Symmetry Factors of Feynman Diagrams

arXiv:0907.0859v2 [hep-ph] 15 Nov 2010

for Scalar Fields

P. V. Dong∗, L. T. Hue†, H. T. Hung‡, H. N. Long§, and N. H. Thao¶
Institute of Physics, VAST, P.O. Box 429, Bo Ho, Hanoi 10000, Vietnam
November 16, 2010

Abstract
We calculate the symmetry factors of diagrams for real and complex scalar
fields in general form using an analysis of the Wick expansion for Greens func-
tions. We separate two classes of symmetry factors: factors corresponding to
connected diagrams and factors corresponding to vacuum diagrams. The sym-
metry factors of vacuum diagrams play an important role in constructing the
effective action and phase transitions in cosmology. In the complex scalar field
theory, diagrams with different topologies can contribute the same, and the
inverse symmetry factor for the total contribution is therefore the sum of the
inverse symmetry factors, i.e., 1/S = i (1/Si ).
P

PACS number(s): 11.15.Bt, 12.39.St.

Keywords: General properties of perturbation theory, Factorization.

1 Introduction
In quantum field theory, physical processes are described by the elements of the S-
matrix, which are in turn given by Feynman diagrams. One important task in calculat-
ing these diagrams is determining their symmetry factors (see, e.g., [1]). Fortunately,
∗
pvdong@iop.vast.ac.vn
†
lthue@grad.iop.vast.ac.vn
‡
hthung@grad.iop.vast.ac.vn
§
hnlong@iop.vast.ac.vn
¶
nhthao@grad.iop.vast.ac.vn

1
there are now various convenient computer programs (for instance, FeynArts [2] or
QGRAF [3]) for constructing Feynman diagrams in different field theories. We note
that QGRAF does not work with vacuum diagrams, which play an important role
in effective theories. In a series of papers, Kastening and coauthors [4] developed an
alternative systematic approach for constructing all Feynman diagrams based on con-
sidering a Feynman diagram as a functional of its graphical elements. We stress that
only real fields were considered in all these papers, and complex fields were outside
the scop
Our aim here is to derive a general formula describing the case of complex scalar
fields (it, of course, would also hold in the case of real fields). This formula turns
out to be easily understood and is therefore very useful for those physicists who have
not developed good skills in computer disciplines. Below, we show that the case of
complex fields has very special features that are absent in the case of real fields.
We return to our questions. What is the symmetry factor? How is it constructed?
We consider a pth-order expansion of the n-point correlation function in a real scalar
theory with the interaction Lint = (λ/4!)φ4 :

(1/p!)(1/4!)p h0|T [φ(x1)φ(x2 )...φ(xn )φ4 (y1 )φ4 (y2 )...φ4 (yp )]|0i, (1)

where the factor (iλ)p and integrations over y1 , y2 , ..., yp are omitted because they are
always presumed in the Feynman rules. Our task is to count the number of different
contractions giving the same expression (corresponding to a Feynman diagram) [5].
This number is equal to N/D, the number of all possible contractions divided by the
number of identical contractions. The overall constant of the diagram then becomes
S −1 ≡ (1/p!)(1/4!)pN/D. The number S, called the symmetry factor of the diagram,
generally differs from unity. Further, the numerator N is a product of p! interchanges
of the vertices y1 , y2 , ..., yp and Ni self-contractions of the vertex yi (i = 1, 2, ..., p)
and placements of contractions into this vertex. The value of Ni is 4! if there is no self-
contraction, 4!/2 if there is one self-contraction (single bubble), and 4!/8 if there are
two self-contractions (double bubble). Hence, N = p! i Ni = [p!(4!)p ]/[2s 8d ], where s
Q

and d are the respective numbers of single-bubble and double-bubble vertices. Because
a double bubble contains two single bubbles, the total number of single bubbles is
β = s + 2d. We can rewrite N = [p!(4!)p ]/[2β 2d ].
In contrast, determining the denominator D is not so easy. Briefly, we evaluate
it as follows. First, we consider the interchange of vertexvertex contractions. If
there are n contractions, then we have n! interchanges. Second, we consider the
interchange of the vertices y1 , ..., yp giving identical contractions, i.e., an identical set
of Feynman propagators. In this case, there are d! interchanges of d-type vertices

2
times g ′ interchanges of the remaining vertices. The result is D = g ′d! n=2,3... (n!)αn ,
Q

where αn is the number of vertex pairs with n contractions. The symmetry factor is
given by
S = g ′d! (n!)αn 2d 2β .
Y
(2)
n

Determining g ′ is nontrivial [6] and sometimes leads to significant problems. In

the literature, only the symmetry factors in the real scalar field theory with connected
diagrams are presented; those for the vacuum diagrams and also those for the complex
fields are absent. We note that vacuum diagrams have applications in particle physics
and cosmology, such as in the effective action and phase transition (see, e.g., [1], [5]
and also [7] for a recent implication). A formula for calculating the symmetry factors
of such diagrams is needed; our aim here is to explicitly derive such a formula for the
symmetry factor in the real scalar theory. We do this by applying Wick’s theorem
and in the process show that the vacuum diagrams are factored [8] explicitly order by
perturbation theory order. We also study its generalizations to complex scalar fields.
We also list symmetry factors corresponding to Feynman diagrams in both theories.
This work is organized as follows. In Sec. 2, we present some notation. In Sec. 3,
we formulate the symmetry factor for the real scalar theory. In Sec. 4, we generalize
the formula to complex fields and also consider the special features existing only in
the complex theory. In Sec. 5, we summarize our results and draw conclusions. The
appendix is devoted to Feynman diagrams and the corresponding symmetry factors
in both theories.

2 Notation
We recall some ingredients of the S-matrix approach. The time-evolution operator is
given in terms of the action as [9]

U(t1 , t2 ) = T exp[iSint (t1 , t2 , ϕ̂)]

( ! )
1 δ δ
= N exp ∆ exp[iSint (t1 , t2 , ϕ)] |... , (3)
2 δϕ δϕ

where symbol |... indicates that after differentiation, the classical fields ϕi are replaced
with the quantized ones ϕ̂i and T and N denote the time-ordering and normal-ordering
operators. The S-matrix is the limit of the time-evolution operator as t1 → −∞ and
t2 → +∞. The c-number function ∆(x, x′ ) (Feynman propagator) is defined as

∆(x, x′ ) = T [ϕ̂(x)ϕ̂(x′ )] − N[ϕ̂(x)ϕ̂(x′ )]. (4)

3
The formula [9]
   
n n
( )
Y 1X δ
 δ X δ δ Y 
T Fi (ϕ̂) = N exp  ∆ + ∆ Fi (ϕi ) |... (5)
i=1
 2 i δϕi δϕi i<k δϕi δϕk i=1 

is useful for our further presentation. We note that the first term in the right-hand
side of (5) is present only in the real field theory.
We recall that every Feynman diagram, as mentioned in the introduction, has a
symmetry factor. In [1, 10], it has the form given by

S = g2β (n!)αn ,
Y
(6)
n=2,3...

where αn is the number of pairs of vertices connected by n identical self-conjugate

lines, β is the number of lines connecting a vertex to itself, and g is the number of
permutations of vertices that leave the diagram unchanged with fixed external lines.
We note that the factor 2β comes from the factor 1/2 in the first term in the r.h.s
of (5). We also note that formula (6) works only for connected diagrams but not for
vacuum diagrams. We derive the symmetry factor in the general case as sketched in
(2) for the real φ4 theory.

3 Symmetry factors in real scalar theory

We consider the model with the interaction Lagrangian
λ 4
Lrint = φ. (7)
4!
It is well known that there is a direct connection between the S-matrix elements and
the Green’s functions defined by the expansion G(x1 , x2 , ..., xn ) = ∞ (p)
p=0 G (x1 , x2 , ..., xn )
P

where the pth-order term has the form

ip ∞ 4
Z
G(p) (x1 , x2 , ..., xn ) = d y1 ...d4 yp h0|T [φ(x1 )...φ(xn )
p! −∞
Lrint (φ(y1 ))...Lrint (φ(yp )]|0i. (8)

This term is called the pth-order Green’s function. The full Green’s function G(x1 , ..., xn )
contains every n–point diagram in the theory, both connected and disconnected.

4
 We recall that the four fields in Lrint (φ(y)) are taken at equal times. Applying (5)
for Lagrangian (7), we obtain
˙ + 3∆
φ4 (y) ∼ T [φ4 (y)] = N[φ4 (y)] + 6N[φ2 (y)]∆ ˙ ∆,
˙ (9)

where ∆˙ ≡ ∆(y, y) denotes the bubble diagram . We let a, b and c denote the three
terms in (9)
˙
a ≡ N[φ4 (y)], b ≡ N[φ2 (y)]∆, c≡∆ ˙ ∆.
˙ (10)
Then we can rewrite (9) as

φ4 ∼ T [φ4 ] = a + 6b + 3c. (11)

Green’s function (8) is invariant under permutations of the interaction Lagrangians.

Hence, the product of these Lagrangians can be expanded as a sum of monomials in
a, b and c such that all terms ap bq ct with given p, q and t are equivalent under inte-
gration. The overall coefficients of the monomials in the expansion can be extracted
using the multinomial formula
p!
(x1 + x2 + · · · + xr )p = xp11 · · · xpr r ,
X
(12)
p1 ,p2 ,...,pr p1 !p2 ! · · · pr !
with p1 + p2 + p3 + · · · + pr = p.

Equation (8) then becomes

!p
1 iλ p! ∞
Z
(p)
d4 y1 ...d4 yp
X
G (x1 , x2 , ..., xn ) =
p! 4! p1 +p2 +p3 =p p1 !p2 !p3 ! −∞

×h0|T [φ(x1 )...φ(xn )a (6b) (3c)p3 ]|0i,

p1 p2
(13)

where the variables in the integrand have the clear meaning

ap1 bp2 cp3 = a(y1 )a(y2 )...a(yp1 )b(yp1 +1 )b(yp1 +2 )...b(yp1 +p2 )c(yp1 +p2 +1 )c(yp1 +p2 +2 )...c(yp ).

For the further presentation, we omit the summations and integrations and repre-
sent the coefficients of b and c by
4! 4!
6= , 3= (14)
2!2! 2!2!2!
The Green’s function can then be rewritten in the form

G(n) (x1 , x2 , ..., xn ) = (iλ)p AB, (15)

5
where
(4!)p2 (4!)p3
A ≡ , (16)
(4!)(p1 +p2 +p3 ) (2!)p2 (2!)p2 (2!)p3 (2!)p3 (2!)p3 p1 !p2 !p3 !
B ≡ h0|T [φ(x1)...φ(xn )ap1 bp2 cp3 ]|0i (17)

We note that the b associated with p2 contains one bubble diagram while the c as-
sociated with p3 , contains two, a double bubble . Hence if we let the β be the
number of lines that connect a vertex to itself, then

β = p2 + 2p3 . (18)

Moreover, these bubbles can be factored out of the T -product in B such that the
T –operator does not act on them:
˙ p2 ∆
B = h0|T [φ(x1 )...φ(xn )(N(φ4 ))p1 (N(φ2 ))p2 ]|0i∆ ˙ 2p3 , (19)

where the double bubbles (as disconnected pieces) are vacuum subdiagrams. We also
note that p2 and p3 simply coincide with the corresponding s and d in the introduction.
The corresponding coefficient A is interpreted as
" #" #
1 1
A= . (20)
(4!) 1 (2!)p2 p1 !p2 !
p 2 (2!)p3 p3 !
β

In this formula, p1 ! and p2 ! are respective numbers of permutations of a and b vertices,

similar to p! in (8). The 4! (powered p1 ) and 2! (powered p2 ) are symmetry factors (the
number of permutations of identical interaction-fields) respectively associated with a
and b, similar to 4! in (7). In total, we obtain the factor p1 !p2 !(4!)p1 (2!)p2 , which
is deduced as the first factor in (20). This factor can be simplified if we use the T -
product expansion for B. The second factor, associated with the bubbles subdiagrams,
is unchanged under T -product: p3 ! is the number of permutations of c vertices; 2!
(powered p3 ) is the number of permutations of the two single bubbles of any c vertex;
β was described above.
Next, to contract B under the T -product, we refer to Eq. (4.45) in [5]. The
number of different contractions that give the same expression is the product of four
types of factors. First, we have p1 !p2 ! interchanges of p1 a and p2 b vertices. Second,
for the placement of contractions into a vertex, we have 4! for a and 2! for b vertices
and therefore (4!)p1 (2!)p2 for p1 a and p2 b vertices (we note that there is no self-
contraction for each vertex). Third, we have 1/ n=2,3... (n!)αn interchanges of vertex-
Q

vertex contractions, where n is the number of contractions and αn is the number of

6
vertex pairs with n contractions. Finally, if we let g ′ be the number of interchanges
of a and b vertices that do not change the diagram topologically, then the factor 1/g ′
should be multiplied to the result. In summary, the total factor contributing to one
diagram is
p1 !p2 !(4!)p1 (2!)p2 1
′ α
A= ′ (21)
g n (n!) (g p3 !)2 (2!)p3 n (n!)αn
β
Q Q
n

Hence, the symmetry factor is given by

S = g2β (2!)d (n!)αn ,

Y
(22)
n

where d = p3 , and g = g ′p3 ! has the same meaning as g ′. We note that any vertex
of a and b directly connected to the external points x1 , x2 , ..., xn is not subject to the
interchanges defining g ′ . The examples in [6] and the followings examples demonstrate
this.
The constructed diagram typically consists of connected pieces (subdiagrams), a
piece connected to x1 , x2 ,..., xn and several pieces disconnected from all the external
points, vacuum bubbles, in which the double bubble is one of the cases. We let Vc
denote the connected piece and Vk denote the various possible disconnected pieces:

...
Vk ∈ , , ,

where k = 1, 2, 3... We suppose that the diagram has nk pieces of the form Vk for each
k in addition to Vc . Let the value of g for the connected piece Vc and disconnected
pieces Vk be gc and gk . It is easy to obtain g = l nl !(gl )nl , where l = c, k and nc = 1,
Q

n1 = p3 . Here, nk ! is the symmetry factor coming from interchanging the nk copies of

Vk . We can therefore rewrite (22) as

nl !(Sl )nl = Sc × nk !(Sk )nk ,

Y Y
S= (23)
l k

where Sl = Sc , Sk is the symmetry factor of Vl having the same form as (22):

l
Sl = gl 2βl (2!)dl (n!)αn ,
Y
(24)
n

where the parameters indexed by l are those of Vl satisfying dl=1 = 1, dl6=1 = 0,

d = l nl dl , β = l nl βl and αn = l nl αnl . We note that there is an additional factor
P P P

7
2! associated with only double bubble. This contradicts formula (6), which is given
in the literature.
In calculating, we note that the symmetry factor of arbitrary diagrams is obtained
from (22) or (23) while that of connected diagrams is given by (24). Because (22)
and (24) have the same form, we can commonly use (22) for both the cases with
the corresponding interpretation of the parameters. The symmetry factors of some
two-point connected diagrams are

S = 2 (g = 1, β = 1, d = 0, αn = 0)

S = 6 (g = 1, β = 0, d = 0, α3 = 1)

S = 4 (g = 1, β = 2, d = 0, αn = 0)

S = 8 (g = 1, β = 3, d = 0, αn = 0)

S = 8 (g = 1, β = 2, d = 0, α2 = 1)

S = 8 (g = 2, β = 2, d = 0, αn = 0)

For some vacuum bubbles, we also have

8
 S = 8 (g = 1, β = 2, d = 1, αn = 0)

S = 16 (g = 2, β = 2, d = 0, α2 = 1)

S = 48 (g = 2, β = 0, d = 0, α4 = 1)

S = 24 (g = 2, β = 1, d = 0, α3 = 1)

For general diagrams, we consider the example

S = 2.2!(8)2 = 256, using (23).

Alternatively, S = 256 (g = 2, β = 5, d = 2, αn = 0), from (22).

More examples of symmetry factors are given in the following sections. In what
follows, if some parameter has its trivial value (such us g = 1 or β = 0), then that
parameter is not listed in parentheses. We next consider the case of complex scalar
fields.

4 Symmetry factors in complex scalar theory

The interaction Lagrangian in the complex scalar theory is
ρ
Lcint = (ϕ∗ ϕ)2 (25)
4
Applying (5), we obtain
˙ + 2∆
(ϕ∗ ϕ)2 ∼ T [(ϕ∗ ϕ)2 ] = N[(ϕ∗ ϕ)2 ] + 4N(ϕ∗ ϕ)∆ ˙ ∆,
˙ (26)

9
where ∆˙ in this case denotes the bubble diagram with arrow . As before, we let
a, b and c denote the corresponding terms. The pth-order Green’s function is
G(p) (x1 , x2 , ..., xn ) = (iρ)p Ac h0|T [ϕ(x1 )...ϕ∗ (xn )ap1 bp2 cp3 ]|0i, (27)
where the integrations and summations are understood and
1
Ac ≡ . (28)
4p1 2p3 p1 !p2 !p3 !
We note that the Green’s function is nonzero only if the number of fields ϕ in (27) is
equal to the number of their complex-conjugate fields ϕ∗ .
Repeating the previous analysis, we obtain the contribution for one diagram
p1 !p2 !4p1 1
′ α
Ac = ′
. (29)
g n (n!) n (g p3 !)2 3 n (n!)αn
p
Q Q

Hence, the symmetry factor in the theory under consideration is given by

S = g2d (n!)αn ,
Y
(30)
n

where d = p3 is the number of double bubbles, and g = g ′ p3 ! is the number of

interchanges of interacting vertices leaving both the diagram and its charged scalar
flows unchanged. As before, we can separate the symmetry factor into subfactors
corresponding to connected and vacuum subdiagrams:
S = Sc × Sv . (31)
The symmetry factor for these subdiagrams has the same form as (30), where d is
nonzero only if it is associated with a double bubble.
We emphasize that there is no factor 2β in (30). We note that n is the number of
identical lines connecting two separated vertices with the same direction. Formula (30)
is simply a generalization of (22) discriminating between the scalar field directions.
We illustrate this with the examples

(a) S = 1 (b) S = 2 (α2 = 1) (c) S = 8 (g = 2, α2 = 2)

10
In the diagram (a) the symmetry factor is 1 because β is zero. In (b), we have only
one set n = 2 and in (c), we have two sets with n = 2. We recall that in the real scalar
theory, we have n = 3 and n = 4 for the corresponding diagrams. Many comparisons
of symmetry factors of third-order diagrams in the real and complex scalar theories
are given in the appendix.
It follows from Eq.(31) that the vacuum diagrams are factored order by perturba-
tion theory order. Hence, the connected Green’s functions, as in the literature, can
be defined by the formula

h0|T [ϕ(x1 ) · · · ϕ(xn ) exp i d4 yLint]|0i

R
< 0|T [ϕ(x1 ) · · · ϕ(xn )]|0 >c = , (32)
h0|T exp i d4 yLint |0i
R

where the vacuum diagrams are contained in the denominator.

We next discuss some special properties of the complex theory. We consider two
contributions with the symmetry factors

S1 = 6 (g = 3!) S2 = 24 (g = 3, α2 = 3)

It is easy to verify that these contributions coincide because ∆(x, y) = ∆(y, x) [11].
Hence, contributions of this type can be determined by only one diagram with the
symmetry factor given by
S −1 = S1−1 + S2−1 , (33)
and therefore S = 24/5.
We note the recently proposed hybrid inflationary scenario [12] in which there are
two scalar fields φ and ϕ with the coupling
λ 2 2
(φ ϕ ). (34)
2
It is easily to verify that our formula is applicable to such interactions.

11
5 Conclusion
We have derived the symmetry factor for both the real and the complex scalar theories:

S = g2β 2d (n!)αn ,
Y
(35)
n

where g is the number of interchanges of vertices leaving the diagram topologically

unchanged, β is the number of lines connecting a vertex to itself (β is zero if the field
is complex), d is the number of double bubbles, and αn is the number of vertex pairs
connected by n-identical lines. Our result revises the usual symmetry factor formula
in the literature. Our result is easily generalized to higher-spin fields.
We have also showed that in the complex scalar theory, diagrams with different
topologies can contribute the same. We also obtained the symmetry factor for con-
tributions of such type.
It is easy to verify that our results are consistent with the symmetry factors in [4].
Our result explicitly shows that the vacuum diagrams, as expected, are factored
order by perturbation theory order.
We recall that determining the symmetry factor is important because it not only is
an important component of modern quantum field theory but also is used to calculate
the effective potential in higher-dimensional theories and cosmological models.

Acknowledgments
This work was supported in part by the National Foundation for Science and Tech-
nology Development (NAFOSTED) under grant No: 103.01.16.09.

References
[1] T. P. Cheng and L. F. Li, Gauge Theory of Elementary Particle Physics, Claren-
don Press (2004).

[2] J. Külbeck, M. Böhm, and A. Denner, Comput. Phys. Com-

mmun. 60, 165 (1991); T. Haln, [arXiv:9905354(hep-ph)];
http://www-itp.physik.uni-karlsruhe.de/feynarts.

[3] P. Nogueira, J. Comput. Phys. 105, 279 (1993); ftp://gtae2.ist.utl.pt/pub/qgraf

12
 [4] B. Kastening, Phys. Rev. D 57, 3567 (1998); Phys. Rev. E 61, 3501 (2000); H.
Kleinert, A. Pelster, B. Kastening, and M. Bachmann, Phys. Rev. E 62, 1537
(2000).

[5] M. E. Peskin and D. V. Schroeder, An Introduction to Quantum Field Theory,

Westview Press (1995).

[6] C. D. Palmer and M. E. Carrington, Can. J. Phys. 80, 847 (2002).

[7] Do T. Huong and Hoang N. Long, “Inflationary Scenario in the Supersymmetric

Economical 3-3-1 Model ”, [arXiv:0807.2346(hep-ph)], Phys. Atom. Nucl., 73, 791
(2010).

[8] L. H. Ryder, Quantum field theory, 2nd edition, Cambridge University Press,
(1998).

[9] A. N. Vasil’ev, Functional methods in Quantum Field Theory and Statistics,

Leningrad University Press, (1976).

[10] Michio Kaku, Quantum Field Theory, A Modern Introduction, Oxford University
Press (1993).

[11] W. Greiner and J. Reinhardt, Field Quantization, Springer (1996), p. 112.

[12] A. D. Linde, Phys. Lett. B 259, 38 (1991).

A Symmetry factors for three-order diagrams in

the real and the complex scalar theories
Diagrams of the real scalar theory are given on the left below. The corresponding
diagrams of the complex scalar theory are given on the right.

13
S = 3072 (g = 3!, β = 6, d = 3) S = 48 (g = 3!, d = 3)

S = 256 (g = 2, β = 5, d = 2) S = 8 (g = 2, d = 2)

S = 12 (β = 1, α3 = 1) S = 2 (α2 = 1)

14
 S = 128 (g = 2, β = 4, d = 1, α2 = 1) S = 4 (g = 2, d = 1)

S = 32 (β = 4, d = 1) S = 2 (d = 1)

S = 32 (β = 3, d = 1, α2 = 1) S = 2 (d = 1)

S = 32 (g = 2, β = 3, α2 = 1) S = 2 (g = 2)
15
 S = 48 (g = 3!, β = 3) S = 3 (g = 3)

S = 8 (β = 3) S=1

S = 32 (g = 2, β = 2, α2 = 2) S = 2 (g = 2)

S = 8 (β = 2, α2 = 1) S=1

16
 S = 8 (g = 2, β = 2) S=1

S = 384 (g = 2, β = 2, d = 1, α4 = 1) S = 16 (g = 2, d = 1, α2 = 2)

S = 48 (β = 2, d = 1, α3 = 1) S = 4 (d = 1, α2 = 1)

S = 24 (g = 2, β = 1, α3 = 1) S = 2 (α2 = 1)

17
 S = 8 (β = 1, α2 = 2) S=1

S = 96 (g = 2, β = 1, α4 = 1) S = 8 (g = 2, α2 = 2)

S = 4 (β = 1, α2 = 1) S = 2 (α2 = 1)

S = 12 (g = 2, α3 = 1) S = 2 (α2 = 1)

18
S = 48 (g = 3!, α2 = 3) S = 24 (g = 3, α2 = 3)

S = 6 (g = 3!)

S = 4 (α2 = 2) S=1

S = 4 (α2 = 2)

19