TTP12-012

SFB/CPP-12-23

Three-loop β-functions for top-Yukawa and

the Higgs self-interaction in the Standard
Model
K. G. Chetyrkin a , M. F. Zoller a
a
Institut für Theoretische Teilchenphysik, Karlsruhe Institute of Technology (KIT), D-76128
Karlsruhe, Germany

Abstract

We analytically compute the dominant contributions to the β-functions for the top-Yukawa
coupling, the strong coupling and the Higgs self-coupling as well as the anomalous dimensions
of the scalar, gluon and quark fields in the unbroken phase of the Standard Model at three-
loop level. These are mainly the QCD and top-Yukawa corrections. The contributions from
the Higgs self-interaction which are negligible for the running of the top-Yukawa and the
strong coupling but important for the running of the Higgs self-coupling are also evaluated.

1 Introduction

Using perturbation theory in any renormalizable quantum field theory comes with the price
that the parameters of this theory, e.g. the couplings and masses, in general depend on the
renormalization scale µ. The precise description of the evolution of these parameters with the
energy scale is an important task in any model. This is done by means of the Renormalization
Group functions, that is the β-functions and anomalous dimensions. The knowledge of the
three-loop contributions to the β-functions for the Standard Model (SM) and its extensions
is important for physics at the very high energy frontier and for cosmology. Here are some
examples: The running of the gauge couplings plays an important role for the construction
of Grand Unified theories of the strong and electroweak interactions. The β-functions for
the scalar self-interaction and for the top quark Yukawa coupling constant are important for
the analysis of Higgs-inflation in the SM [1, 2, 3, 4]. The current investigations of these issues
are based on the two-loop approximation. The inclusion of the next order could be essential.
Further, in a recent work [5], the possibility has been discussed that the SM, supplemented
by the asymptotically safe gravity could play the role of a fundamental, rather than effective
field theory. Within this framework the mass of the Higgs boson has been predicted to be
approximately 126 GeV. The theoretical uncertainty of the prediction is about 2 GeV.
Recent exciting evidence from several SM-like Higgs search channels at both the CERN
Large Hadron Collider and the Fermilab Tevatron [6, 7, 8] point to the possibility of a SM
Higgs boson with a mass in the vicinity of 125 GeV which is in truly remarkable agreement
with the aforementioned prediction. This calls for more precise calculations, in particular of
β-functions, in the SM.
In the present paper we are particularly interested in the evolution of the Higgs self-coupling

1
as well as the top-Yukawa coupling in the SM.
The underlying gauge group of the SM is an SUC (3) × SU(2) × U Y (1) which is spontaneously
broken to SUC (3) × UQ (1) at the electroweak scale. As the renormalization constants for
fields and vertices do not depend on masses and external momenta in the MS-scheme, we will
perform our calculations in the unbroken phase of the SM.
The most important contributions to the running of the Higgs self-coupling λ arise from
the top-Yukawa coupling and the strong sector. All other Yukawa couplings are signifi-
cantly smaller due to to the smallness of the respective quark masses.
√ MFrom the top mass
Mt ≈ 172.9 GeV we get the Yukawa coupling at this scale yt (Mt ) = 2 v t ≈ 1
where v ≈ 246.2 GeV is proportional to the vacuum expectation value of the scalar field Φ
from which results the Higgs field after the spontaneous symmetry breaking: |hΦi| = √v2 .
√
The next Yukawa coupling to be considered would be yb = 2 Mvb ≈ 0.02. The strong cou-
pling at the√ scale of the Z boson mass is√ s
g (MZ ) ≈ 1.22 whereas the electroweak couplings
4πα 4πα
g1 (MZ ) = cos θW ≈ 0.36 and g2 (MZ ) = sin θW ≈ 0.65 give much smaller contributions which
are further suppressed by the isospin and hypercharge factors. For this reason we will con-
sider a simplified version of the SM or - as one could also see it - a minimal extension of QCD
by setting g1 = g2 = 0 in our calculation. For a Higgs mass of 125 GeV the value of the Higgs
self-interaction would be λ(MH ) =≈ 0.13 at the scale of the Higgs mass. The relevance of
this parameter will be examined in section 2.
The outline of the work is as follows. In the next section we discuss the main definitions
and the general setup of our work. Section 3 deals with the technical details, including the
treatment of γ5 . In sections 4 and 5 we present our results for the β-functions of the top-
Yukawa, the strong and the Higgs self-couplings and the relevant field anomalous dimensions.
The numerical influence of the computed three-loop corrections on the evolution of the quartic
Higgs coupling is discussed in section 6. For this analysis we will include the already known
contributions with g1 and g2 at one-loop and two-loop level. Finally, section 7 contains our
conclusions and acknowledgements.
All our results for β-functions and anomalous dimensions can be retrieved from
http://www-ttp.particle.uni-karlsruhe.de/Progdata/ttp12/ttp12-012/

2 General Setup

The Lagrangian of our model consists of three pieces:

L = LQCD + Lyt + LΦ . (1)
The QCD part is defined by
1 1
LQCD = − Gaµν Ga µν − (∂µ Aa µ )2 + ∂µ c̄a ∂ µ ca + gs f abc ∂µ c̄a Ab µ cc
4 2(1 − ξ)
Xi ← →  , (2)
a a
+ / / T q
q̄ ∂ q + gs q̄ A
q 2

where q runs over all quark flavours, the gluon field strength tensor is given by
Gaµν = ∂µ Aaν − ∂ν Aaµ + gs f abc Abµ Acν (3)

2
and f abc are the structure constants of the colour gauge group with the generators T a :
h i
T a , T b = if abc T c . (4)

The complex scalar field Φ and the left-handed parts of the top and bottom quarks tL and
bL are doublets under SU(2):
! !
Φ1 t
Φ= , QL = . (5)
Φ2 b
L

Setting all Yukawa couplings to zero except for the top coupling yt the Lagrangian for the
Yukawa sector is given by
n o
Lyt = −yt t̄R Φ† c QL + Q̄L Φc tR
( ! ! )
t   Φ∗2
= −yt t̄R (Φ2 , −Φ1 ) · + t̄, b̄ · tR (6)
b L −Φ∗1
L
n

 
o
= −yt t̄PR t Φ∗2 + t̄PL t Φ2 − b̄PR t Φ∗1 − t̄PL b Φ1 .

Finally, we have the scalar sector of the model

 2
LΦ = ∂µ Φ† ∂ µ Φ − m2 Φ† Φ − λ Φ† Φ . (7)

The indices L and R indicate the left- and right-handed part of the fields as obtained by the
projectors
1 1
PL = (1 − γ5 ) PR = (1 + γ5 ) . (8)
2 2
This model is renormalized with the counterterm Lagrangians
1 (2g)  2 1 (3g)  
δLQCD = − δZ3 ∂µ Aaν − ∂ν Aaµ − δZ1 gs f abc ∂µ Aaν − ∂ν Aaµ Abµ Acν
4 2
1 (4g) 2  abc b c 2 (2c) (ccg)
− δZ1 gs f Aµ Aν + δZ3 ∂µ c̄a ∂ µ ca + δZ1 gs f abc ∂µ c̄a Ab µ cc (9)
4 
X i ← → h (2q) i h i 
(2q) a a (qqg) (qqg)
+ / /
q̄ ∂ δZ2,L PL + δZ2,R PR q + gs q̄ A T δZ1,L PL + δZ1,R PR q
q 2

for the QCD part,

n   o
(tbΦ)



δLY ukawa = −δZ1 yt t̄PR t Φ∗2 + t̄PL t Φ2 − b̄PR t Φ∗1 − t̄PL b Φ1 (10)

for the Yukawa part and

 2
(2Φ) (4Φ)
δLΦ = δZ2 ∂µ Φ† ∂ µ Φ − m2 δZΦ2 Φ† Φ + δZ1 Φ† Φ (11)

for the Φ-sector. Note that in general the left- and right-handed parts of quark fields and
quark-vertices are renormalized differently.1 So the renormalization constant for the strong
gauge coupling gs can be obtained for example from
(ttg) (ttg)
Z1,L Z1,R
Zgs = q = q (12)
(2t) (2g) (2t) (2g)
Z2,L Z3 Z2,R Z3
1
In our case this is true for the quark fields participating in the Yukawa sector. As we do not consider the
electroweak interaction here and neglect all Yukawa couplings except for yt the light quark fields u,d,s and c
have the same renormalization constant for the left- and right-handed part.

3
or the renormalization constant for yt from
(tbΦ)
Z1
Zyt = q . (13)
(2t) (2t) (2Φ)
Z2,L Z2,R Z2

Here the renormalization constants have been defined in a minimal way as

Z = 1 + δZ, (14)

with δZ containing only poles in the regulating parameter ε = (4 − D)/2 of the dimensional
regularization and D being the engineering space-time dimension. The Higgs self-coupling λ
is related to the Higgs mass at tree level MH via

MH2
λ= , (15)
2v 2
which for MH = 125 GeV yields λ(MH = 125GeV) ≈ 0.13. For the running of the top-
Yukawa coupling the contribution from λ is negligible compared to the top-Yukawa and strong
coupling. The corresponding four-Φ vertex is nevertheless needed for the renormalization at
three-loop level, namely to kill the subdivergence from the fermion loop in diagrams like Fig.
1 (d). The β-function for a coupling X is defined as
∞
dX X 1 (n)
βX = µ2 2
= β
2 )n X
(16)
dµ n=1
(16π

and is given as a power series in all couplings of the model, namely gs , yt and λ. Note that
the β-functions βgs and βyt are proportional to gs and yt respectively whereas βλ has one
part proportional to λ and one part proportional to yt4 with no λ-dependence at all. The
anomalous dimension of a field f is defined as
∞
dlnZf−1 X 1 f (n)
γ2f = −µ2 2
= γ
2 )n 2
, (17)
dµ n=1
(16π

where Zf is the field strength renormalization constant for the respective field.2 The β-
functions for all couplings are independent of the gauge parameter ξ whereas the anomalous
dimensions of the fields are not.

3 Calculation

As we are interested in β-functions it is enough to compute the UV divergent part of all

diagrams in order to determine the necessary renormalization constants. In the MS-scheme
the latter depends only polynomially on external momenta and masses. Therefore most of our
renormalization constants could be computed from massless propagator-like diagrams using
the FORM 3 [9] version3 of the package MINCER [10]. In some diagrams, e.g. with four
2
For an n-point vertex V or a mass m the anomalous dimension is defined as γnV = −µ2 dlnZ
dµ2
V
or
2 dlnZm
γm = −µ dµ2 . For a field we take the inverse renormalization constant.
3
The program can be downloaded from
http://www.nikhef.nl/˜ form/maindir/packages/mincer/mincer.html

4
external Φ-fields where two external momenta are set to zero, this leads to IR divergences
which mix with the UV ones in dimensional regularization. Another convenient method to
compute renormalization constants has been suggested in [11] and elaborated in the context
of three-loop calculations in [12]. The idea is an exact decomposition of all propagators using
an auxiliary mass parameter M 2 :
1 1 −p2 − 2q·p − M 2 1
2
= 2 2
+ 2 2
, (18)
(q + p) q −M q −M (q + p)2
where p is a combination of external and q of internal momenta. This can be done recursively
until the power in the denominator of the last term is high enough for this contribution to
be finite, e.g.
1 1 −p2 − 2q·p (−p2 − 2q·p)2 M2
= + + −
(q + p)2 q 2 − M 2 (q 2 − M 2 )2 (q 2 − M 2 )3 (q 2 − M 2 )2
(19)
M 2 (M 2 + 2p2 + 4q·p) (−p2 − 2q·p − M 2 )3 1
+ + .
(q 2 − M 2 )3 (q 2 − M 2 )3 (q + p)2
As the result is independent of M 2 we can omit the contributions ∼ M 2 in the above decom-
position as long as we introduce counterterms into the Lagrangian to cancel M 2 -dependent
2 (2g) 2 (2Φ)
subdivergences. In our case only a term M2 δZM 2 Aaµ Aa µ and M2 δZM 2 Φ† Φ are possible.4
The first one is not gauge invariant but this does not matter as it is only used for the can-
cellation of subdivergences which works nevertheless. This method effectively amounts to
introducing the same auxiliary mass parameter M 2 in every denominator of propagators and
all possible M 2 -counterterms. We expand in the external momenta5 and arrive at massive
tadpole diagrams with one scale M . Due to the auxiliary mass no IR divergences can ap-
pear while the UV counterterms which we are interested in (that is the ones without any
dependence on the auxiliary mass M) will stay untouched.
For the calculation of massive tadpoles we have used the FORM-based program MATAD [13].
Where possible, i.e. for the propagators and three point functions, we have employed both the
MINCER and the MATAD setups which served as an extra check. To generate the diagrams
we used QGRAF [14] and to compute the colour factors the FORM package COLOR [15].
An important aspect of calculations such as these is the proper treatment of γ5 . As is well
known, a naive treatment of γ5 can be applied if it only appears in an external fermion line.
In fermion loops we have to be more careful. In four dimensions we define
i
γ5 = iγ 0 γ 1 γ 2 γ 3 = εµνρσ γ µ γ ν γ ρ γ σ with ε0123 = 1 = −ε0123 . (20)
4!
In Fig. 1 to 5 we show a few diagrams that had to be calculated for the various ingredients
of our final result. In order to have a contribution from a fermion loop with one γ5 in it
at least four free Lorentz indices or momenta on the external lines of the minimal subgraph
containing this fermion loop are required. These can be indices from the gluon vertices or the
internal momenta from other loops which act as external momenta to the minimal subgraph
containing the fermion loop in question. External momenta of the whole diagram can be set
to zero as the renormalization constants in the MS-scheme do not depend on those.
4
Counterterms ∼ M that would arise for fermions cannot appear because we have no M in the numerators
2 (2c)
of propagators. The ghost mass term M2 δZM 2 c̄a ca does not appear because of the momentum dependence
of the ghost-gluon-vertex.
5
This method also works in massive theories. In this case we expand in the physical masses as well.

5
Consider for example one of the fermion loops in Fig.2 (c). The momenta on the two external
Φ-legs can be set to zero. Then we have two indices from the gluon lines attached to our
fermion loop and one loop momentum going through the two gluons and acting as an external
momentum to the subgraph containing only our fermion loop. This is not enough to have a
non-naive γ5 contribution from this graph.
For this reason diagrams like Fig.1 (a,b,c,d), Fig.2 (a,b,c), Fig.3 (a,b), Fig.4 (a,b,c), Fig.5
(a,b) can be treated naively. Fig.3 (c) has enough indices and momenta but no γ5 in it.
Diagrams like Fig.1 (c,e), Fig.2 (b), Fig.3 (b), Fig.4 (a,c), Fig.5 (a,c) are zero because of an
odd number of γ-matrices in at least one fermion loop. And diagrams like Fig.1 (b), Fig.2 (b),
Fig.3 (a), Fig.4 (b), Fig.5 (b) are zero because of their colour structure. The only problematic
type is Fig.1 (f) which fortunately only contributes a 1ε pole and can therefore be treated as
described in [16]. We use the fact that γ5 anticommutes with every other γ-matrix in four
dimensions and that γ52 = 1. Then we apply relation (20) for the case when one γ5 remains
on each fermion line. The two εµνρσ can be rewritten as a combination of metric tensors
which can be handled in dimensional regularization. The error we make with this treatment
is of order ε and does therefore not affect the pole part of our result.

Φ Φ Φ
(a) (b) (c)

t t t t t t
Φ Φ Φ
(d) (e) (f)

t t t t t t
(tbΦ)
Figure 1: Some diagrams contributing to Z1

4 Results for the β-functions

First we give the results for the three-loop β-functions of couplings λ, yt and gs with the gen-
eral gauge group factors for the strong interacting sector. Below CF and CA are the quadratic
Casimir operators of the quark and the adjoint representation of the corresponding Lie  alge-
bra, dR is the dimension of the quark representation, TF is defined so that TF δ = Tr T a T b
ab

6
 (a) (b) (c)
Φ Φ Φ Φ Φ Φ

Φ Φ Φ Φ Φ Φ
(4Φ)
Figure 2: Some diagrams contributing to Z1

(a) (b) (c)

g g g g g g

(2g)
Figure 3: Some diagrams contributing to Z3

(a) (b) (c)

Φ Φ Φ Φ Φ Φ

(2Φ)
Figure 4: Some diagrams contributing to Z2

(a) (b) (c)

t t t t t t
(2t)
Figure 5: Some diagrams contributing to Z2,L/R

is the trace of two group generators of the quark representation.6 For QCD (colour gauge
group SU(3)) we have CF = 4/3 , CA = 3 , TF = 1/2 and dR = 3. Furthermore we denote the

6 1


For an SU(N ) gauge group these are dR = N , CA = 2TF N and CF = TF N − N
.

7
number of fermions by nf = nl + 1.
(1)
βλ =12 λ2 + 2dR yt2 λ − dR yt4 ,
(2) 1
βλ = − 156 λ3 − 24dR yt2 λ2 − dR yt4 λ + 5dR yt6
2
+ 10CF dR gs2 yt2 λ − 4CF dR gs2 yt4 ,
 
(3) 789
βλ =λ4 (3588 + 2016ζ3 ) + 291dR yt2 λ3 + yt4 λ2 dR + 252ζ3 dR − 36d2R
2
   
1881 13 195 2
+ yt6 λ − dR − 66ζ3 dR + 80d2R + yt8 dR − 12ζ3 dR − dR
8 2 8
  . (21)
2 2 2 2 4 895
+ gs yt λ (−306CF dR + 288ζ3 CF dR ) + gs yt λ CF dR − 324ζ3 CF dR
4
  
19 119 2
+ gs2 yt6 − CF dR + 60ζ3 CF dR + gs4 yt2 λ − C dR + 77CA CF dR
2 2 F

 131 2
−16nf TF CF dR + 72ζ3 CF2 dR − 36ζ3 CA CF dR + gs4 yt4 C dR
2 F

109 2
+48TF CF dR − CA CF dR + 10nf TF CF dR − 48ζ3 CF dR + 24ζ3 CA CF dR
2
The purely λ-dependent parts of eq. (21) have been known for a while [17, 18], the full one-
loop and two-loop result are in agreement with [19, 20] (for an SU(3) colour gauge group).
It has also been a useful check for our setup to see that the same result can be derived from
the four-Φ1 vertex, the four-Φ2 vertex and the (Φ∗1 Φ∗2 Φ1 Φ2 ) vertex.

βy(1)
 
3 1
t
=yt2 + dR − 3CF gs2 ,
yt 4 2
βy(2)
 
t 2 2 4 3 9
=3 λ − 6 yt λ + yt − dR
yt 4 4
   
2 2 5 4 3 2 97 10
+ gs yt 6CF + CF dR + gs − CF − CA CF + nf TF CF ,
2 2 6 3
(3)
βyt
   
285 45 63 45
= − 18 λ3 + yt2 λ2 − dR + yt4 λ + dR
yt 8 4 2 2
 
345 9 107 3 39
+ yt6 − + ζ3 + dR + ζ3 dR + d2R
32 4 32 2 16 (22)
 
57 81
+ 6CF gs2 yt2 λ − gs2 yt4 CF + CF dR
2 8

471 119 717 77
+ gs4 yt2 CF2 − CF2 dR + 25TF CF + CA CF + CA CF dR
16 8 16 4
33 27
− nf TF CF − 4nf TF CF dR − 27ζ3 CF2 + 18ζ3 CF2 dR − ζ3 CA CF
4  2
6 129 3 129 2 11413 2
−9ζ3 CA CF dR ) + gs − C + CA CF − C CF + 46nf TF CF2
2 F 4 108 A

556 140 2 2
+ nf CA TF CF + nf TF CF − 48ζ3 nf TF CF2 + 48ζ3 nf CA TF CF .
27 27
The one-loop and two-loop part of this result have been found before in [19, 21] (for dR = 3,
TF = 21 ) and the contributions of order gs2 , gs4 , yt2 gs2 and gs4 yt to Zyt have been successfully
checked against [22]. In this reference the calculation of ZMt has been performed in the

8
broken phase of the SM with a massive top quark. When comparing these two results one
has to take into account that in the broken SM the top quark mass is to be renormalized as
a product yt (Φ2 + Φ†2 ) so the corresponding top quark mass renormalization constant is
(tbΦ)
(2Φ) Z
ZMt = Zyt Z2 =q 1 .
(2t) (2t)
Z2,L Z2,R

Again the setup could be checked for consistency by using the renormalization of the four
different vertices t-t-Φ2 , t-t-Φ∗2 , t-b-Φ1 , t-b-Φ∗1 for the calculation.
βg(1)
 
11 2
s
=gs2 − CA + nf TF ,
gs 6 3
(2)
βgs
 
2 2 4 17 2 10
= − 2TF gs yt + gs − CA + 2nf TF CF + nf CA TF ,
gs 3 3
(3)
βgs
 
9 7 (23)
= + gs2 yt4 TF + TF dR − gs4 yt2 (3TF CF + 12CA TF )
gs 2 2

2857 3 205 1415
+ gs6 − C − nf TF CF2 + nf CA TF CF + nf CA2 TF
108 A 18 54

22 2 2 79 2 2
− nf TF CF − nf CA TF .
9 27

In the case of yt = 0 this is in agreement with the well-known result [23, 24]. The one-loop
and two-loop parts of eq. (23) are known from [25, 26, 27, 28, 29, 30], the term ∝ gs4 yt2 can be
found in [22] and the full three-loop result has been computed in [31] (for dR = 3, TF = 12 ).
The β-function describing the running of the “mass” parameter m2 in eq. (7) can be computed
from the renormalization constant of the local operator O2Φ := Φ† Φ. An insertion of O2Φ into
a Green’s function, e.g. with two external Φ-fields, is renormalized as [O2Φ ] = ZΦ2 O2Φ where
[O2Φ ] is the corresponding finite operator. From [O2Φ ] = Zm2 O2Φbare and O bare = Z (2Φ) O
2Φ 2 2Φ it
follows that  
(2Φ) −1
Zm2 = Z2 ZΦ2 . (24)
This yields the following contributions to βm2 :
(1)
βm2
=6 λ + dR yt2 ,
m2
(2)
βm2 9
2
= − 30 λ2 − 12dR yt2 λ − dR yt4 + 5CF dR gs2 yt2 ,
m 4
(3)  
βm2 3 99 2 2 4 333 2
=1026 λ + d y
R t λ + y λ dR − 18d + 72ζ d
3 R
m2 2 t
4 R
 
617 (25)
+ yt6 − dR + 24d2R + 15ζ3 dR
16
+ gs2 yt2 λ (−153CF dR + 144ζ3 CF dR )
 
447
+ gs2 yt4 CF dR − 90ζ3 CF dR
8
 
119 2 77
+ gs4 yt2 − CF dR + CA CF dR − 8nf TF CF dR + 36ζ3 CF2 dR − 18ζ3 CA CF dR
4 2
.

9
The one-loop and two-loop parts of this result are in agreement with [19] where they have
been computed before. The purely λ-dependent part can be found in [17, 18]. For dR = 3
and TF = 12 (QCD) we get the following results:
(1)
βλ =12 λ2 + 6 yt2 λ − 3 yt4 ,
(2) 3 4
βλ = − 156 λ3 − 72 yt2 λ2 − y λ + 15 yt6 + 40 gs2 yt2 λ − 16 gs2 yt4 ,
2 t  
(3) 1719
βλ =λ4 (3588 + 2016ζ3 ) + 873 yt2 λ3 + yt4 λ2 + 756ζ3
2
    (26)
117 1599
+ yt6 λ − 198ζ3 − yt8 + 36ζ3 + gs2 yt2 λ2 (−1224 + 1152ζ3 )
8 8
2 4 2 6
+ gs yt λ (895 − 1296ζ3 ) + gs yt (−38 + 240ζ3 )
   
4 2 1820 4 4 626
+ gs yt λ − 32nf − 48ζ3 + gs yt − + 20nf + 32ζ3 .
3 3
To get an idea of the size of these contributions and therefore the significance of our calculation
we evaluate βλ at the scale µ = MZ (with an assumed Higgs mass of 125 GeV and nf = 6)
which yields a value of βλ ∼ (−0.01) at one-loop level. The two and three-loop contributions
change this result by ∼ 1% and ∼ (−0.04)% respectively. To estimate the importance of the
individual terms we introduce the labels
gs yt λ
G= , Y = , L= (27)
gs (µ = MZ ) yt (µ = MZ ) λ(µ = MZ )
and get
   

βλ |µ=MZ = |−1.7 Y 4  10−2 + 5.0

{z } | {zLY 2} |+1.5 L2  10−3
{z }
1 loop 1 loop 1 loop
 

+ −8.5
|
G2 Y 4} |+5.0
{z
Y 6 +3.1 G
{z } | {z
2
LY 2} 10−4
2 loop 2 loop 2 loop

2 6
+ 7.9
| G {z Y } −4.8 Y 8 −5.3{zL2 Y 2} |−3.1 G
| {z } | {z
2
LY 4}
3 loop 3 loop 2 loop 3 loop
 (28)
−2.5
|
G4 Y 4} +2.6
{z |
G
{z
4
LY 2} |−1.7 L3  10−5
{z }
3 loop 3 loop 2 loop
 

+ −7.5 LY 4 +7.8{zL2 Y 4} −6.6

| {z } |
LY 6 +1.1 G
| {z } |
2 2 2
{z
L Y } 10−6
2 loop 3 loop 3 loop 3 loop
 
3 2
+ 5.7
| L {z Y } |+5.9 L4  10−7 .
{z }
3 loop 3 loop

We see that the decrease of the effective four-Φ coupling with increasing energy is induced
by top quark loops. Without quarks there would be an increase. It is also worth noting that
the individual contributions at three-loop level are much larger than the overall effect due
to huge cancellations. Consider for example the five numerically largest three-loop terms at
µ = MZ :  
7.9 G2 Y 6 − 4.8 Y 8 − 3.1 G2 LY 4 − 2.5 G4 Y 4 + 2.6 G4 LY 2 10−5 .

10
The total contribution from these terms is by almost two orders of magnitude smaller than
the size of the largest one.
For the top-Yukawa β-functions we find
βy(1) 9
t
= yt2 − 4 gs2 ,
yt 4
(2)
βyt
 
2 2 4 2 2 4 202 20
=3 λ − 6 yt λ − 6 yt + 18 gs yt + gs − + nf ,
yt 3 9
(3)
βyt
 
3 15 2 2 4 6 339 27 (29)
= − 18 λ + y λ + 99 yt λ + yt + ζ3
yt 8 t 16 4
 
2 2 157 2 4 4 2 4799 27
+ 8 gs yt λ − g y + gs yt − nf − 114ζ3
2 s t 12 2
 
6 2216 140 2 160
+ gs −1249 + nf + n + ζ3 n f .
27 81 f 3
As has already been mentioned above the λ-corrections are negligible here. Evaluating βyt
at the scale µ = MZ (with an assumed Higgs mass of 125 GeV and nf = 6) we get a value
of ∼ (−0.023) at one-loop level which means a decrease of yt and therefore the top mass
with increasing energy. This is due to the QCD corrections. In the absence of QCD the
β
opposite would be the case as we can see from the term ∝ yt2 in yytt . The two and three-loop
corrections are ∼ 16.6% and ∼ 0.7% with respect to the one-loop result and so quite high
compared e.g. to the case of βgs discussed below. We use again the labels (27) to get an
impression of the individual terms:
   

βyt |µ=MZ = −3.7 Y G2 +1.3 Y 3  10−2 + −4.7

| {z } | {z }
Y G4  10−3
| {z }
1 loop 1 loop 2 loop
 
3 2
+ 9.8
| Y {z G} −2.5 Y G6 −2.1 Y 5  10−4
| {z } | {z }
2 loop 3 loop 2 loop
 
3 4
+ 9.3
| Y {z G} −3.1 LY 3 −2.5{z
| {z } |
Y 5 G2} 10−5 (30)
3 loop 2 loop 3 loop
 

+ |6.0{zY 7} +3.0 LY 5 +2.3 L2 Y  10−6

| {z } | {z }
3 loop 3 loop 2 loop
     
3 2 −7 3  −8 2 3 −9
+ 3.9
| LY{z G} 10 + |−1.2{zL Y} 10 + 8.5
 
| L{z Y } 10 .
3 loop 3 loop 3 loop

For the strong coupling we get

βg(1)
 
11 1
s
=gs2 − + nf ,
gs 2 3
(2)
βgs
 
2 2 4 19
= − gs yt + gs −51 + nf , (31)
gs 3
(3)
βgs
 
15 2 4 4 2 6 2857 5033 325 2
= gs yt − 20 gs yt + gs − + nf − n .
gs 2 4 36 108 f

11
In order to numerically compare the higher order corrections to the above β-functions we also
give the evaluation of βgs at the scale µ = MZ and with nf = 6. The one-loop contribution
is ∼ (−0.04) to which the two and three-loop calculations give corrections of ∼ 3.7% and
∼ (−0.02)% respectively. Very small λ-corrections to βgs do not appear until four loops.
With eq. (27) and the above assumptions we get
   
3 −2 5
βgs |µ=MZ = −4.0 G 10
| {z }
+ −1.4 G 10−3
| {z }
1 loop 2 loop
    (32)
+ −6.9
|
Y 2 G3} +1.7
{z
G7 −1.3{z
| {z } |
Y 2 G5} 10−5 + 3.1
| Y
4 3 −6
{z G} 10 .
2 loop 3 loop 3 loop 3 loop

The running of the m2 parameter is given by

(1)
βm2
=6 λ + 3 yt2 ,
m2
(2)
βm2 27 4
= − 30 λ2 − 36 yt2 λ − y + 20 gs2 yt2 ,
m2 4 t
(3)  
βm2 3 297 2 2 4 351 (33)
=1026 λ + y λ + y λ + 216ζ 3
m2 2 t t
4
 
1605
+ yt6 + 45ζ3 + gs2 yt2 λ (−612 + 576ζ3 )
16
   
447 910
+ gs2 yt4 − 360ζ3 + gs4 yt2 − 16nf − 24ζ3 .
2 3
Again we evaluate βm2 at the scale µ = MZ (with an assumed Higgs mass of 125 GeV and
nf = 6) and get a value of ∼ (+0.023) at one-loop level which means an increase of m2 at
higher energy scales. The two and three-loop corrections are ∼ 2.9% and ∼ 0.32%. With the
labels (27) we can estimate the contributions of the individual terms:
     
βm2
|µ=MZ = |1.8{zY 2} 10−2 + 5.3
| {zL} +1.1 G2 Y 2} 10−3 + −2.4 Y 4 −1.9 LY 2  10−4
m2 | {z | {z } | {z }
1 loop 1 loop 2 loop 2 loop 2 loop
 
4 2
{z Y } −7.0
+ |9.4 G |
G2 Y 4} |+3.3
{z
Y 6 −2.4 L2 +1.1 LY 4  10−5
{z } | {z } | {z } (34)
3 loop 3 loop 3 loop 2 loop 3 loop
   
2 2 3
| G{zLY } 10 + 7.1
+ 4.0 −6  L2 Y 2} 10−7 .
| {zL } |+7.0{z
3 loop 3 loop 3 loop

5 Results for the anomalous dimensions

In this section we give the anomalous dimensions of the physical fields in this setup. Note
that because of the SU(2) symmetry Φ1 and Φ2 must have the same anomalous dimension
γ2Φ . The same holds for the left-handed part of top and bottom quarks: γ2,L
t b
= γ2,L . For the
quark flavours q which do not participate in the Yukawa interaction there is no difference
between the left- and right-handed part as they are renormalized by the same Z-factor:

12
γ2q ≡ γ2,L
q q
= γ2,R b
. This also applies to the right-handed part of the bottom quark: γ2,R = γ2q .
All these relations have been tested explicitly during our calculation which provides a nice
additional check.

q (1)
γ2 =gs2 CF (1 − ξ) ,
 
q (2) 4 3 2 17 5 1 2
γ2 =gs − CF + CA CF − 2nf TF CF − ξCA CF + ξ CA CF ,
2 2 2 4

q (3) 3 143 10559
γ2 =6TF CF gs4 yt2 + gs6 C3 − CA CF2 + C 2 CF + 3nf TF CF2 (35)
2 F 4 144 A
1301 20 15 371 2
− nf CA TF CF + n2f TF2 CF + 12ζ3 CA CF2 − ζ3 CA2 CF − ξCA CF
36 9 2 32 
17 3 69 3 5
+ ξnf CA TF CF − ξζ3 CA2 CF + ξ 2 CA2 CF + ξ 2 ζ3 CA2 CF − ξ 3 CA2 CF .
4 2 32 8 16
For yt = 0 this is in agreement with the well-known QCD result [24]. The renormalization
constants for t,b and q can also be found in [22] up to order gs6 and gs4 yt2 where the calculation
has been performed in the broken phase of the SM.
t (1) q (1) 1 2
γ2,L =γ2 + y ,
2 t 
t (2) q (2) 4 1 3
γ2,L =γ2 − yt + dR − 2CF gs2 yt2 ,
4 4
 
t (3) q (3) 33 2 2 4 6 3 29 3 2 3
γ2,L =γ2 − y λ + 6 yt λ + yt − + dR − dR + ζ3 (36)
4 t 2 8 8 2
 
13 5
+ gs2 yt4 CF + CF dR + 6ζ3 CF dR
2 8
 
4 2 51 2 31 3 2
+ gs yt − CF + CA CF − nf TF CF + 6ζ3 CF − 15ζ3 CA CF .
8 8 2
t (1) q (1)
γ2,R =γ2 + yt2 ,
 
t (2) q (2) 1 3
γ2,R =γ2 − yt4 + dR − 4CF gs2 yt2 ,
4 2
 
t (3) q (3) 33 2 2 33 53 3
γ2,R =γ2 − yt λ + 12 yt4 λ + yt6 − + dR − d2R + 3ζ3 (37)
2 16 8 4
 
5
+ gs2 yt4 5CF + CF dR + 12ζ3 CF dR
4
 
4 2 51 2 31 2
+ gs yt − CF + CA CF − 3nf TF CF + 12ζ3 CF − 30ζ3 CA CF .
4 4
Φ (1)
γ2 =dR yt2 ,
Φ (2) 9
γ2 =6 λ2 − dR yt4 + 5CF dR gs2 yt2 ,
4
Φ (3) 45
γ2 = − 36 λ3 − dR yt2 λ2 + 15dR yt4 λ
 2   
25 15
6
+ yt − dR + 6d2R + 3ζ3 dR + gs2 yt4 CF dR − 18ζ3 CF dR
16 8
 
119 77
+ gs4 yt2 − CF2 dR + CA CF dR − 8nf TF CF dR + 36ζ3 CF2 dR − 18ζ3 CA CF dR ,
4 2
(38)

13
The purely λ-dependent part of this has been computed before in [17, 18].

 
g (1) 2 5 4 1
γ2 =gs − CA + nf TF − ξCA ,
3 3 2
 
g (2) 2 2 4 23 2 15 2 1 2 2
γ2 = − 4TF gs yt + gs − CA + 4nf TF CF + 5nf CA TF − ξCA + ξ CA ,
4 8 4
 
g (3) 25
γ2 =gs2 yt4 (9TF + 7TF dR ) − gs4 yt2 6TF CF + CA TF
2
 (39)
4051 5 875 44
+ gs6 − CA3 − 2nf TF CF2 + nf CA TF CF + nf CA2 TF − n2f TF2 CF
144 18 18 9
76 2 2 3 3 2 127 3
− nf CA TF + ζ3 CA + 24ζ3 nf CA TF CF − 18ζ3 nf CA TF − ξCA
9 2  16
9 27 3 7
+2ξnf CA2 TF − ξζ3 CA3 + ξ 2 CA3 + ξ 2 ζ3 CA3 − ξ 3 CA3 .
8 16 16 32

This is also in agreement with [24] for yt = 0 and with [23] for yt = 0, ξ = 0. For dR = 3 and
TF = 12 (QCD) these results are as follows:

q (1) 4
γ2 = (1 − ξ) gs2 ,
3  
q (2) 4 94 4 2
γ2 =gs − nf − 10ξ + ξ ,
3 3

24941 1253 20 1113 (40)
q (3) 4 2 6
γ2 =4 gs yt + gs + − nf + n2f − 26ζ3 − ξ
36 18 27 8

17 207 2 9 2 15
+ ξnf − 18ξζ3 + ξ + ξ ζ3 − ξ 3 ,
2 8 2 4

t (1) q (1) 1 2
γ2,L =γ2 +y ,
2 t
t (2) q (2) 5 8
γ2,L =γ2 − yt4 − gs2 yt2 ,
2 3
33 2 2

3
 (41)
t (3) q (3) 4 6
γ2,L =γ2 − y λ + 6 y t λ + y t 6 + ζ3
4 t 2
   
2 4 67 4 2 25 148
+ gs yt + 24ζ3 + gs yt − nf − ζ3 .
6 6 3

t (1) q (1)
γ2,R =γ2 + yt2 ,
t (2) q (2) 19 4 16 2 2
γ2,R =γ2 − y − g y ,
4 t 3 s t  
t (3) q (3) 33 2 2 4 6 177
(42)
γ2,R =γ2 − yt λ + 12 yt λ + yt + 3ζ3
2 16
   
2 4 35 4 2 25 296
+ gs yt + 48ζ3 + gs yt − 2nf − ζ3 ,
3 3 3

14
 Φ (1)
γ2 =3 yt2 ,
Φ (2) 27 4
γ2 =6 λ2 − y + 20 gs2 yt2 ,
4 t  
Φ (3) 3 135 2 2 4 6 789
(43)
γ2 = − 36 λ − y λ + 45 yt λ + yt + 9ζ3
2 t 16
   
2 4 15 4 2 910
+ gs yt − 72ζ3 + gs yt − 16nf − 24ζ3 ,
2 3

 
g (1) 2 2 3
γ2 =gs −5 + nf − ξ ,
3 2
 
g (2) 2 2 4 207 61 135 9 2
γ2 = − 2 gs yt + gs − + nf − ξ+ ξ ,
4 6 8 4
g (3) 91
γ2 =15 gs2 yt4 − gs4 yt2 (44)
 4
12153 7831 215 2 81 3429
+ gs6 − + nf − n + ζ3 − 33ζ3 nf − ξ
16 36 27 f 2 16

243 729 2 81 2 189 3
+9ξnf − ξζ3 + ξ + ξ ζ3 − ξ .
8 16 16 32

6 The evolution of the quartic Higgs coupling

The quartic Higgs coupling λ is of special interest as it is directly related to the Higgs mass
MH . If we assume that the SM is valid up to some high energy scale Λ, then the value of
MH should meet the constraints

mmin < MH < mmax .

Here the upper limit is related to the well-known fact that the running Higgs self-coupling
develops a Landau pole7 if MH is large [32,33,34]. For Λ = MP lanck = 1018 GeV the estimated
value of mmax is around 175 GeV [32, 33, 34, 35], which is already excluded by experiments
carried out at the LHC and the Tevatron.
The lower limit mmin follows from the requirement of the vacuum stability [36, 37, 38]. In
order to find mmin one should construct the effective Higgs potential V [φ] including radiative
corrections and sum possible large logarithms using the standard method of the Renormal-
ization Group (for a review see, e.g. [39]). Once this has been done, the condition that the
potential V [φ] does not develop a deeper minimum in addition to the standard one for all
values of φ < Λ fixes mmin .
In our analysis we will use a simplified approach for finding mmin , namely the requirement
that the running coupling constant λ(µ) stay non-negative for all µ less than Λ. It has been
shown in [33,40] that the simplified approach is essentially equivalent to the one based on the
use of the effective potential provided the instabilty of V [φ] can only happen at φ ≫ MZ .

7
This is true in the one-loop approximation. At two loops the Landau pole is replaced by an ultraviolet
metastable fixed point with the resulting fixed point value of λ being outside the weak coupling region.

15
 (3)
In this section we investigate the effect of the three-loop result βλ on the running of λ and
therefore its effect on the stability of the electroweak vacuum in the SM. For this we also
include the electroweak contributions up to the two-loop level. The two-loop β-functions
for the SM gauge couplings have been derived in [27, 28, 29, 30]. The two-loop results for
the Yukawa-couplings and λ can be found in [19, 21, 20]. Now we add the three-loop results
derived in the previous section and investigate the effect this has on the evolution of our
couplings (for a recent similar analysis, using the two-loop running, see, e.g. [41, 42]).
To find starting values for the running of the couplings we should account for the fact that
the physical parameters (e.g. pole masses) are related to the ones in the MS-scheme in a
non-trivial way (see e.g. [43, 44, 45]). For example eq. (15) is only valid at tree level. For
the higher order corrections we take the electroweak ones at one-loop and the QCD ones at
two-loop level from [43, 44]. These matching relations depend on the exact values of αs (MZ )
and the pole mass Mt of the top quark and of course the mass of the Higgs boson MH . For
the latter we consider the cases MH = 124 GeV and MH = 126 GeV. For the other two we
use the values

αs (MZ ) = 0.1184 ± 0.0007 and Mt = 172.9 ± 0.6 ± 0.9 GeV [46]. (45)

One should also keep in mind that the matching relations themselves receive contributions
from not yet known higher order corrections . The corresponding uncertainty in mmin has
been estimated in [4, 41] and found to be about 2 GeV.
Fig. 6 shows the evolution of λ in this framework up to the Planck scale. To estimate the
dependence of the λ-running on the parameters αs (MZ ) and Mt we give the shifted curves
for λ(µ) when we change these parameters by ±σ as given in eq. (45).8 As the two- and
three-loop curves are very close together we zoom in on the region where λ crosses over to
negative values in Fig. 7. In this plot we give the αs -uncertainty for the two- and three-loop
curves to compare between this uncertainty and the shift from two to three loops. Note that
there is a considerable difference between MH = 124 GeV and MH = 126 GeV which means
that the evolution of λ is very sensitive to the value of the Higgs mass. Given a fixed value
for MH the largest uncertainty lies in the exact value of the top mass. The second largest
uncertainty comes from αs . The total effect due to the three-loop part of the β-functions is
somewhat smaller than latter as can be seen best in Fig. 7. Still, it is worthy of note that
the three-loop corrections to the β-functions presented here enhance the stability of the SM
electroweak vacuum.
The smallness of the three-loop correction to βλ seems to be somewhat coincidental as the
(3)
aforementioned cancellations of individual terms in βλ depend strongly on the value of
MH . Finding a Higgs with a mass of 124 to 126 GeV would therefore mean an excellent
convergence of the perturbation series for βλ . Another intriguing consequence of a Higgs
mass in that region is the uncertainty whether λ becomes indeed negative at high scales
or not. If we take e.g. MH = 126 GeV and αs = 0.1184 and decrease9 the top mass from
Mt = 172.9 GeV to Mt = 171.25 GeV (Mt = 171.16 GeV without the three-loop corrections),
then λ stays positive up to the Planck scale MP lanck in our framework. The same effect can be
achieved for MH = 126 GeV and Mt = 172.9 GeV by increasing αs = 0.1184 by 6.5 σαs (7 σαs
8
In order not to make the plot too crowded these shifted curves are only given for the two-loop result.
The difference to the three-loop result is similar to the one between the two- and three-loop curves for
αs (MZ ) = 0.1184 and Mt = 172.9 GeV.
9
Smaller values for Mt or larger values for αs increase the stability of the vacuum.

16
 m H =124GeV

m t = 172.9 GeV
0.06 Αs HM Z L = 0.1184

0.04
Λ HΜL

0.02 m
t =1
71
.4
Ge
V

0.00 m
t =1
74.
4G
eV
Αs HM Z L = 0.1191

- 0.02 Αs HM Z L = 0.1177

4 6 8 10 12 14 16 18
Log10 @ Μ  GeV D

m H =126 GeV

m t = 172.9 GeV
Αs HM Z L = 0.1184
0.06

0.04
Λ HΜL

0.02 mt
=1
71.4
GeV

0.00
mt
=1
74.4 Αs HM Z L = 0.1177
GeV

Αs HM Z L = 0.1191
- 0.02
4 6 8 10 12 14 16 18
Log10 @ Μ  GeV D

Figure 6: Evolution of λ with the scale µ: 2 loop (dashed, blue) and 3 loop (continuous, red)
results; Uncertainties with respect to the two-loop result: ±1σαs , ±1σMt (dotted)

without the three-loop corretions). A combined scenario for MH = 126 GeV would be a shift

17
 m H =124GeV

m t = 172.9 GeV
0.002 Αs HM Z L = 0.1184
Αs H M
ZL = 0.11
91 H 3
loopL
Αs H M
ZL = 0.11
91 H 2
0.001 3 lo o p loopL

2 lo o
p
Λ HΜL

0.000

Αs H M
ZL = 0.11
77 H 3
- 0.001 loopsL
Αs H M
ZL = 0.11
77 H 2
loopL

- 0.002

9.20 9.25 9.30 9.35 9.40 9.45 9.50

Log10 @ Μ  GeV D

m H =126 GeV

m t = 172.9 GeV
0.002
Αs HM L Αs HM Z L = 0.1184
Z = 0.11
9 1 H 3 loo
pL

Αs HM L
Z = 0.11
9 1 H 2 loo
0.001 pL
3 lo o p

2 lo o p
Λ HΜL

0.000

- 0.001 Αs HM L
Z = 0.1
1 77 H 3 lo
opsL

Αs HM L
Z = 0.11
77 H 2 lo
- 0.002 opL

10.60 10.65 10.70 10.75 10.80 10.85 10.90

Log10 @ Μ  GeV D

Figure 7: Evolution of λ with the scale µ: 2 loop (dashed, blue) and 3 loop (continuous, red)
results; Uncertainties with respect to the two- and three-loop results: ±1σαs (dotted)

18
of Mt = 172.9 GeV by −1 σMt = −1.5 GeV and of αs = 0.1184 by +1 σαs = +0.0007 GeV
which would also make λ positive up to the Planck scale.
Thus, we conclude that at present no definite answer can be given to the question whether
the SM vacuum is stable all the way up to the Planck scale or not. If indeed a SM Higgs
boson is found with a mass of 124 to 126 GeV, this is a good motivation for determining
αs (MZ ) and Mt as accurately as possible as well as calculating the SM β-functions to the
highest achievable accuracy.

7 Conclusions

We have computed the three-loop corrections to the evolution the top-Yukawa coupling, the
strong coupling and the quartic Higgs self-coupling in the unbroken SM with the numerically
small gauge coupling constants g1 and g2 and all Yukawa couplings except for yt set to zero.
The implications of our calculation on the stability of the electroweak vacuum in the SM can
be summarized as follows:
• The total effect of the three-loop terms is relatively small which is not self-evident as
(3)
the individual terms in βλ are much larger than the final value due to significant
cancellations for a Higgs mass in the vicinity of 125 GeV.
• The evolution of λ is very sensitive to the values of the Higgs mass, the top mass and
αs (MZ ). If we take e.g. MH = 126 GeV and decrease the top mass by about 1.7 GeV,
then λ stays positive up to the Planck scale (a similar observation has been made
in [41]). This is a very good motivation for high precision measurements of αs (MZ )
and Mt . With the latter values known more precisely, the account of the the three-
loop effects in the evolution of the quartic Higgs self-coupling would be essential in
considering the problem of the stability of the electroweak vacuum in the SM.
• In this context it may also be useful to calculate both the electroweak contributions to
(3)
βλ and βy(3)t
at three-loop level as well as the matching of experimentally measurable
on-shell parameters and MS-parameters to a higher accuracy.
We thank Luminita Mihaila, Jens Salomon and Matthias Steinhauser for useful discussions
and informing us on the results of [31] before their publication. We thank Johann Kühn for
valuable comments and support. Last but not least, we are grateful to Fedor Bezrukov and
Mikhail Shaposhnikov for gently drawing our attention to the subject, numerous discussions
and providing us with their version of a Mathematica package to perform one-loop matching
in the SM.
In conclusion we want to mention that all our calculations have been performed on a SGI
ALTIX 24-node IB-interconnected cluster of 8-cores Xeon computers using the thread-based
[47] version of FORM [9]. The Feynman diagrams have been drawn with the Latex package
Axodraw [48].
This work has been supported by the Deutsche Forschungsgemeinschaft in the Sonder-
forschungsbereich/Transregio SFB/TR-9 “Computational Particle Physics”.

19
References

[1] F. L. Bezrukov and M. Shaposhnikov, The Standard Model Higgs boson as the inflaton.
Phys. Lett. B659 (2008) 703–706, arXiv:0710.3755 [hep-th].
[2] F. L. Bezrukov, A. Magnin, and M. Shaposhnikov, Standard Model Higgs boson mass
from inflation. Phys. Lett. B675 (2009) 88–92, arXiv:0812.4950 [hep-ph].
[3] A. De Simone, M. P. Hertzberg, and F. Wilczek, Running Inflation in the Standard
Model. Phys. Lett. B678 (2009) 1–8, arXiv:0812.4946 [hep-ph].
[4] F. Bezrukov and M. Shaposhnikov, Standard Model Higgs boson mass from inflation:
two loop analysis. JHEP 07 (2009) 089, arXiv:0904.1537 [hep-ph].
[5] M. Shaposhnikov and C. Wetterich, Asymptotic safety of gravity and the Higgs boson
mass. Phys. Lett. B683 (2010) 196–200, arXiv:0912.0208 [hep-th].
[6] ATLAS Collaboration, G. Aad et al., Combined search for the Standard Model Higgs
boson using up to 4.9 fb-1 of pp collision data at sqrt(s) = 7 TeV with the ATLAS
detector at the LHC. Phys. Lett. B710 (2012) 49–66, arXiv:1202.1408 [hep-ex].
[7] CMS Collaboration, S. Chatrchyan et al., Combined results of searches for the
standard model Higgs boson in pp collisions at sqrt(s) = 7 TeV.
arXiv:1202.1488 [hep-ex].
[8] TEVNPH (Tevatron New Phenomina and Higgs Working Group)
Collaboration, Combined CDF and D0 Search for Standard Model Higgs Boson
Production with up to 10.0 fb-1 of Data. arXiv:1203.3774 [hep-ex].
[9] J. A. M. Vermaseren, New features of FORM. arXiv:math-ph/0010025.
[10] S. G. Gorishnii, S. A. Larin, L. R. Surguladze, and F. V. Tkachov, MINCER: Program
for multiloop calculations in quantum field theory for the SCHOONSCHIP system.
Comput. Phys. Commun. 55 (1989) 381–408.
[11] M. Misiak and M. Munz, Two loop mixing of dimension five flavor changing operators.
Phys. Lett. B344 (1995) 308–318, arXiv:hep-ph/9409454.
[12] K. G. Chetyrkin, M. Misiak, and M. Munz, Beta functions and anomalous dimensions
up to three loops. Nucl. Phys. B518 (1998) 473–494,
arXiv:hep-ph/9711266 [hep-ph].
[13] M. Steinhauser, MATAD: A program package for the computation of massive tadpoles.
Comput. Phys. Commun. 134 (2001) 335–364, arXiv:hep-ph/0009029.
[14] P. Nogueira, Automatic Feynman graph generation.
J. Comput. Phys. 105 (1993) 279–289.
[15] T. Van Ritbergen, A. Schellekens, and J. Vermaseren, Group theory factors for
Feynman diagrams. International Journal of Modern Physics A 14 (1999) no. 1, 41–96.
cited By (since 1996) 70.
[16] G. ’t Hooft and M. J. G. Veltman, Regularization and Renormalization of Gauge
Fields. Nucl. Phys. B44 (1972) 189–213.
[17] E. Brezin, J. C. Le Guillou, J. Zinn-Justin, and B. G. Nickel, Higher oder contributions
to critical exponents. Phys. Lett. 44A (1973) 227–228.

20
[18] E. Brezin, J. C. Le Guillou, and J. Zinn-Justin, Addendum to wilson’s theory of critical
phenomena and callan-symanzik equations in 4-epsilon dimensions.
Phys. Rev. D9 (1974) 1121–1124.
[19] M.-x. Luo and Y. Xiao, Two loop renormalization group equations in the standard
model. Phys. Rev. Lett. 90 (2003) 011601, arXiv:hep-ph/0207271 [hep-ph].
[20] M. E. Machacek and M. T. Vaughn, Two-loop renormalization group equations in a
general quantum field theory: (III). Scalar quartic couplings.
Nucl. Phys. B 249 (1985) no. 1, 70–92.
[21] M. E. Machacek and M. T. Vaughn, Two-loop renormalization group equations in a
general quantum field theory (II). Yukawa couplings.
Nucl. Phys. B 236 (1984) no. 1, 221–232.
[22] M. Steinhauser, Higgs decay into gluons up to O(alpha**3(s) G(F)m**2(t)).
Phys.Rev. D59 (1999) 054005, arXiv:hep-ph/9809507 [hep-ph].
[23] O. Tarasov, A. Vladimirov, and A. Zharkov, The gell-mann-low function of QCD in
the three-loop approximation. Phys. Lett. B 93 (1980) no. 4, 429–432.
[24] S. Larin and J. Vermaseren, The Three loop QCD Beta function and anomalous
dimensions. Phys. Lett. B303 (1993) 334–336, arXiv:hep-ph/9302208 [hep-ph].
[25] D. J. Gross and F. Wilczek, Ultraviolet Behavior of Non-Abelian Gauge Theories.
Phys. Rev. Lett. 30 (1973) 1343–1346.
[26] H. D. Politzer, Reliable Perturbative Results for Strong Interactions?
Phys. Rev. Lett. 30 (1973) 1346–1349.
[27] M. Fischler and J. Oliensis, Two-loop corrections to the beta function for the
Higgs-Yukawa coupling constant. Phys. Lett. B 119 (1982) no. 4, 385–386.
[28] D. R. T. Jones, Two-loop β function for a G1 × G2 gauge theory.
Phys. Rev. D 25 (1982) 581–582.
[29] I. Jack and H. Osborn, General background field calculations with fermion fields.
Nucl. Phys. B 249 (1985) no. 3, 472–506.
[30] M. E. Machacek and M. T. Vaughn, Two-loop renormalization group equations in a
general quantum field theory: (I). Wave function renormalization.
Nucl. Phys. B 222 (1983) no. 1, 83–103.
[31] L. N. Mihaila, J. Salomon, and M. Steinhauser, Gauge coupling beta functions in the
standard model to three loops. Phys. Rev. Lett. 108 (2012) 151602.
[32] L. Maiani, G. Parisi, and R. Petronzio, Bounds on the Number and Masses of Quarks
and Leptons. Nucl. Phys. B136 (1978) 115.
[33] N. Cabibbo, L. Maiani, G. Parisi, and R. Petronzio, Bounds on the Fermions and
Higgs Boson Masses in Grand Unified Theories. Nucl. Phys. B158 (1979) 295–305.
[34] M. Lindner, Implications of Triviality for the Standard Model.
Z. Phys. C31 (1986) 295.
[35] T. Hambye and K. Riesselmann, Matching conditions and Higgs mass upper bounds
revisited. Phys.Rev. D55 (1997) 7255–7262, arXiv:hep-ph/9610272 [hep-ph].

21
[36] N. Krasnikov, Restriction of the Fermion Mass in Gauge Theories of Weak and
Electromagnetic Interactions. Yad. Fiz. 28 (1978) 549–551.
[37] P. Q. Hung, Vacuum Instability and New Constraints on Fermion Masses.
Phys. Rev. Lett. 42 (1979) 873.
[38] H. D. Politzer and S. Wolfram, Bounds on Particle Masses in the Weinberg-Salam
Model. Phys. Lett. B82 (1979) 242–246.
[39] M. Sher, Electroweak Higgs Potentials and Vacuum Stability.
Phys. Rept. 179 (1989) 273–418.
[40] C. Ford, D. Jones, P. Stephenson, and M. Einhorn, The Effective potential and the
renormalization group. Nucl. Phys. B395 (1993) 17–34,
arXiv:hep-lat/9210033 [hep-lat].
[41] J. Elias-Miro, J. R. Espinosa, G. F. Giudice, G. Isidori, A. Riotto, et al., Higgs mass
implications on the stability of the electroweak vacuum.
Phys. Lett. B709 (2012) 222–228, arXiv:1112.3022 [hep-ph].
[42] Z.-z. Xing, H. Zhang, and S. Zhou, Impacts of the Higgs mass on vacuum stability,
running fermion masses and two-body Higgs decays. arXiv:1112.3112 [hep-ph].
[43] J. Espinosa, G. Giudice, and A. Riotto, Cosmological implications of the Higgs mass
measurement. JCAP 0805 (2008) 002, arXiv:0710.2484 [hep-ph].
[44] R. Hempfling and B. A. Kniehl, On the relation between the fermion pole mass and MS
Yukawa coupling in the standard model. Phys.Rev. D51 (1995) 1386–1394,
arXiv:hep-ph/9408313 [hep-ph].
[45] A. Sirlin and R. Zucchini, Dependence of the Higgs coupling hMS(M) on mH and the
possible onset of new physics. Nucl. Phys. B 266 (1986) no. 2, 389–409.
[46] K. Nakamura et al., The review of particle physics. J. Phys. G (2010) no. 37, 075021.
[47] M. Tentyukov and J. A. M. Vermaseren, The multithreaded version of FORM.
arXiv:hep-ph/0702279.
[48] J. A. M. Vermaseren, Axodraw. Comput. Phys. Commun. 83 (1994) 45–58.

22