You are on page 1of 10

arXiv:physics/9807056v1 [physics.

class-ph] 30 Jul 1998

When a mass term does not represent a mass
V. Faraoni1 and F.I. Cooperstock2
1 Inter–University

Centre for Astronomy and Astrophysics

Post Bag 4, Ganeshkhind PO, Pune 411 007, India
2 Department

of Physics and Astronomy, University of Victoria

P.O. Box 3055, Victoria, B.C. Canada V8W 3P6

The definition of mass of a scalar field in a curved space has often
been generalized by grouping coupling terms between the field and the
Ricci curvature with non–curvature–related mass terms. In a broader
point of view, one sees that a common misunderstanding resulting
from such an identification leads one, in the case of the spin 2 field,
to regard the cosmological constant as a non–vanishing mass of cosmological origin for the graviton. Similarly, there are inconsistencies
for the spin 1 field. Instead, the intrinsic mass of a field should be
regarded as being independent of the background curvature.

To appear in European Journal of Physics.


1 Introduction The generalization of the Klein–Gordon equation for a scalar field φ on a curved spacetime [1] 2φ − m2 φ − ξRφ = 0 (1. Despite some attempts [8. For the case of spin 2 fields (gravitational waves). (1.2) is sometimes introduced.2) has led to problems of interpretation. in the wave equations for the electromagnetic field and for gravitational waves propagating on a curved background spacetime. the identification of field– curvature coupling terms with mass terms frequently led to attribute to the graviton a mass of cosmological origin related to the cosmological constant [2]–[7]. and is a guide for the study of more complicated wave equations. In this case. 5]. particularly when the Ricci curvature of the background spacetime is constant. (1.1).1) is the prototype of the wave equation on a curved spacetime. 2 . direct couplings with the Ricci and Riemann tensors appear. Various authors have grouped the scalar field– curvature coupling term ξRφ with the mass term m2 φ in Eq. The interpretation of the terms coupling the field with the curvature as mass terms has also been extended to wave equations for fields of higher spin. Eq. Equation (1. The interpretation of terms like ξRφ in Eq. the past and recent misunderstandings on this issue still await clarification in the literature. respectively.1) presents the possibility of a non–minimal coupling between the scalar field and the Ricci curvature R of spacetime. In fact.1) as mass terms leads to properties of the “mass” µ which are physically unacceptable. Analogously. (1. The proper interpretation is to regard the mass of the field as being independent of the background curvature. an effective mass µ given by µ2 = m2 + ξR (1.

(1. The second contribution has support inside the light cone.1) (with m = 0) is not conformally invariant. the tail–free propagation property experienced by scalar waves in flat space is transferred to the conformally flat de Sitter space. If ξ 6= 1/6. It may appear that. particles with nonzero mass necessarily propagate strictly inside the light cone. (1.3) propagate on the light cones of de Sitter space.1) with m = 0 and ξ = 1/6 in this space: 2φ − 1 Rφ = 0 . whatever definition of mass is adopted in curved spacetimes. It appears reasonable to require that. arbitrarily large cosmological constant Λ). The definition (1.3) is conformally invariant and the de Sitter universe is conformally flat: therefore the second contribution to the Green function G(x′ . (1. (1.1). The definition given by Eq. Consider the de Sitter space with arbitrarily large (constant) Ricci curvature (i. In general. Equation (2. and consider Eq. (2. Eq. (2.3) Equation (1. x) of Eq.2) gives a value of the effective mass µ which is arbitrarily large. in this case. In other words.2 The Klein–Gordon field Intuitively. one would be inclined to consider the mass of a particle as an intrinsic characteristic which does not depend on whether the particle is in flat space or in curved space.3) vanishes (no tails) and the waves propagate strictly on the light cones [14]–[17]. the conclusions reached using our example with ξ = 1/6 are not valid and.1) is. However. x) of Eq. at most. and describes timelike propagation of waves. we have an arbitrarily large effective mass µ = (R/6)1/2 . for a general given spacetime. and is associated with tails (violations of Huygens’ principle [9]–[12] – see [13] for a pedagogical introduction).2) of mass of a scalar field in curved spacetime does not have this property. the sum of two contributions. Therefore. the first contribution has support on the light cone. but the solutions of Eq.3) propagate on light cones. the solutions of the wave equation (2. (1. The Green function G(x′ . and describes lightlike propagation of the waves.2) clearly does not satisfy this requirement. the previous example suggests some 3 . This result is well known in the investigations of Huygens’ principle in curved spacetimes and of “tails” of radiation [9]–[12]. the second contribution is different from zero even when m = 0 in Eq.e. 6 (2.

4) that the cosmological constant provides a mass for the photon field. (1.1) ∇[µ Fνρ] = 0 . the Einstein equivalence principle [18] applied to the scalar field φ restricts the value of ξ to be 1/6. in the absence of sources. (3. (3. In the Lorentz gauge ∇α Aα = 0. In other words.2) and using Eq.4) According to the identification underlying Eq.2). The previous example shows clearly the unphysical properties of the mass µ defined by Eq. the photon field can be described using the four–vector potential Aµ . In this space.3) reduces to 2Aµ − ΛAµ = 0 .2). (3. (3.3) We consider again the de Sitter universe. 4 (3. in which the Ricci tensor is given by Rρσ = Λgρσ . (3.2). (3.2) Alternatively. (1. (1. satisfies the curved space Maxwell equations: ∇ν Fµν = 0 .2).caution in the use of Eq. Further support for this identification is given in the next section. (3. (1. the term “mass” of the scalar field φ in Eq. a result that is now well established [19. 21]. 3 The electromagnetic field We now consider the Maxwell field tensor Fµν which. Eq. Aµ satisfies the wave equation 2Aµ − Rµν Aν = 0 . 20. One can actually derive a wave equation for the Maxwell tensor by applying the operator ∇µ to Eq. rejecting the alternative definition (1.1): 2Fµν + 2Rαµνβ F βα + 2Rβ[µ Fν] β = 0 . one would conclude from Eq. the intrinsic mass is independent of the background curvature.1) should be used exclusively for the coefficient m. (3.5) . Therefore. However.

one would read a √ in the case of a spin 1 field in de Sitter spacetime.7) for the field Fµν would provide a mass 4Λ/3. which can be seen as follows: The Maxwell equations (3. However. (1. (3. we know that the Maxwell field is massless according to any useful definition of mass for a field in a curved space. (3.10) ∂µ F µν = −m2 Aν . it is easy to derive a wave equation for Fµν : 5 . (3. leading to a “mass” (4Λ/3)1/2 for the photon field. Since the propagation of electromagnetic waves in this case is restricted to the light cones. This conclusion is incorrect. the equations for the Proca 4–potential are ∂ µ Aµ = 0 . the electromagnetic field propagates on the light cone [14.10) and (3.11) 2Fµν − m2 Fµν = 0 .1). so that the “mass” depends on whether one chooses to consider the potential or the field.2) in four dimensions are conformally invariant and the de Sitter space is conformally flat.2) leads to another inconsistency. 3 (3. the wave equamass Λ from Eq. µ .8) 2Aµ − m2 Aµ = 0 . (3.5) to in which a “mass term” appears. (3. 16. as expressed in Eq. (3. According to this identification. (3.7) which reduces Eq.11) and from the identity ∂[ρ Fµν] = 0. The Proca field Fµν satisfies ¿From Eqs. Thus. (3.12) where m is the mass of the Proca potential.In the de Sitter space one has Rµνρσ = 2Λ gµ[ρ gσ]ν . 17]. 3 (3. In flat space.4) for the vector potential Aq tion (3. The identification of the linear term in the field as a mass term.6) 2Fµν − 4Λ µν F =0. (3. This situation is markedly different from the case of the massive spin 1 field in Minkowski space.9) Fµν ≡ ∂µ Aν − ∂ν Aµ .

By contrast. A Yukawa potential is also obtained by postulating a finite range of gravitation arising from a nonvanishing graviton mass [4]. and there is no trace of the ambiguity encountered in Eqs. due to the non–commutativity of covariant derivatives. The presence of terms coupling the field tensor with the Riemann and Ricci tensors follows from the fact that. contractions of the field Fµν with the Riemann tensor inevitably appear in the wave equation for Fµν in addition to the term 2Fµν . This conclusion is incorrect and. Previous claims [8. However.12) provide the same value for the mass of the Proca field. the issue has not been clarified in the literature. An exact solution 6 .9) and (3.7).4) and (3. This feature has been known for many years and its astrophysical consequences have been explored [22]. The old argument supporting the idea of the cosmological constant as endowing the graviton with a mass relies on the fact that the weak field limit of the Einstein equations with a cosmological constant produces a Yukawa (instead of a Coulombic) potential [2]–[4]. one can take the combination of covariant derivatives ∇µ 2Aν − ∇ν 2Aµ to generate 2Fµν . (3. in a flat space. It has been explicitly stated or implicitly suggested that the cosmological constant endows the graviton with a rest mass [2]–[7]. 4 Gravitational waves A recurring point in the literature involves the effect of the cosmological constant in determining the character of gravitational waves which propagate in curved spacetimes. ∇µ 2Aν − ∇ν 2Aµ is precisely 2Fµν and hence the mass which is identified from the equation for the 4–vector potential is the same as the mass that is identified from the equation for Fµν . in the derivation of the wave equation for the field tensor Fµν in a curved space.Equations (3. 5] against the identification of the cosmological constant as the mass of the graviton [23] proceed as follows. despite some attempts [8].

Eq.3) (where h ≡ g (0)µν hµν ) gives 2hµν − 2Rαµνβ hαβ + Rµρ hν ρ + Rνρ hµ ρ − 2Λhµν = 0 . (4.4) The last term in the left hand side of Eq. The claim that this proves that the propagation describes a massless field on this basis is unfounded: if one proceeds to substitute the form (3. [24]). in Ref. (4. and one is presented with a quandary. Equation (4. one obtains 2hµν − 2Λ hµν = 0 .4) is susceptible to being interpreted as a mass term. This solution was perturbed as (0) gµν = gµν + hµν . 8. 3 (4.1) with the gauge choice ∇ ν  hµν 1 (0) h =0 − gµν 2  (4. the substitution has removed all trace of Λ and the propagation equation is reduced to the same form as it would have in a spacetime free of any cosmological constant.2) (0) where hµν (|hµν | ≪ |gµν |) describe gravitational waves. (4.4) is reduced to 2hµν − 2Rαµνβ hαβ = 0 . Rµν − 1 gµν R + Λgµν = 0 2 (4.6) agrees with Eq. e.1) was considered.g.3) and the additional constraint h = 0 (the propagation equations for ∇ν hµν .5) According to Ref. (4. 7 .21) of Ref. h and a proof that the constraints ∇ν hµν = 0.6) in which a “mass term” re–appears in the wave equation for gravitational waves. The linearization of the Einstein equations (4.6) of the Riemann tensor. However. h = 0 can be imposed in a globally hyperbolic spacetime can be found. in Ref. (4. 8 it is noted that Rµρ hν ρ + Rνρ hµ ρ = 2Λhµν .(0) gµν of the Einstein equations with cosmological constant and no matter. Using this result. (2.. 21 and is obtained by imposing (4.

While the presence of Λ affects the propagation of scalar. The connection between the cosmological constant and the mass of the graviton is excluded. (1.2) may be useful from the mathematical point of view in some cases.6) as a mass term is inappropriate. it is seen that the correct conclusion was reached in Ref. it should not be identified with the physical mass of the field. 8 .Armed with our argument from the previous section. since such an identification would lead to unacceptable physical properties for the field. in part. Acknowledgment This work was supported. it does not endow the fields with intrinsic mass. Although the introduction of the quantity µ in Eq. Thus. vector and tensor fields in curved spacetime. but the proof was incomplete. (4. 8. we recognize that the identification of the second term in the left hand side of Eq. by the Natural Sciences and Engineering Research Council of Canada.

695–708. 220–259. J. 413–420. Class. [15] Deser S and Nepomechie R I 1984. Inst. [7] Curtis Michel F 1996. Phys. 2. 1087–1092. [4] Freund P G O. 153–188. Greek indices assume the values 0. and 2 ≡ gµν ∇µ ∇ν . Int. (NY) 9. 1. Astrophys. [10] De Witt B S and Brehme R W 1960. The Riemann tensor is given in terms of the Christoffel symbols by Rµνρ σ = Γσµρ. Grav. J. Phys. [3] Bergmann P G 1968. Found. 157. 625–632. Ann. 167–169. 12. Latin indices assume the values 1. Lett. [12] G¨ unther P 1988. Lectures on Cauchy’s Problem in Linear Partial Differential Equations (New York: Dover). A 66. [5] Treder H–J and Yourgrau W 1978.References [1] The metric signature is – + + +.. Phys. 466.. [9] Hadamard J 1952. Pa˘ vsi˘c M and Recami E 1978. [11] Friedlander F G 1975. (NY) 154. Phys. [8] Treder H–J 1968. Ann. [2] Tonnelat M A 1965. Phys. Maheshwari A and Schonberg E 1969.ν −Γσνρ. 8. The Wave Equation on a Curved Spacetime (Cambridge: Cambridge University Press). [17] Noonan T W 1995. Astrophys.µ +Γαµρ Γσαν −Γανρ Γσαµ . J. 1. 396–420. Lett. The Ricci tensor is Rµρ ≡ Rµνρ ν . Les Th´eories Unitaires de l’Electromagnetism et de la Gravitation (Paris: Gauthier–Villars). 3. 9–12. 857–867. Phys. 1. Ann. A 170. Math. 2. 33. Henri Poincare 20. [13] Faraoni V and Sonego S 1992. [14] McLenaghan R G 1974. 660–667. Phys. 3. ∇µ is the covariant derivative operator. Phys. [16] Sonego S and Faraoni V 1991. Quant. 9 . J. Huygens’ Principle and Hyperbolic Equations (New York: Academic Press). [6] Caldirola P. 25–36. Theor. Theor. Int. J. Round [square] brackets denote [anti]symmetrization.

The second argument relies on the perturbation of a flat background. [24] Higuchi A 1989. Grav. 518–529. 1. Grav. Phys. 10 .1) if Λ 6= 0. Publ. [21] Grib A A and Rodrigues W A 1995. Acta 68. Pac. 6. 1185–1187. 397–406. Class. [19] Sonego S and Faraoni V 1993. Quant.[18] Will C M 1993. Helv. different from the one that we consider in this paper. 273–276. [20] Grib A A and Poberii E A 1995. Cosmol. 380–395. [22] Zwicky F 1957. Grav. which is unacceptable since the flat metric that is perturbed is not a solution of Eq. Class. 69. 10. Astron. [8]. a second argument is presented. (4. Quant. Soc. Theory and Experiment in Gravitational Physics (Cambridge: Cambridge University Press). [23] In Ref.