A Projection operators

Renormalization of Einstein Gravity Through a Derivative-Dependent Field Redefinition


This work explores an alternative solution to the problem of renormalizability in Einstein gravity. In the proposed approach, Einstein gravity is transformed into the renormalizable theory of four-derivative gravity by applying a field redefinition containing an infinite number of higher derivatives. It is also shown that the current-current amplitude is invariant with the field redefinition, and thus the unitarity of Einstein gravity is preserved.

Quantum gravity; higher derivative gravity; renormalization


1 Introduction

The development of a quantum field theory of gravity based on the Einstein-Hilbert (Einstein) Lagrangian has been problematic because the traditional methods of renormalization cannot be used to eliminate the ultraviolet divergences that appear in perturbation theory [1, 2]. Nonrenormalizable terms appear at two loops [3] or at one loop when coupled to matter [1]. Alternatively, generalizations of the Einstein Lagrangian that include higher-derivative terms, namely and , are renormalizable to all orders in perturbation theory [4, 5, 6, 7, 8], and the dimensionless couplings of the higher-derivative terms are asymptotically free [9, 10, 11]. Also, the essential dimensionless coupling given by the product of the cosmological constant and Newton’s constant is claimed to be asymptotically free [9, 12, 13].

Despite these desirable properties, higher-derivative gravity has a major drawback: in flat-space perturbation theory the higher-derivative terms give rise to a massive spin-two ghost, so the theory is not unitary [4, 7, 8, 14]. It has been suggested that higher-order loop effects may render the massive ghost unstable [6, 15, 16], making the theory unitary for asymptotic states, but a rigorous proof of this is lacking.

It is now understood that the Einstein Lagrangian and its higher-derivative extensions may be regarded as the lowest-order terms in the effective field theory of general relativity [17], the theory containing all generally-covariant functions of the metric and its derivatives [18]. One approach for studying the asymptotic behavior of an effective field theory, referred to as asymptotic safety, is to show that only a subset of the possible couplings are essential, and that they are attracted to a fixed point in the ultraviolet [19, 20, 21, 22, 23]. Non-Gaussian (i.e., non-zero) fixed points have been found by dimensional continuation [19, 24, 25], the approximation [26, 27], the lattice approach [28, 29], and various truncations of the functional renormalization group equation [30, 31, 32, 33, 34].

However, by definition an asymptotically safe effective-field theory of gravity will include higher-derivative terms with essential couplings, so the corresponding -matrix may not be unitary in flat-space perturbation theory [20]. This problem may be avoided if the renormalization group flow near the non-Gaussian fixed point drives the ghost mass to infinity [13, 34]. Another possibility is that the ghost pole in the propagator of the truncated effective Lagrangian is an artifact of the truncation [19, 14]. For example, it has been shown that unitarity arises only when higher-derivative terms of all orders are included [35, 15, 14]. However, because such actions arise from the expansion of entire functions, they are nonlocal. Until these issues are resolved, it is important to develop alternative methods to study quantum effects in gravity.

This paper explores an alternative method of renormalizing Einstein gravity based on field redefinitions. The equivalence theorem states that physical observables such as amplitudes and -matrix elements are independent of field redefinitions [36, 37, 38, 39, 40, 41]. A simple example is a linear field redefinition , which rescales the propagator by and the sources by . Meanwhile, the current-current amplitude, given by the product of two sources and the propagator, is independent of [38, 39]. The property of renormalizability, on the other hand, is determined by the derivative or momentum dependence of the propagator and vertices, which in general is not invariant under field redefinitions [42, 1, 38, 39]. For example, a derivative-dependent field redefinition would alter the momentum dependence of the propagator and vertices, while the current-current amplitude, and thus unitarity, would remain invariant. It follows that a derivative-dependent field redefinition can alter renormalizability without affecting unitarity.

Drawing from recent results, [43] in this paper this invariance is exploited to obtain a quantum theory of Einstein gravity that is both renormalizable and unitary. Specifically, it is shown that Einstein gravity can be transformed into the renormalizable theory of four-derivative gravity by applying a field redefinition that contains an infinite number of higher derivatives. It is further shown that the current-current amplitude, which embodies the property of unitarity, is invariant with the field redefinition. Thus, the field redefintion renders the theory renormalizable while preserving the unitarity of the Einstein theory.

The following calculations assume natural units, a metric signature of (+ - - -), curvature tensor of , Ricci tensor defined by and scalar curvature by , where is the metric tensor.

2 Field redefinition and renormalizability

The classical action of Einstein gravity is


where . Consider the following local field redefinition of the metric:


The Einstein Lagrangian transforms as


where denotes terms with six or more derivatives. Equation (3) becomes


The field redefinition has introduced terms with four derivatives of the metric. This leads to propagator and vertex functions, respectively, which vary as and at large momentum. Therefore, to the transformed Lagrangian is renormalizable in four dimensions. A more detailed proof of the perturbative renormalizability of has been provided by Stelle [4].

Because Einstein gravity is nonpolynomial in the metric, there will be an infinite number of terms in the expansion of Eq. (3). The terms will be cubic and higher in the metric. As a result, the degree of divergence of the vertex functions will be unbounded and the theory will no longer be renormalizable. To maintain renormalizability at higher orders, the field redefinition must be supplemented with additional higher derivative terms as


where is given by Eq. (2). In this case, the Einstein Lagrangian transforms as


As shown above, terms of order transform Einstein gravity into the renormalizable theory of four-derivative gravity. The role of is to cancel the higher derivative vertex functions generated by terms such that the transformed Lagrangian remains equivalent to the Lagrangian of four-derivative gravity. This leads to the condition


which can be solved for to obtain


Using the expression for the second order variation of the Einstein Lagrangian, [7, 44] this can be written as

This procedure can be applied at each order of the transformation to ensure the transformed Lagrangian is equivalent to the renormalizable Lagrangian of four derivative gravity. The end result is a renormalizable theory obtained from a field redefinition containing an infinite number of higher derivative terms.

It may also be possible to transform the Einstein theory into another renormalizable model, such as non-local nonpolynomial gravity [35, 15], or local polynomial superrenormalizable gravity [45, 16, 46, 47]. This leads to potential ambiguity in calculating the quantum corrections. However, according to the equivalence theorem, physical observables such as -matrix elements and beta functions of essential couplings are invariant under arbitrary local field redefinitions [36, 37, 38, 39, 40, 41]. Therefore, in principle, all renormalizable models obtained from the Einstein theory by a local field redefinitions are equally valid. Four-derivative gravity is merely the simplest extension of Einstein gravity sufficient to obtain renormalizability.

3 Propagator

To probe the unitary properties of the theory, it is necessary to derive the propagator. This process is greatly simplified using the momentum space projection operators for symmetric rank 2 tensors described in the appendix, which project out the spin-0, spin-1, and spin-2 components of the field [4, 6]. Taking the gravitational field as , in momentum space the quadratic part of can be written in terms of the projection operators as [4, 13, 9]


In the weak field approximation, the field redefinition in Eq. (2) reduces to


The Lagrangian transforms as


where represents vertices with six or more derivatives. Noting the orthogonality properties of the projection operators, namely , the transformed Lagrangian simplifies to


where and . The invariance of under infinitesimal coordinate transformations of the form


leads to a gauge invariance


which makes the propagator of divergent. This issue is resolved by supplementing with a gauge-fixing term as


A particularly useful gauge which leads to a propagator in which all parts vary as at large momentum is the so-called Julve-Tonin gauge [13, 4, 6, 48]


where is a constant. The total Lagrangian can then be written as [4, 6, 48]




The propagator, obtained by inverting , is then


It can be seen that all parts of the propagator vary as at large momenta.

4 Current-current amplitude and unitarity

The unitarity of the theory can be understood by expanding the propagator into partial fractions. For example, for


The field redefinition has formally introduced additional massive graviton states. The first term corresponds to the massless spin-2 graviton, while the second and third terms, respectively, correspond to massive spin-2 and spin-0 states. Note that for , reduces to the propagator of the Einstein theory. The conditions for unitarity at tree level can be determined from the current-current transition amplitude given by [49, 50, 51, 52]


where is the stress-energy tensor. Unitarity requires the imaginary part of the residue of at the poles to be positive [49, 50, 51, 52]. While the residues of the massless spin-2 and the spin-0 state are positive, the residue of the massive spin-2 state is negative, which would normally violate the unitarity condition. However, noting that the sources couple linearly to the fields as , the linear field redefinition in Eq. (11) also requires the sources in to be redefined as


As a result, the amplitude is invariant under the field redefinition,


That is, only the propagator of the massless spin-2 state appears in the amplitude. Since the imaginary part of the on-shell residue of this portion of the propagator is positive, the unitarity condition is satisfied at tree level.

Beyond tree level, unitarity is preserved provided that the field redefinition is modified to include radiative corrections to the masses. For example, at one-loop order radiative corrections lead to a dressed propagator of the form [9, 16]


where and are the renormalized masses. This dressed propagator is obtained from the field redefinition in Eq. (11) by replacing the bare masses as


Importantly, as long as this replacement is also made in the source redefinition of Eq. (22), the contribution of the massive states to the amplitude in Eq. (23) vanishes, and unitarity is preserved at one-loop order.

In addition to the transformation of the Lagrangian and the redefinition of the source, there is a Jacobian associated with the field redefinition. For local transformations, the Jacobian can be written as a ghost Lagrangian of the form [42, 39]


where and are the ghost fields. Since is linear in the second derivatives of , the ghost acquires a kinetic term but does not couple to the physical field . Therefore, the ghost contributes only an overall contant to the generating functional and thus has no physical effect.

5 Summary

This work aims to develop a quantum theory of gravity that is both unitary and power-counting renormalizable. The approach is to transform Einstein gravity into the renormalizable theory of four-derivative gravity through a field redefinition containing an infinite number of higher derivatives. Importantly, it is also shown that the current-current amplitude is invariant with the field redefinition, and thus the unitarity of the Einstein theory is preserved.

Appendix A Projection operators

The derivation of the graviton propagator is considerably simplified using the momentum space projection operators for symmetric rank 2 tensors. The complete set of projection operators in momentum space is [4, 6]

where and , respectively, are the transverse and longitudinal vector projection operators given by

The orthogonality relations are

where and .


  1. G. ’t Hooft and M. Veltman. One loop divergencies in the theory of gravitation. Ann. Inst. Henri Poincaré, 20:69–94, 1974.
  2. E. Alvarez. Quantum gravity: an introduction to some recent results. Rev. Mod. Phys., 61:561–604, 1989.
  3. M. H. Goroff and A. Sagnotti. Quantum gravity at two loops. Phys. Lett. B, 160:81, 1985.
  4. K. S. Stelle. Renormalization of higher-derivative quantum gravity. Phys. Rev. D, 16:953–969, 1977.
  5. B. L. Voronov and I. V. Tyutin. On renormalization of r2 gravitation. Yad. Fiz.(Sov. Journ. Nucl. Phys.), 39:998, 1984.
  6. I. Antoniadis and E. T. Tomboulis. Gauge invariance and unitarity in higher-derivative quantum gravity. Phys. Rev. D, 33:2756, 1986.
  7. I. L. Buchbinder, S. D. Odintsov, and I. L. Shapiro. Effective Action in Quantum Gravity. IOP Publishing, 1992.
  8. G. de Berredo-Peixoto and I. L. Shapiro. Higher derivative quantum gravity with gauss-bonnet term. Phys. Rev. D, 71:064005– 064020, 2005.
  9. E. S. Fradkin and A. A. Tseytlin. Renormalizable asymptotically free quantum theory of gravity. Nucl. Phys. B, 201:469–491, 1982.
  10. E. Tomboulis. Renormalizability and asymptotic freedom in quantum gravity. Phys. Lett. B, 97:77–80, 1980.
  11. I. G. Avramidi. Asymptotic behavior of the quantum theory of gravity with higher order derivatives. Yad. Fiz, 44:255–263, 1986.
  12. A. Codello and R. Percacci. Fixed points of higher-derivative gravity. Phys. Rev. Lett., 97:221301–221304, 2006.
  13. J. Julve and M. Tonin. Quantum gravity with higher derivative terms. II Nuovo Cimento B Series, 46:137–152, 1978.
  14. E. T. Tomboulis. Renormalization and unitarity in higher derivative and nonlocal gravity theories. Mod. Phys. Lett. A, 30:1540005, 2015.
  15. I. L. Shapiro. Counting ghosts in the ”ghost-free” non-local gravity. Phys. Lett. B, 744:67, 2015.
  16. L. Modesto. Super-renormalizable or finite lee–wick quantum gravity. Nucl. Phys. B, 909:584, 2016.
  17. C. P. Burgess. Quantum gravity in everyday life: General relativity as an effective field theory. Living Rev. Relativity, 7:5–56, 2004.
  18. S. Weinberg. Effective field theory, past and future. arXiv preprint, arXiv:0908.1964:1–24, 2009.
  19. S. Weinberg. Ultraviolet divergences in quantum theories of gravitation. In S.W. Hawking and W. Israel, editors, General Relativity: An Einstein Centenary Survey. Cambridge University Press, Oxford, 1979.
  20. M. Niedermaier and M. Reuter. The asymptotic safety scenario in quantum gravity. Living Rev. Relativity, 9:5–173, 2006.
  21. R. Percacci. Asymptotic safety. In D. Oriti, editor, Approaches to quantum gravity. Cambridge University Press, Cambridge, 2009.
  22. N. Christiansen, D. F. Litim, J. M. Pawlowski, and A. Rodigast. Fixed points and infrared completion of quantum gravity. Phys. Lett. B, 728:114, 2014.
  23. P. Dona, A. Eichhorn, and R. Percacci. Matter matters in asymptotically safe quantum gravity. Phys. Rev. D, 89:084035, 2014.
  24. H. Kawai, Y. Kitazawa, and M. Ninomiya. Ultraviolet stable fixed point and scaling relations in (2+)-dimensional quantum gravity. Nucl. Phys. B, 404:684–714, 1993.
  25. M. Niedermaier. Dimensionally reduced gravity theories are asymptotically safe. Nucl. Phys. B, 673:131–169, 2003.
  26. L. Smolin. A fixed point for quantum gravity. Nucl. Phys. B, 208:439–466, 1982.
  27. R. Percacci. Further evidence for a gravitational fixed point. Phys. Rev. D, 73:041501–041504, 2006.
  28. J. Ambjørn, J. Jurkiewicz, and R. Loll. Emergence of a 4d world from causal quantum gravity. Phys. Rev. Lett., 93:131301–131304, 2004.
  29. J. Ambjørn, A. Görlich, J. Jurkiewicz, and R. Loll. Nonperturbative quantum de sitter universe. Phys. Rev. D, 78:063544–063560, 2008.
  30. M. Reuter. Nonperturbative evolution equation for quantum gravity. Phys. Rev. D, 57:971–985, 1998.
  31. O. Lauscher and M. Reuter. Flow equation of quantum einstein gravity in a higher-derivative truncation. Phys. Rev. D, 66:025026– 025075, 2002.
  32. M. Reuter and F. Saueressig. Renormalization group flow of quantum gravity in the einstein-hilbert truncation. Phys. Rev. D, 65:065016–065041, 2002.
  33. M. R. Niedermaier. Gravitational fixed points from perturbation theory. Phys. Rev. Lett., 103:101303–101306, 2009.
  34. D. Benedetti, P. F. Machado, and F. Saueressig. Asymptotic safety in higher-derivative gravity. Mod. Phys. Lett. A, 24:2233–2241, 2009.
  35. T. Biswas, E. Gerwick, T. Koivisto, and A. Mazumdar. Towards singularity-and ghost-free theories of gravity. Phys. Rev. Lett., 108:031101, 2012.
  36. J. S. R. Chisholm. Change of variables in quantum field theories. Nucl. Phys., 26:469–479, 1961.
  37. R. E. Kallosh and I. V. Tyutin. The equivalence theorem and gauge invariance in renormalizable theories. Yad. Fiz., 17:190–209, 1973.
  38. J. Wudka. Electroweak effective lagrangians. Int. J. Mod Phys B, 9:2301–2361, 1994.
  39. C. Arzt. Reduced effective lagrangians. Phys. Lett. B, 342:189–195, 1995.
  40. M. B. Einhorn and J. Wudka. Effective beta-functions for effective field theory. J. High Energ. Phys., 2001:1–13, 2001.
  41. S. Weinberg. The Quantum Theory of Fields, Volume I: Foundations. Cambridge University Press, 2005.
  42. G. ’t Hooft and M. Veltman. Diagrammar. Technical Report 73-9, CERN, 1973.
  43. B. A. Slovick. Renormalization and asymptotic freedom in quantum gravity through the equivalence theorem. arXiv preprint, arXiv:1309.5945, 2013.
  44. H. W. Hamber. Quantum Gravitation: The Feynman Path Integral Approach. Springer, 2009.
  45. M. Asorey, J. L. Lopez, and I. L. Shapiro. Some remarks on high derivative quantum gravity. Int. J. Mod. Phys. A, 12:5711, 1997.
  46. L. Modesto and I. L. Shapiro. Superrenormalizable quantum gravity with complex ghosts. Phys. Lett. B, 755:279, 2016.
  47. L. Modesto, L. Rachwal, and I. L. Shapiro. Renormalization group in super-renormalizable quantum gravity. arXiv preprint, arXiv:1704.03988:1–16, 2017.
  48. A. Accioly, S. Ragusa, H. Mukai, and E. de Rey Neto. Algorithm for computing the propagator for higher derivative gravity theories. Int. J. Theo. Phys., 39:1599, 2000.
  49. F. C. P. Nunes and G. O. Pires. Extending the barnes-rivers operators to d= 3 topological gravity. Phys. Lett. B, 301:339, 1993.
  50. C. Pinheiro, G. O. Pires, and N. Tomimura. Some quantum aspects of three-dimensional einstein-chern-simons-proca massive gravity. II Nuovo Cimento B, 111:1023, 1996.
  51. A. Accioly, A. Azeredo, and H. Mukai. Propagator, tree-level unitarity and effective nonrelativistic potential for higher-derivative gravity theories in d dimensions. J. Math. Phys., 43:473, 2002.
  52. S. F. Hassan, R. A. Rosen, and A. Schmidt-May. Ghost-free massive gravity with a general reference metric. J. High Energy Phys., 2:1, 2012.
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minumum 40 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description