Revisiting the decoupling effects in the running of the Cosmological Constant

We revisit the decoupling effects associated with heavy particles in the renormalization group running of the vacuum energy in a mass-dependent renormalization scheme. We find the running of the vacuum energy stemming from the Higgs condensate in the entire energy range and show that it behaves as expected from the simple dimensional arguments meaning that it exhibits the quadratic sensitivity to the mass of the heavy particles in the infrared regime. The consequence of such a running to the fine-tuning problem with the measured value of the Cosmological Constant is analyzed and the constraint on the mass spectrum of a given model is derived. We show that in the Standard Model (SM) this fine-tuning constraint is not satisfied while in the massless theories this constraint formally coincides with the well known Veltman condition. We also provide a remarkably simple extension of the SM where saturation of this constraint enables us to predict the radiative Higgs mass correctly. Generalization to constant curvature spaces is also given.


I. INTRODUCTION
It is widely accepted that our today's universe is undergoing the phase of the accelerated expansion which is commonly explained by the presence of the Cosmological Constant (CC) Λ. However, the value of Λ required by experiment is in a contradiction with the values emerging from the physics scales associated with known phase transitions in the universe so that severe fine-tuning has to be applied which is at heart of the CC problem. To recall the main aspects of this problem we begin with the Standard Model (SM) formulated on the classical curved background. In order to construct a renormalizable gauge theory in an external gravitational field one starts from the classical action (with ϕ as the Higgs doublet field) The renormalization procedure for the theory (1) consists of the renormalization of the SM matter fields, couplings and masses, non-minimal coupling ξ and the gravitational couplings a 1,2,3,4 , G vac and Λ vac . We are going to work in the low energy domain of the gravitational physics and, for that reason, the short distance effects from the higher derivative terms a 1,2,3,4 , in (1) are not important for our considerations, and so we start with the usual bare Hilbert-Einstein action with coupling constants G vac , Λ vac supplemented with non-minimal coupling ξ: The bare quantities are defined with the superscript "0". Let us focus on the CC itself which, as we mentioned above, must be renormalized and the connection with experimentally measured value ρ phys is achieved via the renormalization condition (see Eq.(8) below) imposed on the vacuum energy density: at some energy scale µ so that ρ vac Λ (µ). Moreover, in the presence of the dynamical cosmological background characterized by the time-dependent Hubble parameter H(t), ρ vac Λ can be dynamical ρ vac Λ (µ, t) that will be reflected in the evolution of ρ vac Λ via: which is the important ingredient for cosmological evolution. Currently, there is no consensus on whether ρ vac Λ (µ, t) depends on t. Even if it is time-dependent, to understand the precise form of (4) one may go back to the time-independent RG problem: where g i and m i are the dimensionless couplings and masses respectively. The g i and m i are also supplemented with their own Renormalization Group (RG) equations. Besides ρ vac Λ , the physical vacuum energy ρ phys consists of several additional parts. One of these parts is "induced" contribution ρ ind (µ) to the vacuum energy density arising from the vacuum condensates. For example, if ϕ vac is the value of the Higgs field ϕ(x) which minimizes the Higgs potential V(ϕ) the Higgs condensate contribution (at the classical level) to the vacuum energy is Besides the vacuum and induced terms we may have additional effects from the higher derivative terms in (1) as well as corrections from quantum gravity. Again, these contributions can be classified as coming from purely quantum effects and therefore expected to be µ-dependent and some also time-dependent due to the expanding background, and therefore contributing to (4). All in all, the physical value is measured at the cosmological RG scale µ c , which is experimentally given by µ c = O(10 −3 ) eV, as The problem now is that if we use the experimental Higgs mass M H = 125 GeV, then the corresponding value ρ ind 10 8 GeV 4 . In order to keep the QFT consistent with astronomical observations, one has to demand that the parts contributing to the ρ phys should cancel with the accuracy dictated by the current data. For example, if we neglect all the ... terms in (8), the ρ vac Λ and ρ ind should cancel with the precision of 55 decimal orders. This is the CC fine-tuning problem [1,2].
To understand deeper this tuning, one has to take into account the decoupling effects due to massive particles. Clearly, we expect that contribution to the RG running from the particle of mass m should change dramatically, as we go from µ m to µ m regime. Moreover, requiring the absence (or, at least, reduction) of the tuning may provide a constraint on the spectrum of the particle physics models.
In this paper, we deal with the time-independent classical curved background and will derive the RG evolution of ρ vac Λ and ρ ind of the form (5) taking into account the decoupling effects due to massive particles by using the mass-dependent RG formalism. We also elucidate the implication of the leading residual effects on the RG evolution of ρ phys due to the heaviest particles in the SM and present a simple phenomenological extension of the SM predicting correctly the Higgs mass.
The paper is structured as follows. In the next Sec.II we briefly discuss the RG running of the CC in the simple φ 4 -theory highlighting the necessary RG formalism we use later and also discuss the basic issue of decoupling in the RG running. In Sec.III we extend the RG approach to the full SM, in both, mass-independent and mass-dependent RG schemes. Sec.IV deals with applications of the derived heavy-mass threshold effects within and beyond the SM and Sec.V presents our conclusions. In Appendices we provide the technical details, as well as generalize the flat spacetime results to the spaces with constant curvature.

II. RG RUNNING OF THE COSMOLOGICAL CONSTANT
To prepare for the discussion of the RG dependence of the CC and to setup the formalism, let us consider the skeleton Lagrangian for the real scalar: The schematic contributions to the one-loop effective potential, up to the 4 external legs, are shown in Fig.1 . Correspondingly to this diagramatic picture and for a general QFT, the renormalized effective potential can be split into two pieces: the φ-independent (vacuum) term corresponding to the diagram Fig.1(a) and the φ-dependent "scalar" term connected with diagrams Fig.1(b,c): where the parameter ρ vac Λ = ρ vac Λ (µ) depends on the vacuum cosmological constant and will be defined in the next section. In order to understand the origin of this splitting, one introduces the functional called the effective action of the vacuum Γ vac . It is part of the full effective action which is left when the mean scalar field φ is set to zero: Γ vac = Γ[φ = 0]. Thus, it is a pure quantum object which only depends on the set of parameters P = m, λ, ... of the classical theory. At the functional level, the generating functional W for the vacuum-to-vacuum transition amplitude is where the source J is set to zero. In this way, the functional Γ vac is the generator of the proper vacuum-to-vacuum diagrams.
The RG-invariance of the full renormalized effective potential reads (where γ m m 2 = β m 2 ): Using (10), we now show that Eq. (12) is, in fact, a sum of two independent RG equations, To prove this, notice that from the RG-invariance of the renormalized effective action follows the µ-independence of the renormalized functional Γ vac and, therefore, we arrive at the second identity (14) for the vacuum part of the effective potential, while the first identity is then the result of the subtraction of (14) from (12). We will illustrate this point later.
The net result is that the vacuum and matter parts of the effective potential are overall µindependent separately and no cancelation between them is expected.

A. Vacuum part of the CC
Let us compute the V vac (m 2 , λ, ρ vac Λ , µ) object at the one-loop level. We start from As it is well-known the Λ 0 vac -dependent part has exactly the form of the bare vacuum energy density (3) 1 : In the standard QFT, the loop-divergent terms in the vacuum density are absorbed by the bare cosmological constant term (ρ 0 Λ ) vac of the Hilbert-Einstein action. For this, we split the bare term (ρ 0 Λ ) vac as where the counterterm δρ vac Λ depends on the regularization and the renormalization scheme. Specifically, the one-loop effects encoded inV (1) vac modify this relation as follows: ) . (19) In terms of Feynman diagrams this is just the vacuum bubble shown in Fig.1(a). Integrating this equation and adding (ρ 0 Λ ) vac we obtain the vacuum energy density (18) as: The pole of the Gamma function in 4 dimensions Γ(1 − n 2 ) ∼ 2/(n − 4) so for n → 4: Equation (21) is divergent and needs a subtraction. If we adopt the MS subtraction scheme, the counterterm δρ vac Λ gets fixed in such a way that the renormalized vacuum energy density at 1-loop is This is the result for V vac (m 2 , λ, ρ vac Λ , µ) at 1-loop. Notice that it is a pure quantum object that (to one-loop order) depends only on the parameter m of the classical Lagrangian and does not depend (to this order) on λ.
It is clear from (22) that the cosmological constant is renormalized according to: and µ ∂ρ vac This is the expression for β ρ vac Λ calculated in the MS scheme. In writing this equation we used the renormalized mass because the RG equations must involve only finite (renormalized) quantities. However, so far, we only computed the vacuum bubble in the free theory, where renormalized and bare masses are the same. In the interacting theory, we have to take care about the renormalization of the mass m itself. The leading mass correction is based on the correction to the scalar propagator shown in Fig.1(b) and after the standard calculation we arrive at: The addition of the interactions modifies the renormalization of the cosmological constant according to the two-loop diagram shown in Fig.2 which leads to: To put this expression in the renormalized form we have to replace the bare mass by the renormalized one using (25), while we can use renormalized quartic coupling λ 0 = λ to this order. One obtains: where we observe that the leading term is written in terms of the renormalized mass. It is clear from (27) that the cosmological constant is then renormalized according to: Note again that there is no correction to the RG to the leading order in λ. Basically, each of the two bubbles in Fig.2 acts as a mass correction to the other one and gets reabsorbed into the renormalized mass.

B. Decoupling effects
By definition, the RG equation (28) holds in the region µ m and to go to the opposite regime µ m would require to take into account: 1) the contribution of heavy particles at the energies near their mass, 2) the residual effects from the heavy particles at energies well below their mass.
It is well-known that the decoupling of heavy particles does not hold in a mass-independent scheme like the MS, and for this reason they must be decoupled by hand using the sharp cut-off procedure or some of the mass-dependent schemes. The quantum effects of the massive particles are, in principle, suppressed at low energies by virtue of the Appelquist-Carazzone theorem [4], so that in the region below the mass of the particle its quantum effects become smaller. At this point we need the relation between the IR and the UV regions which would require to extend the Wilson RG for the quantitative description of the threshold effects, and to apply a mass-dependent RG formalism.
On purely dimensional grounds, in the regime µ m one expects the corrections to the CC of the type µ 2 m 2 . These corrections can be seen from the fact that in a mass-dependent subtraction scheme a heavy mass m enters the β-functions through the dimensionless combination µ/m, so that the CC, being a dimension-4 quantity, is expected to have the β-function corrected as follows: where a,b and c are some coefficients, m light is some light mass m light µ, and the dots stand for terms suppressed by higher order powers of µ/m 1. Equipped with the necessary RG formalism and expectation of decoupling behavior based on the dimensional analysis, we will show how one can deal with the decoupling effect in the full SM and how to calculate explicitly the coefficients a, b, c for any model.

III. RG RUNNING OF THE COSMOLOGICAL CONSTANT IN THE STANDARD MODEL
Before discussing the mass-dependent RG schemes relevant for decoupling, let us recall the results in the usual MS scheme.

A. Mass-independent (MS) scheme
The renormalized effective potential of the SM, V, can be written in the 't Hooft-Landau gauge and the MS scheme as [5,6] where λ i ≡ (g, g , λ, h t ) runs over all dimensionless couplings and V 0 , V 1 are the tree level potential and the one-loop correction respectively, namely and coefficients n i , κ i , κ m i , and c i defined in Table.I. M 2 i (φ) are the tree-level expressions for the background-dependent masses of the particles that enter in the one-loop radiative corrections, is the SM analogue of the renormalized cosmological constant ρ vac Λ (µ) for the real scalar field discussed in the previous section. As discussed in the previous section we may split the effective potential into two pieces: the φ-independent (vacuum) term and the φ-dependent "scalar" term Various pieces satisfy the RG equations (12) and (13,14) with λ → λ i and these equations are valid for any value of φ. However, for the extremum value φ = φ defined via ∂V scal (φ) ∂φ φ = 0, the term containing anomalous dimension of the Higgs γ φ drops out and (13) reads: Using the tree-level potential (31), it is useful to define parameter The running of this parameter reads: By equating the terms with the different powers of φ : it is straightforward to check that the requirement (12) applied to the full one-loop effective potential (30) leads to  (14). These equations show explicitly that the vacuum V vac and scalar V scal parts of the full effective potential satisfy independent RG equations. For the extremum value φ = φ we have to drop the γ φ terms from (39) and (40) and subtracting these equations appropriately we obtain the running of the ρ ind : Combining (42) and (41) we finally obtain where we used and φ 2 = 2m 2 /λ. Eq.(43) is the central equation valid in the UV regime of massless and massive theories, theories with the spontaneous symmetry breaking (SSB) and without. This equation defines, in a compact form, the total running of the implicit µ−dependences on the l.h.s. by balancing them with the explicit µ−dependences on the r.h.s [8].

B. Mass-dependent scheme
Now, following the discussion above, we may generalize approach to the mass-dependent RG scheme. As we discussed above the decoupling of heavy particles does not hold in a massindependent MS scheme and here we recall how to get around this problem.
The basic issue can be seen in the computations of 2→ 2 scattering amplitude in a simple φ 4 -theory: which is just the potential of (31) limited to one real scalar. The exemplary scattering amplitude is shown in Fig.3 where p = p 1 +p 2 is the total incoming momenta. Expanding in terms of the external momentum p, it is only the term p = 0 which is divergent since every power of p effectively gives one less power of k for large k.  Computing the logarithmically-divergent integral using, for example, dimensional regularization, we obtain: where we see explicitly that p 2 -terms are finite.
In the MS renormalization scheme, one chooses counterterms (c.t.) in such a way as to remove the divergent 2/ pole and scale independent number −γ E +log(4π) and therefore, by construction, the counterterms are mass-independent. Also one introduces the arbitrary mass parameter µ MS to make equation dimensionally correct so that finally: From the RG equation applied to the 4-point function: one now derives the β-function of the theory: In a mass-dependent renormalization scheme, the counterterms are mass-dependent and can be chosen, for example, to subtract from (46), in addition to the divergent pole and scale independent numbers, also the log[m 2 − x(1 − x)p 2 ] evaluated at the p 2 = −µ 2 where µ is yet another arbitrary scale. After this additional finite subtraction, (46) will be replaced by the corresponding expression in the momentum subtraction scheme (MOM) as Again, µ−dependence will determine the beta function of the theory through the RG-equation and we obtain: (51) which in the µ m region, reproduces the decoupling behavior µ 2 /m 2 we discussed in (29). Now, we need to generalize the above derivation in the mass-dependent scheme for the simple φ 4 -theory to the full SM including the loops of the W, Z, t and Goldstones. In Appendix A, we show that the appropriate generalization of (39) is given by: in agreement with [9]. For the single real scalar case discussed above, we have to 1-loop γ φ = 0 and, from Table(I) we have n i = 1 and κ i = 3λ/2 so that we reproduce (51). Notice that when performing the sum over Goldstones, the parameter (M 2 phys ) i becomes the physical mass of the vector boson corresponding to the Goldstone of type i.
Similarly, in Appendix A we also show that the generalization of (40) is given by: We therefore conclude, that in this mass-dependent scheme, the corresponding MOM expression for the MS running of ρ ind in (42) takes the following form: Now it remains to derive the vacuum part, Eqs. (24) and (41), in the mass-dependent scheme. To accomplish that, one starts from the simple observation that the expression for the unrenormalized vacuum density (21) can be brought to the following form: In above, A 0 (m) is the one-point Passarino-Veltman function with the properties given in the Appendix A. Now, using the relation and the fact that V vac (m 2 , λ, ρ vac Λ , µ) satisfies the RG equation (14) we obtain for the running of the vacuum part (41) in the MOM scheme 3 The essence of (56) is to ensure that once the finite subtraction (defining the MOM scheme) was made for β ρ ind , the same finite subtraction is made for β ρ vac Λ . Putting everything together we achieve the generalization of (43) to the mass-dependent scheme. The Eq.(58) is the master equation describing the running of CC in any regime, non-decoupling and decoupling one, which is valid both in the UV and the IR regime. In the UV regime we recover (43) while in the IR regime we obtain the running of ρ vac Λ + ρ ind including the decoupling effects associated with the mass thresholds as which will be used in the next section to show the application of such running.

IV. APPLICATIONS
Here we give some examples for the application of the derived behavior of the CC (58).

A. Standard Model
• Massive case m 0: In (58) we derived the running of ρ ind + ρ vac Λ valid in the regime of decoupling of heavy particles µ 2 M 2 i ( φ ) while for the light particles (m 2 light ) j µ 2 , we can simply use (43). Working in the region where (m 2 light ) j µ 2 M 2 i ( φ ), we may combine these asymptotic results to obtain: 3 The "vacuum bubble" in Fig.1(a) is independent of the external momentum. In order to have an external momentum probe one needs to consider this "vacuum bubble" with external fields, such as for example the graviton legs. Then, to obtain the beta function for the ρ vac Λ in the MOM scheme, one has to repeat the same steps as in the φ 4 theory above. First, one has to calculate the renormalization of the quantum corrections to the n-point function of gravitons, then make a finite subtraction of the value of this quantity at p 2 = −µ 2 and, finally, calculate the derivative µ∂/∂µ of the form-factors.
This program was carried out in [10] for the contributions of the loop of massive scalar to the propagator (2-point function) of the gravitational perturbation h µν on the flat background g µν = η µν + h µν with the result that in this approach one cannot reveal the beta functions for ρ vac Λ and the form of the decoupling remained unclear.
The above expression is exactly of the form of (29) and proves the expected decoupling behavior in the effective theories. The light masses m light may be, again, generated by the Higgs vev m light ( φ ) such as a mass for, say, charm quark, or may be a new mass parameters in the SM Lagrangian related, for example, to the neutrino masses. As the µ-scale slides down the energy, more and more SM masses will migrate from the m 4 light -term to the inside of the brackets in the µ 2 -term.
Notice that, when performing the sum over Goldstones, both beta functions for ρ vac Λ (57) and ρ ind (54) have the term ∼ M 4 H /(2/M 2 W + 1/M 2 Z ) but since it comes with the opposite sign it cancels in (60). This demonstrates the importance of considering the RG running of the total ρ vac Λ + ρ ind parameter rather than RG running of these contributions separately. 4 . The µ 2 M 2 i term in the running of ρ vac Λ + ρ ind provides the leading RG effect due to the heavy SM particles and we may demand it to vanish as to reduce the fine-tuning in the physical value of the CC at the µ c = O(10 −3 ) eV. This requirement, however, leads to the prediction m H ≈ 550 GeV, inconsistent within the experimental value of m H ≈ 125 GeV.
As discussed in [12], heavy mass terms µ 2 M 2 i may also affect nucleosynthesis if we choose µ ∼ T, because they would induce vacuum energy density ∼ (T 2 M 2 i )/(4π) 2 much bigger than the energy density of radiation ρ rad at the typical energy of nucleosynthesis T ∼ 10 −4 GeV. On the other hand, the m 4 light -and µ 4 -terms obey the constraint ρ rad < ρ vac Λ + ρ ind in the energy interval relevant for nucleosynthesis. To avoid the problem, either we have to use alternative choice µ ∼ H or, again, sufficient amount of fine-tuning should be arranged among the various µ 2 M 2 i terms. Since, as we saw, the heavy SM spectrum does not have this tuning, our results imply that SM has to be extended or µ ∼ H choice is preferred over the µ ∼ T [12].
• Massless case m = 0: In the massless limit of (60) m = 0 ( i.e. all the terms with κ m i absent), from (57) we have ρ vac Λ = const and only ρ ind runs with µ. In this case, the µ 2 -term can be related to the Veltman condition as we now show.
In the massless theory at the tree-level φ = 0, which means that the tree-level mass of the Higgs is zero and the electroweak symmetry needs to be broken radiatevely. For this to happen, we need to balance the tree-level potential against the 1-loop contribution, so that for consistent perturbative expansion we have to impose that the value of the Higgs quartic couplings at the electroweak scale is parametrically given as λ ∼ (g 4 , g 4 , y 4 t ). This allows us to neglect the λ-terms in (60) associated with the Higgs and Goldstones and we obtain (i = W, Z, t and neglecting the light masses m light ): where in the last line we used the fact that in the massless theory with only one background field φ, any mass can be written as M 2 i (φ) = κ i φ 2 . Notice that in the last line the µ 2proportional term is nothing but the generalization of the well known Veltman condition i.e. the requirement of the absence of the quadratic divergence 5 for the Higgs mass (cancellation of the prefactor of the φ 2 -term). This means that within this class of models fine-tuning problem of the Higgs mass is linked to the fine-tuning problem of the Cosmological Constant value 6 .

B. Standard Model with extra massless real scalar
Let us now consider the simplest extension of the SM by adding one extra massless real scalar S: so that contribution from the Higgs background to the mass of the scalar S is given by In this model, the solution to the Veltman condition (61) reads 7 Working in the parameter space of the model where S =0, see [14] for details, leads to the scalar mass M S = √ λ HS v EW ≈ 550 GeV which we already noticed above in the massive version of the SM where the role of the scalar S was played by the Higgs. Remarkably, with the mass of the scalar S satisfying the Veltman condition, we correctly predict the one-loop induced Higgs mass from the Coleman-Weinberg potential This provides an interesting example of how the demand for the absence of leading RG effects in the running of the ρ ind due to the heavy particles may provide the hints on the possible extensions of the SM. Moreover, in this model there is no problem with nucleosynthesis for either of the choices for the RG scale µ ∼ T or µ ∼ H.

C. Standard Model in the constant curvature space
In our final example, we work with the full renormalized version of the Hilbert-Einstein action (2) containing additional coupling constants κ = (16πG) −1 = M 2 pl /2, and non-minimal coupling ξ. We consider the Standard Model in the constant curvature space R µν = (R/4)g µν and working in the linear curvature approximation in Appendix B we show that appropriate generalization of (43) is given by where with parameters κ R i defined in Table.II in Appendix B. Generalizing to the mass-dependent scheme we obtain (see Appendix B for details): where masses m 2 i have corrections from the non-minimal Higgs coupling ξ (44). The result 69) generalizes effective theory expansion (29) to the constant curvature space which also appears via explicit calculations on the expanding cosmological background where vacuum energy is dynamical [15]. The result (69) generalizes the flat space result to possibility of, for example, curvature-induced running of the vacuum energy and curvatureinduced phase transitions [16][17][18][19][20][21].

V. CONCLUSIONS
We revisited the decoupling effects associated with heavy particles in the RG running of the vacuum energy in the mass-dependent renormalization scheme. We derived the universal oneloop beta function of the vacuum energy ρ ind + ρ vac Λ , arising from the Higgs vacuum and the Cosmological Constant term in the entire energy range, valid in the UV and in the IR regime. We have shown that although ρ vac Λ and ρ ind run separately, it is only the sum ρ ind + ρ vac Λ that exhibits behavior consistent with the decoupling theorem.
At the energy scale lower than the mass of the particle, the leading term in the RG running of ρ ind + ρ vac Λ is proportional to the square of the mass of the heavy particle which leads to the enhanced RG running and, consequently, severe fine-tuning problem with the measured value of the Cosmological Constant. We showed that the condition of absence of this leading effect is not satisfied in the SM, while in the massless theories, where Higgs mass is generated radiatively via Coleman-Weinberg mechanism, this constraint formally coincides with Veltman condition. We provided a simple extension of the SM with addition of one massless real scalar where condition of absence of leading effect in β ρ ind allowed us to predict the radiative Higgs mass correctly.
Finally, we also provided the generalization to the constant curvature space in the linear curvature approximation finding the effective field theory expansion that also appears via explicit calculations on the expanding cosmological background. In view of this, our results also might have impact on models based on the dynamical cosmological constant which seem to be favoured by the new cosmological observations [22,23].
To achieve this, we can use the results of [24], where the finite part of (δθ os − δθ MS )| fin was provided in terms of the one-and two-point Passarino-Veltman functions: which are connected as and In the MOM scheme the following relation is valid which stems from the fact that Moreover, the unrenormalized form of A 0 was used in (55). With the expressions above we can easily reconstruct the external momentum dependence of the renormalized form-factor we are looking for. We use the one-loop result for the quartic coupling [24] (notice that all masses are physical): Using (A2), it is easy to show that (A9) leads to: To obtain the corresponding object in the MOM scheme, we only have to reinstate B 0 (M h ; M 1 , M 2 ) → B 0 (p; M 1 , M 2 ), make a finite subtraction at p 2 = −µ 2 and calculate the derivative µ∂/∂µ 8 . After that, we arrive at (52) Similarly, the Higgs mass term is corrected as [24]: (δm 2 os − δm 2  [26].
where we showed only logarithmic term relevant for us and defined with the parameters n i , κ i , κ m i and κ R i shown in Table.II.M 2 i (φ) are the tree-level expressions for the background-dependent and curvature-dependent masses of the particles that enter in the oneloop radiative corrections. Also κ = (16πG) −1 = M 2 pl /2 and ξ is the non-minimal coupling. It is convenient to redefine M 2 i ≡M 2 i − R/6 so that (up to R 2 -terms) We again split the potential to vacuum and φ-dependent pieces V(φ, m 2 , λ i , ρ vac Λ , κ, ξ, µ) = V scal (φ, m 2 , λ i , ξ, µ) + V vac (m 2 , λ i , ρ vac Λ , κ, ξ, µ) .
(B6) and the RG equations (12 -14) get now modified as follows: These equations are valid for any value of φ. However, for the extremum value φ = φ defined via ∂V scal (φ) ∂φ φ = 0, the term containing anomalous dimension of the Higgs γ φ will drop out and we have: Using the tree-level potential (B2), it is useful to define parameter ρ ind (µ) = V 0 ( φ ) = − (m 2 (µ)−ξ(µ)R) 2 2λ(µ) which will play the role similar to the ρ vac Λ (µ). The running of this parameter reads: where SM βand γ-functions are given above in (A14), while with β ξ as reported in [26]. For the extremum value φ = φ we have to drop the γ φ terms from (B13), (B14) and (B17) and combining everything to linear order in R we finally obtain where we used with φ 2 = 2(m 2 −ξR) λ . Notice that for the Goldstones M 2 χ ( φ ) ∼ R and therefore M 4 χ ( φ ) ∼ R 2 so that to linear R order they do not contribute to the sum. The masses m 2 i formally look identical to the flat space analogues M 2 i ( φ ) in (44) but, however, contain the curvature corrections via the φ : Generalizing to the mass-dependent scheme we obtain: Since the mass of the ghost is equal to the corresponding vector boson mass, the ghost cancels the unphysical gauge mode (see Table.II) and we obtain: