , in ) of
(1, . . . , n ). The conclusions from this formula are similar to the ones from (9.16) in Sect. 9.
For instance, the only non-forbidden transition from an n -particle + n -antiparticle state
is into n -particle + n -antiparticle state; the both states may differ only in the spacetime
positions of the (anti)particles in them. This result is quite natural as we are dealing with
free particles/fields.
In particular, if Xn denotes any state containing n particles and/or antiparticles, n =
0, 1, . . . , then (17.5) says that
h Xn | X0 i = n0 ,
(17.6)
which expresses the stability of the vacuum.
We shall end the present section with a simple example. Consider the one (anti)particle
+
states p
+
0 (p)( X0 ) and 0 (p)( X0 ). Applying (16.2), (16.3), (16.4) and (15.21), we find
(p0 := m2 c2 + p2 ):49




+
+
P +
Q +
0 (p)( X0 ) = p 0 (p)( X0 )
0 (p)( X0 ) = q0 (p)( X0 )
(17.7)




P 0 + (p)( X0 ) = p 0 + (p)( X0 ) Q 0 + (p)( X0 ) = +q0 + (p)( X0 )
Recall (see Sect. 12), the operator +
0 (k) creates a particle with 4-momentum k and charge
p q, while
creates a particle with 4-momentum k and charge +q, where, in the both cases, k0 = m2 c2 + k 2 .
In Heisenberg picture and in terms of the Heisenberg creation/annihilation operators, in equations (17.8)
the terms proportional to (x p x p ) are absent and tildes over all operators should be added. Equations (17.7) remain unchanged in Heisenberg picture (in terms of the corresponding Heisenberg operators).
48

0 + (k)
49





n
o +

p
0 (p)( X0 )
L (x) +
(p)(
X
)
=
(x
p

x
p
)

i~
p

0
p
p
n


o +
L (x) 0 + (p)( X0 ) = (x p x p ) i~ p
p
0 (p)( X0 ) .
p
p

(17.8)

+
These results agree completely with the interpretation of +
0 (k) and 0 (k) as operators
creating one (anti)particle states.

18.

Conclusion

The main results of this paper, dealing with a study of free Hermitian or non-Hermitian
scalar fields, may be formulated as follows:
The creation and annihilation operators in momentum representation in momentum picture
are introduced without an explicit appeal to the Fourier transform of the field operator(s). However, they are (up to constant phase factor and, possibly, normalization)
identical with the known ones introduced in momentum representation in Heisenberg
picture.
The quantization with commutators, not by anticommutators, is derived from the field
equations (in momentum picture) without involving the spin-statistics theorem (or
other additional condition), if one stars from a suitable Lagrangian.
The (system of) field equation(s) in terms of creation and annihilation operators is derived.
It happens to be similar to a kind of paracommutation relations.50
An analysis of the derivation of the standard commutation relations is given. It is shown
that, under some explicitly presented conditions, they are equivalent to the (system of)
field equation(s) and are not additional to it conditions in the theory.
In forthcoming paper(s), we intend to investigate other free fields, like vector and spinor
ones, in momentum picture.

References
[1] Bozhidar Z. Iliev. Pictures and equations of motion in Lagrangian quantum field theory.
In Frank Columbus, editor, Progress in Mathematical Physics, pages ???? Nova Science
Publishers, Inc., Suite, 2004. To appear.
http://www.arXiv.org e-Print archive, E-print No. hep-th/0302002, February 2003.
[2] Bozhidar Z. Iliev. Momentum picture of motion in Lagrangian quantum field theory.
http://www.arXiv.org e-Print archive, E-print No. hep-th/0311003, November 2003.
[3] N. N. Bogolyubov and D. V. Shirkov. Introduction to the theory of quantized fields.
Nauka, Moscow, third edition, 1976. In Russian. English translation: Wiley, New York,
1980.
[4] J. D. Bjorken and S. D. Drell. Relativistic quantum mechanics, volume 1 and 2. McGrawHill Book Company, New York, 1964, 1965. Russian translation: Nauka, Moscow, 1978.
50
More precisely, elsewhere we shall show how the parabose-commutation relations for free arbitrary scalar
field can be derived from the Klein-Gordon equations in terms of creation and annihilation operators.

[5] Paul Roman. Introduction to quantum field theory. John Wiley&Sons, Inc., New YorkLondon-Sydney-Toronto, 1969.
[6] Lewis H. Ryder. Quantum field theory. Cambridge Univ. Press, Cambridge, 1985.
Russian translation: Mir, Moscow, 1987.
[7] A. I. Akhiezer and V. B. Berestetskii. Quantum electrodynamics. Nauka, Moscow,
1969. In Russian. English translation: Authorized English ed., rev. and enl. by the
author, Translated from the 2d Russian ed. by G.M. Volkoff, New York, Interscience
Publishers, 1965. Other English translations: New York, Consultants Bureau, 1957;
London, Oldbourne Press, 1964, 1962.
[8] Pierre Ramond. Field theory: a modern primer, volume 51 of Frontiers in physics. Reading, MA Benjamin-Cummings, London-Amsterdam-Don Mills, Ontario-Sidney-Tokio, 1
edition, 1981. 2nd rev. print, Frontiers in physics vol. 74, Adison Wesley Publ. Co.,
Redwood city, CA, 1989; Russian translation from the first ed.: Moscow, Mir 1984.
[9] N. N. Bogolubov, A. A. Logunov, and I. T. Todorov. Introduction to axiomatic quantum
field theory. W. A. Benjamin, Inc., London, 1975. Translation from Russian: Nauka,
Moscow, 1969.
[10] N. N. Bogolubov, A. A. Logunov, A. I. Oksak, and I. T. Todorov. General principles of
quantum field theory. Nauka, Moscow, 1987. In Russian. English translation: Kluwer
Academic Publishers, Dordrecht, 1989.
[11] J. D. Bjorken and S. D. Drell. Relativistic quantum fields, volume 2. McGraw-Hill Book
Company, New York, 1965. Russian translation: Nauka, Moscow, 1978.
[12] C. Itzykson and J.-B. Zuber. Quantum field theory. McGraw-Hill Book Company, New
York, 1980. Russian translation (in two volumes): Mir, Moscow, 1984.
[13] Silvan S. Schweber. An introduction to relativistic quantum field theory. Row, Peterson and Co., Evanston, Ill., Elmsford, N.Y., 1961. Russian translation: IL (Foreign
Literature Pub.), Moscow, 1963.
[14] Bozhidar Z. Iliev. On operator differentiation in the action principle in quantum field theory. In Stancho Dimiev and Kouei Sekigava, editors, Proceedings of the 6th International
Workshop on Complex Structures and Vector Fields, 36 September 2002, St. Knstantin
resort (near Varna), Bulgaria, Trends in Complex Analysis, Differential Geometry and
Mathematical Physics, pages 76107. World Scientific, New Jersey-London-SingaporeHong Kong, 2003.
http://www.arXiv.org e-Print archive, E-print No. hep-th/0204003, April 2002.
[15] A. M. L. Messiah. Quantum mechanics, volume I and II. Interscience, New York, 1958.
Russian translation: Nauka, Moscow, 1978 (vol. I) and 1979 (vol. II).
[16] P. A. M. Dirac. The principles of quantum mechanics. Oxford at the Clarendon Press,
Oxford, fourth edition, 1958. Russian translation in: P. Dirac, Principles of quantum
mechanics, Moscow, Nauka, 1979.
[17] E. Prugovecki. Quantum mechanics in Hilbert space, volume 92 of Pure and applied
mathematics. Academic Press, New York-London, second edition, 1981.
[18] Bozhidar Z. Iliev. On momentum operator in quantum field theory.
http://www.arXiv.org e-Print archive, E-print No. hep-th/0206008, June 2002, June
2002.

[19] H. S. Green. A generalized method of field quantization. Phys. Rev., 90(2):270273,

1953.
[20] D. V. Volkov. On quantization of half-integer spin fields. Zh. Eksperim. i Teor. Fiz.
(Journal of experimental and theoretical physics), 36(5):15601566, 1959. In Russian.
English translation: Soviet Phys.JETP, vol. 9, p. 1107, 1959.
[21] D. V. Volkov. S-matrix in the generalized quantization method. Zh. Eksperim. i Teor.
Fiz. (Journal of experimental and theoretical physics), 38(2):518523, 1960. In Russian.
English translation: Soviet Phys.JETP, vol. 11, p. 375, 1960.
[22] O. W. Greenberg and A. M. I. Messiah. Symmetrization postulate and its experimental
foundation. Phys. Rev., 136B(1B):248267, 1964.
[23] O. W. Greenberg and A. M. I. Messiah. Selection rules for parafields and the absence
of para particles in nature. Phys. Rev., 138B(5B):11551167, 1965.
[24] R. F. Streater and A. S. Wightman. PCT, spin and statistics and all that. W. A.
Benjamin, Inc., New York-Amsterdam, 1964. Russian translation: Nauka, Moscow,
1966.
[25] Res Jost. The general theory of quantized fields. American Mathematical Society, Rhode
Island, 1965. Russian translation: Mir, Moscow, 1967.
[26] A. B. Govorkov. Parastatistics and internal symmetries. In N. N. Bogolyubov, editor,
Physics of elementary particles and atomic nuclei, volume 14, No. 5, of Particles and
nuclei, pages 12291272. Energoatomizdat, Moscow, 1983. In Russian.
[27] G. C. Wick. Calculation of scattering matrix. Physical Review, 80(2):268, 1950. Russian
translation in [30, pp. 245253].

[28] W. Pauli and V. Weisskopf. Uber

die Quantizierung der skalaren relativistischen Wellengleichung (On the quantization of the scalar relativistic wave equation). Helv. Phys.
Acta, 7:709731, 1934. In German. Russian translation in [31, pp. 194214].
[29] J. Schwinger. The theory of quantized fields. I. Physical Review, 82(6):914, 1951. Russian
translation in [30, pp. 115137].
[30] Novejshee razvitie kvantovoj electrodinamiki. Sbornik statej (The newest development of
quantum electrodynamics. Collection of papers), editor Ivanenko D. D., Moscow, 1954.
IL (Foreign Literature Pub.). In Russian.
[31] Wolfgang Pauli. Works on quantum theory. Articles 19281958. Nauka Publ., Moscow,
1977. In Russian.