Classical and Quantum Gravity TOPICAL REVIEW Testing general relativity with present and future astrophysical observations To cite this article: Emanuele Berti et al 2015 Class. Quantum Grav. 32 243001   View the article online for updates and enhancements. Related content Black hole based tests of general relativity Kent Yagi and Leo C Stein - I-Love-Q relations for neutron stars in dynamical Chern Simons gravity Toral Gupta, Barun Majumder, Kent Yagi et al. - TESTING UNIVERSAL RELATIONS OF NEUTRON STARS WITH A NONLINEAR MATTER GRAVITY COUPLING THEORY Y.-H. Sham, L.-M. Lin and P. T. Leung - Recent citations Spontaneous tensorization from curvature coupling and beyond Fethi M. Ramazanolu - Scalar field as a null dust Valerio Faraoni and Jeremy Côté - Global monopole in Palatini f(R) gravity J. R. Nascimento et al - This content was downloaded from IP address 186.217.236.55 on 23/04/2019 at 21:03 https://doi.org/10.1088/0264-9381/32/24/243001 http://iopscience.iop.org/article/10.1088/0264-9381/33/5/054001 http://iopscience.iop.org/article/10.1088/1361-6382/aa9c68 http://iopscience.iop.org/article/10.1088/1361-6382/aa9c68 http://iopscience.iop.org/article/10.1088/0004-637X/781/2/66 http://iopscience.iop.org/article/10.1088/0004-637X/781/2/66 http://iopscience.iop.org/article/10.1088/0004-637X/781/2/66 http://dx.doi.org/10.1103/PhysRevD.99.084015 http://dx.doi.org/10.1103/PhysRevD.99.084015 http://dx.doi.org/10.1140/epjc/s10052-019-6829-x http://dx.doi.org/10.1103/PhysRevD.99.064053 https://oasc-eu1.247realmedia.com/5c/iopscience.iop.org/510997659/Middle/IOPP/IOPs-Mid-CQG-pdf/IOPs-Mid-CQG-pdf.jpg/1? Topical Review Testing general relativity with present and future astrophysical observations Emanuele Berti1,2, Enrico Barausse3,4, Vitor Cardoso2,5, Leonardo Gualtieri6, Paolo Pani2,6, Ulrich Sperhake1,7,8, Leo C Stein8,9, Norbert Wex10, Kent Yagi11,12, Tessa Baker13, C P Burgess5,14, Flávio S Coelho15, Daniela Doneva16, Antonio De Felice17,18, Pedro G Ferreira13, Paulo C C Freire10, James Healy19, Carlos Herdeiro15, Michael Horbatsch1, Burkhard Kleihaus20, Antoine Klein1, Kostas Kokkotas16, Jutta Kunz20, Pablo Laguna21, Ryan N Lang22,23,24, Tjonnie G F Li25,26, Tyson Littenberg27, Andrew Matas28, Saeed Mirshekari29, Hirotada Okawa2, Eugen Radu15, Richard O’Shaughnessy19,22, Bangalore S Sathyaprakash30, Chris Van Den Broeck31, Hans A Winther13, Helvi Witek7, Mir Emad Aghili1, Justin Alsing32, Brett Bolen33, Luca Bombelli1, Sarah Caudill22, Liang Chen1, Juan Carlos Degollado15, Ryuichi Fujita2, Caixia Gao1, Davide Gerosa7, Saeed Kamali1, Hector O Silva1, João G Rosa15, Laleh Sadeghian22, Marco Sampaio15, Hajime Sotani34 and Miguel Zilhao19 1Department of Physics and Astronomy, The University of Mississippi, University, MS 38677-1848, USA 2CENTRA, Departamento de Física, Instituto Superior Técnico, Universidade de Lisboa, Avenida Rovisco Pais 1, 1049 Lisboa, Portugal 3 CNRS, UMR 7095, Institut d’Astrophysique de Paris, 98bis Bd Arago, 75014 Paris, France 4 Sorbonne Universités, UPMC Univ Paris 06, UMR 7095, 98bis Bd Arago, 75014 Paris, France 5 Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L 2Y5, Canada 6Dipartimento di Fisica, “Sapienza” Università di Roma & Sezione INFN Roma 1, P.le A. Moro 2, 00185 Roma, Italy 7 Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, UK 8 Theoretical Astrophysics 350-17, California Institute of Technology, Pasadena, CA 91125, USA 9Cornell Center for Astrophysics and Planetary Science, Cornell University, Ithaca, NY 14853, USA 10Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, D-53121 Bonn, Germany 11Department of Physics, Princeton University, Princeton, NJ 08544, USA Classical and Quantum Gravity Class. Quantum Grav. 32 (2015) 243001 (179pp) doi:10.1088/0264-9381/32/24/243001 0264-9381/15/243001+179$33.00 © 2015 IOP Publishing Ltd Printed in the UK 1 http://dx.doi.org/10.1088/0264-9381/32/24/243001 12 Department of Physics, Montana State University, Bozeman, Montana 59717, USA 13Astrophysics, University of Oxford, DWB, Keble Road, Oxford, OX1 3RH, UK 14Department of Physics & Astronomy, McMaster University, Hamilton ON, Canada 15Departamento de Física da Universidade de Aveiro and CIDMA Campus de Santiago, 3810-183 Aveiro, Portugal 16 Theoretical Astrophysics, Eberhard Karls University of Tübingen, Tübingen 72076, Germany 17 ThEPʼs CRL, NEP, The Institute for Fundamental Study, Naresuan University, Phitsanulok 65000, Thailand 18 Thailand Center of Excellence in Physics, Ministry of Education, Bangkok 10400, Thailand 19 Center for Computational Relativity and Gravitation, School of Mathematical Sciences, Rochester Institute of Technology, 85 Lomb Memorial Drive, Rochester, NY 14623, USA 20 Institut für Physik, Universität Oldenburg, D-26111 Oldenburg, Germany 21 Center for Relativistic Astrophysics and School of Physics, Georgia Institute of Technology, Atlanta, GA 30332, USA 22 Leonard E Parker Center for Gravitation, Cosmology, and Astrophysics, University of Wisconsin-Milwaukee, Milwaukee, WI 53211, USA 23Department of Physics, University of Florida, Gainesville, FL 32611, USA 24Department of Physics, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA 25Department of Physics, The Chinese University of Hong Kong, Shatin, N. T., Hong Kong, People’s Republic of China 26 LIGO — California Institute of Technology, Pasadena, CA 91125, USA 27 Center for Interdisciplinary Exploration and Research in Astrophysics (CIERA) & Dept. of Physics and Astronomy, Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208, USA 28 CERCA/Department of Physics, Case Western Reserve University, 10900 Euclid Ave, Cleveland, OH 44106, USA 29 ICTP South American Institute for Fundamental Research & Instituto de Física Teórica, UNESP — Universidade Estadual Paulista, Rua Dr. Bento T. Ferraz 271 — 01140-070 São Paulo, SP, Brazil 30 School of Physics and Astronomy, Cardiff University, 5, The Parade, Cardiff CF24 3AA, UK 31Nikhef, Science Park, 1098 XG Amsterdam, The Netherlands 32 Imperial Centre for Inference and Cosmology, Imperial College London, Prince Consort Road, London SW7 2AZ, UK 33Department of Physics, Grand Valley State University, Allendale, MI 49401- 9403, USA 34Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan E-mail: eberti@olemiss.edu Received 9 February 2015, revised 5 June 2015 Accepted for publication 3 August 2015 Published 1 December 2015 Abstract One century after its formulation, Einsteinʼs general relativity (GR) has made remarkable predictions and turned out to be compatible with all experimental tests. Most of these tests probe the theory in the weak-field regime, and there are theoretical and experimental reasons to believe that GR should be modified when Class. Quantum Grav. 32 (2015) 243001 Topical Review 2 mailto:eberti@olemiss.edu http://crossmark.crossref.org/dialog/?doi=10.1088/0264-9381/32/24/243001&domain=pdf&date_stamp=2015-12-01 http://crossmark.crossref.org/dialog/?doi=10.1088/0264-9381/32/24/243001&domain=pdf&date_stamp=2015-12-01 gravitational fields are strong and spacetime curvature is large. The best astro- physical laboratories to probe strong-field gravity are black holes and neutron stars, whether isolated or in binary systems. We review the motivations to consider extensions of GR. We present a (necessarily incomplete) catalog of modified theories of gravity for which strong-field predictions have been com- puted and contrasted to Einsteinʼs theory, and we summarize our current understanding of the structure and dynamics of compact objects in these theories. We discuss current bounds on modified gravity from binary pulsar and cosmo- logical observations, and we highlight the potential of future gravitational wave measurements to inform us on the behavior of gravity in the strong-field regime. Keywords: general relativity, black holes, neutron stars, compact binaries, gravitational waves Contents 1. Introduction 5 1.1. Taxonomy of proposed extensions of GR . . . . . . . . . . . . . . . . . . . . . . . 7 1.2. Compact objects in modified theories of gravity . . . . . . . . . . . . . . . . . . 11 1.3. Present and future tests of strong gravity . . . . . . . . . . . . . . . . . . . . . . . 12 2. Extensions of GR: motivation and overview 14 2.1. A compass to navigate the modified-gravity atlas . . . . . . . . . . . . . . . . . 14 2.2. Scalar–tensor gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.1. The Bergmann–Wagoner formulation . . . . . . . . . . . . . . . 18 2.2.2. Scalar–tensor theories with multiple scalar fields . . . . . . . 20 2.2.3. Horndeski gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.3. Metric f(R) theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.4. Quadratic gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.1. EdGB gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.4.2. Chern–Simons gravity. . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.5. Lorentz-violating theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.5.1. Einstein-Æther . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.5.2. Khronometric theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 2.5.3. Horǎva gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.5.4. n-DBI gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.6. Massive gravity and Galileons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.7. Gravity with auxiliary fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.8. GR and quantum mechanics: an EFT approach. . . . . . . . . . . . . . . . . . . 38 2.8.1. Power-counting and the semiclassical approximation. . . . . 40 2.8.2. Modified gravity seen through EFT glasses . . . . . . . . . . . 42 2.9. Open problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3. Black holes 46 3.1. BHs in GR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Class. Quantum Grav. 32 (2015) 243001 Topical Review 3 3.2. Scalar–tensor theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.2.1. Real scalars and no-hair theorems. . . . . . . . . . . . . . . . . . 47 3.2.2. Complex scalars: new hairy rotating BHs . . . . . . . . . . . . 48 3.2.3. Evading no-hair theorems in Horndeski/Gauss–Bonnet gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 3.2.4. BHs surrounded by matter . . . . . . . . . . . . . . . . . . . . . . . 52 3.2.5. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.3. f(R) theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.4. Quadratic gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.4.1. Perturbative solutions in the slow-rotation limit . . . . . . . . 53 3.4.2. EdGB theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.4.3. dCS theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.5. Lorentz-violating theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 3.6. Massive gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.7. Gravity with auxiliary fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 3.8. Parametrized phenomenological deviations from the Kerr metric . . . . . . . 64 3.9. BH mimickers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.10. BHs as strong-gravity laboratories for exotic fields . . . . . . . . . . . . . . . 68 3.10.1. Collapse of self-interacting scalar fields . . . . . . . . . . . . . 68 3.10.2. Superradiant instabilities: BHs as observatories for beyond- standard-model physics . . . . . . . . . . . . . . . . . . . . . . . . 70 3.11. Open problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 4. Neutron stars 74 4.1. General-relativistic stellar models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.2. Scalar–tensor theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 4.3. f(R) theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.4. Quadratic gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 4.4.1. EdGB theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.4.2. dCS theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 4.5. Lorentz-violating theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4.6. Massive gravity and Galileons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.7. Gravity with auxiliary fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 4.8. Strong-field tests of gravity with universal relations in NSs and quark stars (QSs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 4.9. NS sensitivities in modified gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.10. Open problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 5. Compact binaries 104 5.1. Scalar–tensor theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 5.1.1. Analytical calculations . . . . . . . . . . . . . . . . . . . . . . . . 104 5.1.2. Numerical relativity simulations . . . . . . . . . . . . . . . . . . 109 5.2. f(R) theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 5.3. Quadratic gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5.4. Lorentz-violating theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.5. Massive gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.6. Open problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Class. Quantum Grav. 32 (2015) 243001 Topical Review 4 6. Binary pulsar and cosmological tests of GR 123 6.1. Tests of gravity from radio pulsars . . . . . . . . . . . . . . . . . . . . . . . . . . 123 6.1.1. Open problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 6.2. Testing GR with cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 6.2.1. Theory: the linear regime . . . . . . . . . . . . . . . . . . . . . . 131 6.2.2. Theory: the nonlinear regime . . . . . . . . . . . . . . . . . . . . 134 6.2.3. Observations, current and future . . . . . . . . . . . . . . . . . . 138 6.2.4. Open problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 7. Gravitational wave tests 139 7.1. Science opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 7.2. Parameter estimation and model selection . . . . . . . . . . . . . . . . . . . . . 142 7.3. Direct versus parametrized tests of gravity . . . . . . . . . . . . . . . . . . . . . 143 7.3.1. Implementation of direct tests: the TIGER pipeline . . . . . 147 7.4. Waveform and astrophysical systematics . . . . . . . . . . . . . . . . . . . . . . 153 7.4.1. Stellar mass objects . . . . . . . . . . . . . . . . . . . . . . . . . . 153 7.4.2. Supermassive BHs . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 8. Discussion and conclusions 160 1. Introduction Einsteinʼs theory of general relativity (GR), together with quantum mechanics, is one of the pillars of modern physics. The theory has passed all precision tests to date with flying colors. Most of these tests—with the possible exception of binary pulsar observations—are probes of weak-field gravity; more precisely, they probe gravity at intermediate length (1 μm ℓ 1 AU ∼ 1011 m) and therefore intermediate energy scales. Laboratory experi- ments and astrophysical observations verify the so-called ‘Einstein equivalence principle’ (i.e. the weak equivalence principle supplemented by local Lorentz invariance and local position invariance) and they set constraints on hypothetical weak-field deviations from GR, as encoded in the parametrized post-Newtonian (PPN) formalism (see [1] for an introduction, and [2] for a review of the state of the art on experimental tests of GR). The conceptual foundations of GR are so elegant and solid that when asked what he would do if Eddingtonʼs expedition to the island of Principe failed to match his theory, Einstein famously replied: ‘I would feel sorry for the good Lord. The theory is correct.’ Chandrasekhar made a similar private remark to Clifford Will when Will was a postdoc in Chicago: ‘Why do you spend so much time and energy testing GR? We know that the theory is right.’ Giving up the fundamental, well tested principles underlying Einsteinʼs theory has dramatic consequences, often spoiling the beauty and relative simplicity of Einsteinʼs theory. However, there is growing theoretical and experimental evidence that modifications of GR at small and large energies are somehow inevitable. From a theoretical point of view, GR is a purely classical theory. Power-counting arguments indicate that GR is not renormalizable in the standard quantum field theory sense. Strong-field modifications may provide a solution to this problem: it has long been known that the theory becomes renormalizable if we add quadratic curvature terms—i.e., high- energy/high-curvature corrections—to the Einstein–Hilbert action [3]. Furthermore, high- energy corrections can avoid the formation of singularities that are inevitable in classical GR, as shown by the Hawking–Penrose singularity theorems [4]. Candidate theories of quantum Class. Quantum Grav. 32 (2015) 243001 Topical Review 5 gravity (such as string theory and loop quantum gravity) make specific and potentially tes- table predictions of how GR must be modified at high energies. From an observational point of view, cosmological measurements are usually interpreted as providing evidence for dark matter and a nonzero cosmological constant (‘dark energy’). This interpretation poses serious conceptual issues, including the cosmological constant problem (‘Why is the observed value of the cosmological constant so small in Planck units?’) and the coincidence problem (‘Why is the energy density of the cosmological constant so close to the present matter density?’). No dynamical solution of the cosmological constant problem is possible within GR [5]. It seems reasonable that ultraviolet corrections to GR would inevitably ‘leak’ down to cosmological scales, showing up as low-energy (infrared) corrections. The arguments summarized above suggest that GR should be modified at both low and high energies. This is a serious challenge for theorists. Einsteinʼs theory is the unique interacting theory of a Lorentz-invariant massless helicity-2 particle [6, 7], and therefore new physics in the gravitational sector must introduce additional degrees of freedom. Any addi- tional degrees of freedom must modify the theory at low and/or high energies while being consistent with GR in the intermediate-energy regime, i.e. at length scales 1 μm ℓ 1011 m, where the theory is extremely well tested. Laboratory, Solar System and binary-pulsar experiments verify the Einstein equivalence principle to remarkable accuracy; they force PPN parameters to be extremely close to their GR values; and (as we will see below) they place stringent bounds on popular extensions of GR, such as scalar–tensor theories and Lorentz- violating theories (see [2, 8] for reviews). Some confusion exists about how to link tests of gravity that take place in the very different regimes described above. For example, though it is agreed that strong-field constraints on GR do not rule out cosmological modifications (or vice versa), it is not immediately obvious how to express this statement quantitatively, except perhaps in specific models. One method of resolving this problem was recently put forward in [9]. There the authors place a wide range of laboratory, astrophysical and cosmological systems on a two-dimensional parameter space, where the axis quantities are the approximate gravitational potential and Kretschmann scalar of a system. The Kretschmann scalar is used because it gives a rough measure of how relativistic the system is and does not vanish in vacuum (the diagnostic power of the Ricci scalar is limited for this reason). Many orders of magnitude of Kretschmann curvature separate the classic PPN tests of gravity from both the strong-field regime and the cosmological regime, so we cannot simply take existing Solar System constraints as comprehensive. The main focus of this review is on present and future tests of strong-field gravity. It is useful to classify ‘tests of strong-field gravity’ as belonging to two qualitatively different categories. ‘External tests’ are laboratory experiments, astrophysical and cosmological observations that can be used to determine whether GR (as opposed to any of the numerous proposed extensions) is the correct theory of gravity. ‘Internal tests’ are observations that tell us whether some key predictions of GR (e.g. the Kerr solution of the Einstein equations in vacuum, or the radiative dynamics of compact objects) are ‘internally’ consistent with astrophysical observations. Compact objects such as black holes (BHs) and neutron stars (NSs) are our best natural laboratories to constrain strong gravity. In these celestial bodies gravity prevails over all other interactions, and collapse leads to large-curvature, strong-gravity environments (see e.g. [10]). The Kerr metric is a solution of the vacuum field equations in a large class of modified gravity theories, but theories that differ from GR generically predict different dynamics and different gravitational-wave (GW) signatures when compact objects are displaced from equilibrium and/or when they merge. This is the reason why a large part of this review will be devoted to the structure and dynamics of BHs and NSs, whether isolated or in binary systems. Class. Quantum Grav. 32 (2015) 243001 Topical Review 6 1.1. Taxonomy of proposed extensions of GR To frame external tests in terms of hypothesis testing, one would like to have one or more valid alternatives to GR. What constitutes a ‘valid alternative’ is, of course, a matter of taste. From our perspective (i.e., in terms of tests of strong-field gravity) the alternative should be a cosmologically viable fundamental theory passing intermediate energy tests, with a well- posed initial value formulation, and field equations that follow from an action principle. Furthermore, the theory should be simple enough to make definite, calculable prediction in the strong-field regime: ideally, it should allow us to predict the structure and dynamics of compact objects and the gravitational radiation that they emit, whether isolated or in binary systems. This is a very stringent set of requirements. There are countless attempts to modify GR [11–17], but (for the reasons listed above) in several cases the modifications introduce some screening mechanism in order to be viable at intermediate energies. Screening mechanisms include chameleons, symmetrons, dilatons, MOND-like dynamics, the Vainshtein mechan- ism, etcetera, depending on whether the screening is set by the local value of the field or by its derivatives [18]. Section 2 reviews various theories that have been explored in some detail as phenom- enological alternatives to GR in the strong-field regime. The section begins with a discussion of Lovelockʼs theorem, a ‘uniqueness theorem’ for the field equations of GR. Uniqueness is based on a small set of definite assumptions. The interest of Lovelockʼs theorem from a pragmatic point of view is that it can be ‘turned around,’ and used to classify extensions of GR based on which of the underlying assumptions of Lovelockʼs theorem they violate. Within this classification framework, we list and discuss several theories that have been seriously considered as plausible alternatives to GR in the context of strong-field tests. This selection is necessarily incomplete, and the authors of this review have different opinions on the intrinsic merits, viability and aesthetic appeal of these theories. The main criterion we used to choose these particular theories is that they are simple enough to make definite (and sometimes ‘orthogonal’) predictions for the strong-field dynamics of compact objects. The theories we discuss include: (1) scalar–tensor theories and their generalizations (including tensor-multiscalar and Horndeski theories); (2) f(R) theories; (3) theories whose action contains terms quadratic in the curvature, including in particular Einstein-dilaton-Gauss–Bonnet (EdGB) and dynamical Chern–Simons (dCS) theories; (4) Lorentz-violating theories, including Einstein-Æther, Hořava and n-Dirac–Born–Infeld (n-DBI) gravity; (5) massive gravity theories; (6) theories involving nondynamical fields, including the Palatini formulation of f(R) gravity and Eddington-inspired Born–Infeld (EiBI) gravity. This broad classification will be a leitmotif of the review. Table 1 lists some key references to the literature on the various theories listed above, plus others that are not considered in depth here. The table is an incomplete (but hopefully useful) ‘birdʼs eye’ reference guide for further study. Similar tables following the same classification scheme will support our discussion of the structure and stability of compact objects. Since we do not have a full theory of quantum gravity, an effective field theory (EFT) approach is often invoked when constructing phenomenological alternatives to GR [19, 20]. For example, not all theories of gravity with action quadratic in the curvature (item 3 in the Class. Quantum Grav. 32 (2015) 243001 Topical Review 7 Table 1. Catalog of several theories of gravity and their relation with the assumptions of Lovelock’s theorem. Each theory violates at least one assumption (see also figure 1), and can be seen as a proxy for testing a specific principle underlying GR. Theory Field content Strong EP Massless graviton Lorentz symmetry Linear Tmn Weak EP Well- posed? Weak-field constraints Extra scalar field Scalar–tensor S ⨯ ✓ ✓ ✓ ✓ ✓[34] [35–37] Multiscalar S ⨯ ✓ ✓ ✓ ✓ ✓[38] [39] Metric f(R) S ⨯ ✓ ✓ ✓ ✓ ✓[40, 41] [42] Quadratic gravity Gauss–Bonnet S ⨯ ✓ ✓ ✓ ✓ ✓? [43] Chern–Simons P ⨯ ✓ ✓ ✓ ✓ ⨯✓? [44] [45] Generic S/P ⨯ ✓ ✓ ✓ ✓ ? Horndeski S ⨯ ✓ ✓ ✓ ✓ ✓? Lorentz-violating Æ-gravity SV ⨯ ✓ ⨯ ✓ ✓ ✓? [46–49] Khronometric/ Hořava–Lifshitz S ⨯ ✓ ⨯ ✓ ✓ ✓? [48–51] n-DBI S ⨯ ✓ ⨯ ✓ ✓ ? none ([52]) Massive gravity dRGT/Bimetric SVT ⨯ ⨯ ✓ ✓ ✓ ? [17] Galileon S ⨯ ✓ ✓ ✓ ✓ ✓? [17, 53] Nondynamical fields Palatini f(R) — ✓ ✓ ✓ ⨯ ✓ ✓ none Eddington–Born–Infeld — ✓ ✓ ✓ ⨯ ✓ ? none Others, not covered here TeVeS SVT ⨯ ✓ ✓ ✓ ✓ ? [37] f R m( ) ? ⨯ ✓ ✓ ✓ ⨯ ? f(T) ? ⨯ ✓ ⨯ ✓ ✓ ? [54] Note. See text for details of the entries. Key to abbreviations: S: scalar; P: pseudoscalar; V: vector; T: tensor; ?: unknown; ?✓ : not explored in detail or not rigorously proven, but there exist arguments to expect ✓. The occurrence of ?⨯✓ means that there exist arguments in favor of well-posedness within the EFT formulation, and against well-posedness for the full theory. Weak-field constraints (as opposed to strong-field constraints, which are the main topic of this review) refer to Solar System and binary pulsar tests. Entries below “Others, not covered here” are not covered in this review. C lass. Q uantum G rav. 32 (2015) 243001 TopicalR eview 8 Table 2. Catalog of BH properties in several theories of gravity. The column ‘solutions’ refers to asymptotically-flat, regular solutions. Legend: ST —‘scalar–tensor’; º GR—‘same solutions as in GR’; É GR—‘GR solutions are also solutions of the theory’; NR—‘non rotating’; SR—‘slowly rotating’; FR—‘fast rotating/generic rotation’; ?—unknown or uncertain. Theory Solutions Stability Geodesics Quadrupole Extra scalar field Scalar–tensor GRº [55–60] [61–67] — — Multiscalar/Complex scalar GRÉ [56, 68, 69] ? ? [68, 69] Metric f(R) GRÉ [58, 59] [70, 71] ? ? Quadratic gravity Gauss–Bonnet NR [72–74]; SR [75, 76]; FR [77] [78, 79] SR [75, 80, 81]; FR [77] [76, 82] Chern–Simons SR [83–85]; FR [86] NR [87–90]; SR [79] [74, 91] [85] Generic SR [80] ? [80] equation (3.12) Horndeski [92–94] ? [95, 96] ? ? Lorentz-violating Æ-gravity NR [97–99] ? [98, 99] ? Khronometric/ Hořava–Lifshitz NR, SR [98–101] ? [102] [98, 99] ? n-DBI NR[103, 104] ? ? ? Massive gravity dRGT/Bimetric É GR, NR [105–108] [109–112] ? ? Galileon [113] ? ? ? Nondynamical fields Palatini f(R) ≡GR — — — Eddington–Born–Infeld ≡GR — — — C lass. Q uantum G rav. 32 (2015) 243001 TopicalR eview 9 Table 3. Catalog of NS properties in several theories of gravity. Symbols and abbreviations are the same as in table 2. Theory Structure Collapse Sensitivities Stability Geodesics NR SR FR Extra scalar field Scalar–tensor [26, 114–118] [116, 119, 120] [121–123] [124–131] [132] [133–143] [122, 144] Multiscalar ? ? ? ? ? ? ? Metric f(R) [145–157] [158] [159] [160, 161] ? [162, 163] ? Quadratic gravity Gauss–Bonnet [164] [164] [82] ? ? ? ? Chern–Simons ≡GR [27, 45, 165–167] ? ? [166] ? ? Horndeski ? ? ? ? ? ? ? Lorentz-violating Æ-gravity [168, 169] ? ? [170] [48, 49] [162] ? Khronometric/ Hořava–Lifshitz [171] ? ? ? [48, 49] ? ? n-DBI ? ? ? ? ? ? ? Massive gravity dRGT/Bimetric [172, 173] ? ? ? ? ? ? Galileon [174] [174] ? [175, 176] ? ? ? Nondynamical fields Palatini f(R) [177–181] ? ? ? — ? ? Eddington–Born–Infeld [182–188] [182, 183] ? [183] — [189, 190] ? C lass. Q uantum G rav. 32 (2015) 243001 TopicalR eview 10 list) are acceptable: the equations are of second order in the strong-coupling limit (a very desirable feature, given that higher-order derivatives are vulnerable to the so-called Ostro- gradskii instability [21]) only if the quadratic invariants appear in the special ‘Gauss–Bonnet’ combination. To avoid higher-order derivatives in the equations of motion one must generally assume that couplings are small, and work in an EFT framework. A more detailed discussion of EFTs and further references can be found in section 2.8. 1.2. Compact objects in modified theories of gravity Investigations of compact objects, binary pulsars, cosmology and gravitational radiation vary in depth and scope for the various classes of theories listed above. The best studied examples include scalar–tensor theories and some forms of quadratic gravity. Sections 3–5 are devoted to a discussion of isolated BHs, isolated NSs and compact binary systems in various theories. Isolated BHs. In section 3 we discuss BHs, one of the most striking predictions of GR. There is a consensus in the astronomy community that the massive compact objects in galactic centers, as well as the compact objects with mass larger than about 3Me found in some low-mass x-ray binaries, are well described by the Kerr solution in GR. However, this ‘BH paradigm’ rests on somewhat shaky foundations. From a theoristʼs point of view, one of the most convincing arguments in favor of the BH paradigm is that the alternatives are either unstable (as in the case of dense star clusters, fermion stars or naked singularities), unnatural (e.g. ‘exotic’ matter violating some of the energy conditions), contrived (such as gravastars), more implausible than BHs as the end- point of gravitational collapse (boson stars) or nearly indistinguishable from Kerr. The experimental evidence that astronomical BH candidates possess event horizons (more correctly, apparent horizons) rather than solid surfaces usually rests on plausibility arguments based on accretion physics [22, 23]. All of these arguments are model-dependent, and they leave room for some skepticism (see e.g. [24]). Strictly speaking, any tests that probe the Kerr metric alone (such as tests based on matter accretion or ray tracing of photon trajectories) are of little value as internal tests of GR. The reason is that a large number of extensions of GR admit the Kerr metric as a solution, and the theories that do not (e.g. EdGB, dCS and some Lorentz-violating gravity theories) predict BH solutions that differ from GR by amounts that may not be astrophysically measurable. Despite this somewhat pessimistic caveat, many ‘quasi-Kerr metrics’ have been proposed to perform GR tests, and we will review these proposals in section 3. Most deformations of the Kerr metric should be viewed as unnatural strawmen: they often have serious pathologies, and they are therefore unacceptable even for the limited scope of parametrizing deviations from the Kerr metric [25]. The prospects for testing GR with BHs look brighter when we recall that all extensions of GR predict different dynamics and different GW signatures when compact objects are perturbed away from equilibrium and/or when they merge. These arguments suggest that the most promising way to verify whether the compact objects in galactic centers or low-mass x-ray binaries are actually Kerr BHs is via direct observation of gravitational radiation, especially in the strong-field merger/ringdown phase. Last but not least, astrophysical BHs can be used to constrain modifications of GR in a different way. Many proposed modifications of Einsteinʼs theory and extension of the standard model of particle physics predict the existence of light bosonic degrees of freedom. Light bosons can trigger a superradiant instability, that extracts angular momentum from rotating BHs. By setting the superradiant instability timescale equal to the typical timescale Class. Quantum Grav. 32 (2015) 243001 Topical Review 11 for accretion to spin up the hole (say, the Salpeter time) one can get very stringent constraints on the allowed masses of light bosons (e.g. axions, Proca fields or massive gravitons). Table 2 is a quick reference guide to BH solutions and stability in various modified theories of gravity, organized in the same way as table 1. Isolated NSs. In section 4 we discuss NS solutions and their stability in various extensions of GR. Among other topics, we review the possibility that NSs in scalar–tensor theory may significantly deviate from their GR counterparts in the presence of ‘spontaneous scalarization’ (a phase transition akin to spontaneous magnetization [26]), we discuss controversial claims on the existence of NSs in f(R) theories, and we review the somewhat surprising ‘no-hair’ properties of NSs in quadratic gravity. A major problem in carrying out strong-gravity tests with NSs is the degeneracy between our ignorance of the equation of state (EOS) of high-density matter and strong-gravity effects. A possibility to lift the degeneracy consists of using universal relations between the moment of inertia, Love number (a measure of tidal deformability) and quadrupole moment of a NS— the so-called ‘I–Love–Q’ relations [27]—as well as EOS-independent relations between the lowest three multipole moments and those of higher order [28, 29]. Section 4.8 overviews the promises and challenges of this approach. A property of isolated NSs that plays an important role in many extensions of GR is their ‘sensitivity.’ The sensitivity is a measure of how the gravitational mass of the NS (or any self- gravitating object) varies as it moves within the nonhomogeneous extra field(s) mediating the gravitational interactions—or in other words, a measure of the violation of the strong equivalence principle (SEP) in the theory in question. Section 4.9 is a review of sensitivity calculations, that play an important role in binary dynamics. In table 3 we give a quick reference guide to NS solutions and their stability in various modified theories of gravity. Compact binaries. In preparation for binary pulsar tests (covered in section 6) and GW tests (section 7), in section 5 we review calculations of compact binary dynamics in some extensions of GR. The equations of motion and GW fluxes have been derived using the post- Newtonian (PN) expansion—an expansion in powers of v/c, where v is the orbital velocity of the binary—in scalar–tensor theory, f(R) gravity, specific forms of quadratic gravity (including EdGB and dCS) and Lorentz-violating theories. In comparison, numerical work is much less developed: at the moment of writing this review, simulations of compact binary mergers were carried out only for some of the simplest scalar–tensor theories. 1.3. Present and future tests of strong gravity Sections 6 and 7 capitalize on the material covered in previous sections. Section 6 reviews present astrophysical tests of GR, more specifically those coming from binary pulsar and cosmological observations. Section 7 focuses on the potential payoff of future GW obser- vations, and on how astrophysical modeling will affect our ability to perform tests of strong- field gravity in this context. The first part of section 6 is an overview of the spectacular progress of GR tests from binary pulsars. These extraordinary natural laboratories can be utilized to probe with high precision various nonradiative strong-field effects, as well as radiative aspects of gravity [30]. For instance, pulsars are now able to test Einstein’s quadrupole formula for GW emission to an accuracy of less than 0.1%. They provide stringent bounds on dipolar radiation and on violations of the SEP by strongly self-gravitating bodies (the best tests coming from pulsar- Class. Quantum Grav. 32 (2015) 243001 Topical Review 12 white dwarf systems), and they tightly constrain hypothetical violations of local Lorentz invariance of gravity. The near future in this field is particularly bright. Facilities such as the five-hundred meter aperture spherical radio telescope (FAST) and the Square Kilometer Array (SKA) are expected to come online soon. They should provide drastic improvements in the precision of current tests, qualitatively new tests with already known systems, and the dis- covery of many new ‘pulsar laboratories’ (possibly including the first pulsar-BH system). The second part of section 6 reviews cosmological tests of GR. In the last few decades, a remarkable wealth of astronomical data has constrained the expansion rate of the Universe and provided accurate maps of large-scale structure and the cosmic microwave background, placing ever-tightening constraints on cosmological parameters. In particular, anisotropies in the cosmic microwave background encode information on the geometry of the Universe, its material constituents and the initial conditions for structure formation. If GR is assumed to be correct, 96% of the material content of the Universe must consist of dark matter and dark energy. Since the evidence for these dark constituents of the Universe is purely gravitational, there have been countless attempts at finding theories in which dark matter and dark energy arise from modifications of gravity. These modifications affect the expansion rate of the Universe, but they should also affect gravitational clustering in a way that might be distin- guishable from GR. The proliferation of alternative theories of gravity has led to the devel- opment of model-independent cosmological tests of modified gravity somewhat similar to the PPN framework, which are now one of the primary drivers for future surveys of large scale structure. In the linear regime, these model-independent tests can be grouped in three classes, corresponding to three manifestations of a gravity theory: the action, the field equations derived from that action, and the combinations of those field equations which influence observable quantities. Sections 6 reviews these tests as well as recent progress in the nonlinear regime, where screening effects are important and numerical simulations are necessary. Last but not least, in section 7 we turn our attention to the future of strong-gravity tests, focusing on the promise of GW observations by Earth- and space-based detectors. The main target for both classes of detectors is the inspiral and merger of compact binaries. A technique called matched filtering, based on a careful monitoring of the GW phase to extract the (generally weak) signal from the detectorʼs noise, is used to observe these systems and to measure their parameters. GR makes very specific and testable predictions on the GW phasing of compact binaries as they inspiral, and on the oscillation frequencies of the compact objects that they produce as a result of the merger. If observed, any deviations from these predictions may identify problems in Einsteinʼs theory, and even point us to specific ways in which it could be modified. There are several comprehensive reviews on GW-based tests of GR. In particular, the recent Living Reviews in Relativity article by Yunes and Siemens [31] provides an excellent intro- duction to the literature on GR tests with Earth-based detectors (such as Advanced LIGO/Virgo, LIGO-India and KAGRA) and Pulsar Timing Arrays, and the review by Gair et al [32] expounds the great potential of future space-based detectors such as eLISA. We find it unnecessary to reproduce that material here, and therefore we focus on aspects that are not covered in detail in those reviews, namely: (1) the data analysis implementation of GR tests in advanced Earth-based detectors (the TIGER framework), arguably our best hope to constrain modified gravity using GW observations in the near future; and (2) an analysis of how astrophysical effects can limit (or sometimes enhance) our ability to test strong-field gravity with GW observations. As a rule, in this paper we use geometrical units where the gravitational constant and the speed of light are set to unity: GN = c = 1. Factors of GN and c are occasionally reinstated for clarity, and in isolated cases (e.g. in section 2.8) we switch to units such that c 1 = = . We adopt the mostly positive signature for the metric, and the same conventions as in Misner et al [33] for the Riemann tensor. Class. Quantum Grav. 32 (2015) 243001 Topical Review 13 2. Extensions of GR: motivation and overview 2.1. A compass to navigate the modified-gravity atlas There are countless inequivalent ways to modify GR, many of them leading to theories that can be designed to agree with current observations. Cosmological observations and funda- mental physics considerations suggest that GR must be modified at very low and/or very high energies. Experimental searches for beyond-GR physics are a particularly active and well motivated area of research, so it is natural to look for a guiding principle: if we were to find experimental hints of modifications of GR, which of the assumptions underlying Einsteinʼs theory should be abandoned? Such a guiding principle can be found by examining the building blocks of Einsteinʼs theory. Lovelockʼs theorem [191, 192] (the generalization of a theorem due to Cartan [193]) is particularly useful in this context. In simple terms, the theorem states that GR emerges as the unique theory of gravity under specific assumptions. More precisely, it can be articulated as follows: In four spacetime dimensions the only divergence-free symmetric rank-2 tensor constructed solely from the metric gμν and its derivatives up to second differential order, and preserving diffeomorphism invariance, is the Einstein tensor plus a cosmological term. Figure 1. This diagram illustrates how Lovelock’s theorem serves as a guide to classify modified theories of gravity. Each of the yellow boxes connected to the circle represents a class of modified theories of gravity that arises from violating one of the assumptions underlying the theorem. A theory can, in general, belong to multiple classes. See table 1 for a more precise classification. Class. Quantum Grav. 32 (2015) 243001 Topical Review 14 Lovelockʼs theorem suggests a natural route to Einsteinʼs equations G g T8 , 2.1( )p+ L =mn mn mn where G R Rg1 2 º -mn mn mn is the Einstein tensor and Tμν is the matter stress–energy tensor. Indeed, the divergence-free nature of the Einstein tensor (that follows from the Bianchi identities) implies that Tμν is also divergence free, ∇μT μν = 0. This property is necessary for geodesic motion and it guarantees the validity of the weak equivalence principle, i.e. the universality of free fall (see [194, 195] for further discussion). If we assume that the equations of motion for the gravitational field and the matter fields follow from a Lagrangian, the arguments above single out the Einstein–Hilbert action S x g R S g 1 16 d , , 2.2M 4 ( )⎡⎣ ⎤⎦òp = - + Y mn where Ψ collectively denotes the matter fields, which couple minimally to gμν, so that SM reduces to the standard model action in a freely falling frame. As it stands, Lovelockʼs theorem seems to leave little room for modifying the gravita- tional theory (2.2). However, when analyzed in detail, the theorem contains a number of nontrivial assumptions [196]. Giving up each of these assumptions provides a way to cir- cumvent the theorem and gives rise to different classes of modified theories of gravity, as illustrated in figure 1. Specifically, there are at least four inequivalent ways to circumvent Lovelockʼs theorem: (1) Additional fields. Dynamical fields. The simplest and most beaten path to circumvent Lovelockʼs theorem consists of adding extra degrees of freedom. This leaves more options to construct the left-hand side of Einsteinʼs equations (2.1), including more than just the metric and connection. Lifting this assumption paves the way for countless possibilities, where the metric tensor gμν is coupled to extra fundamental (scalar, vector, tensor) fields. Similar corrections arise from lifting the assumption of second differential order35. Because of the coupling with extra dynamical fields, these theories usually violate the SEP [2]. It is not straightforward to construct theories with extra fields nonminimally coupled to gravity that avoid instabilities associated to the new degrees of freedom, as generically predicted by Ostrogradskiʼs theorem [21]. Because such degrees of freedom remain undetected to date, a major challenge for these theories has been to tame the behavior of the extra fields, so as to evade current experimental constraints related to their existence [2]. Nondynamical fields. Lovelockʼs theorem implicitly assumes that the matter stress– energy tensor Tμν enters the field equations (2.1) linearly. By dropping this assumption, it is possible to construct theories where the left-hand side of equation (2.1) is precisely the Einstein tensor, whereas the right-hand side is a nonlinear combination of Tμν such that its covariant divergence vanishes, i.e., that ∇μT μν = 0 remains an identity [197]. These theories satisfy the weak equivalence principle and are equivalent to GR in vacuum, but differ from it in the coupling to matter. Due to such nonlinear couplings, they resolve some of the curvature singularities that afflict fluid collapse and early time cosmology in GR [198]. The only theories belonging to this class known to date are special classes of theories which modify GR by adding only auxiliary (i.e. nondynamical) fields, the prototypical example being the Palatini formulation of f ( ) gravity [11, 199]. Here 35 Indeed, higher-order equations can always be brought to second-order form by adding an arbitrary number of (effective) extra fields. A representative example is metric f(R) gravity [11], see section 2.3. Class. Quantum Grav. 32 (2015) 243001 Topical Review 15 g = mn mn , where mn denotes the Ricci tensor built from the connection, to distinguish it from the Ricci tensor Rμν in the metric formalism: see the discussion below equation (2.50). (2) Violations of diffeomorphism invariance. Lorentz invariance. One particular form of diffeomorphism invariance, namely Lorentz invariance, has been tested with remarkable precision in the standard model sector, and it is widely believed to be a necessary ingredient of viable gravitational theories. However, if we assume that Lorentz invariance is just an emergent symmetry that is broken at high energies in the gravitational sector, a new class of gravity theories can be built. Some of these theories were found to possess a better ultraviolet behavior than GR [200]. Violations of Lorentz invariance are typically encoded in some extra field(s), so that theories of this class usually also belong to category (1) above. Massive gravity. The assumption of diffeomorphism invariance is also crucial because it implies that gravity should be mediated by a massless spin-2 field. Understanding how the graviton can acquire mass is a century-old problem, and strong constraints on the graviton mass are in place [201]. Massive gravity theories are currently under intense scrutiny, mostly because of their applications in the context of the cosmological constant problem (see [17] for a review). (3) Higher dimensions. Even retaining all other assumptions of Lovelockʼs theorem, the Einstein–Hilbert action (2.2) is not unique in higher dimensions. Gravitational theories built in dimensions other than four have a strong theoretical interest for several reasons, including the formulation of consistent string theories or understanding how the field equations depend on an extra parameter, i.e. the spacetime dimension36. These theories may even offer a resolution of the hierarchy problem, because they predict that the fundamental Planck mass can be several orders of magnitude smaller than the effective four-dimensional M 10 GeVPl 19» . Some extra-dimensional models are severely constrained from an experimental point of view (see e.g. [205]), and their relevance for beyond-GR effects in astrophysics is limited. However, some quantum-gravity corrections might be accessible through astrophysical observations, as we discuss in section 2.8. GR in higher dimensions leads naturally to additional fields if the theory is reduced to D = 4 dimensions: additional scalar and gauge fields emerge by performing a Kaluza–Klein or dimensional reduction from D > 4 to D = 4 dimensions. We discuss theories in higher dimensions only marginally and refer the interested reader to other reviews, e.g. [15]. (4) WEP violations. The requirement that the left-hand side of Einsteinʼs equation be divergence-free is dictated by the desire of having a divergence-free Tμν and, in turn, by the weak equivalence principle. Various classes of theories that circumvent Lovelockʼs theorem only by postulating a nonminimal coupling to the matter sector (and thus violating the weak equivalence principle) have been proposed (see e.g. [206], and [14] for a review). Nevertheless, because the equivalence principle has been tested with the astonishing precision of one part in 1013 [207], we will seldom discuss theories where it is violated. 36 The Cauchy problem in D-dimensional Gauss–Bonnet gravity was first investigated by Choquet-Bruhat in [202] (see also [38]). Reall et al showed that Lovelock theories in D > 4 spacetime dimensions allow for acausal propagation of physical degrees of freedom in some backgrounds, including BH spacetimes [203]. They concluded that higher-dimensional Lovelock theories may or may not be hyperbolic depending on the background spacetime. Willison [204] showed that Lovelock gravity is locally well-posed in arbitrary backgrounds, but global hyperbolicity is still an open problem. Class. Quantum Grav. 32 (2015) 243001 Topical Review 16 Although rather elementary, the classification proposed above has the virtue of sim- plicity. In this review we are mainly interested in understanding how to test GR, and especially what we can test, rather than attempting a comprehensive classification of alternative theories. Our point of view is therefore very practical: any modified theory of gravity will necessarily fall into one of the categories above, and therefore it will violate one or more of the fundamental principles underlying GR; these violations will determine the new effects predicted by the theory, and the payoff of a hypothetical observation of these effects. In table 1 we give a schematic (and necessarily incomplete) summary of proposed extensions of GR that are of interest for astrophysics, i.e. those that provide potential means to test the fundamental principles of GR with current and near-future astrophysical observations. In the rest of this review we discuss these theories (and their implications for experimental verifications of GR in the weak-field and strong-field regimes) in more detail. Regardless of the manner in which Lovelockʼs theorem is violated, all these theories face a common challenge: how to modify the behavior of gravity at extreme energy scales, while leaving the (tightly constrained) intermediate energy regime unchanged? The hypothetical solution to this problem is termed ‘screening.’ When used in a general sense, the word is simply a label for unknown physics, much like the phrase ‘dark energy.’ Three concrete kinds of screening mechanisms are known, corresponding to density-dependent modifications of the three kinds of terms appearing in the action of a scalar field: (i) kinetic terms (including derivative self-interactions), (ii) potential terms, and (iii) couplings to matter fields [18]. These modifications lead to, respectively, the symmetron/dilaton screening mechanisms, the cha- meleon mechanism and the Vainshtein mechanism; see section 6.2 for further mathematical details. It is likely that these exhaust the possibilities for a gravity theory with one additional scalar field. However, it is far from clear that one of these three mechanisms can be embedded in every gravity theory in the current literature. One could argue that any theory lacking an ‘in-built’ screening mechanism is disfavored or, at best, incomplete. However, given the rapidly evolving nature of this research area, it would seem hasty to discard all nonscreening theories at this stage. If GR is not the fully correct theory of gravity, then we are forced to accept one of the following propositions: (a) the true theory must incorporate one of the three known mechanisms; or (b) there exist yet-unknown screening mechanisms, which require more than a single scalar field to operate; or (c) deviations from GR do exist in the intermediate energy regime, but are below the current detection threshold of PPN and binary pulsar constraints. It is worth noting that, in addition to the strong-field and cosmological tests of gravity described in this review, screening mechanisms have spawned a wave of new laboratory and astrophysical tests of gravity. Laboratory examples include experiments to detect the cha- meleon mechanism using cold atom interferometry [208, 209], and the ‘afterglow’ of a chameleon field interacting with the electromagnetic field inside a radio frequency cavity [210]. New astrophysical tests include searches for a potential mismatch between distance indicators such as cepheid variables and tip-of-the-red-giant-branch stars in unscreened dwarf galaxies [211]. Class. Quantum Grav. 32 (2015) 243001 Topical Review 17 2.2. Scalar–tensor gravity One of the most natural extensions of GR is scalar–tensor gravity, in which one or more scalar degrees of freedom are included in the gravitational sector of the theory, through a nonminimal coupling (i.e., the Ricci scalar in the Einstein–Hilbert action is multiplied by a function of the scalar field(s)). Several reviews provide extensive discussions on the subject, see e.g. [39, 196, 212–214]. Scalar fields with nonminimal couplings to gravity appear in several contexts, such as in string theory [215], in Kaluza–Klein-like theories [216] or in braneworld scenarios [217, 218]. They also have important applications in cosmology [15]. Therefore, scalar– tensor gravity is a good framework to study phenomenological aspects of several possible fundamental theories. 2.2.1. The Bergmann–Wagoner formulation. The most general action of scalar–tensor gravity with one scalar field which is at most quadratic in derivatives of the fields was studied by Bergmann and Wagoner [219, 220], and can be written (after an appropriate field redefinition) as: S x g R g U S g 1 16 d , , 2.34 M( )( )( ) ( ) ( ) ⎡ ⎣⎢ ⎤ ⎦⎥ ⎡⎣ ⎤⎦òp f w f f f f f= - - ¶ ¶ - + Ymn m n mn where ω and U are arbitrary functions of the scalar field f, and SM is the action of the matter fields Ψ. When ω(f) = ωBD is constant and U(f) = 0, the theory reduces to (Jordan–Fierz-) Brans–Dicke gravity [221–223], an extension of GR which was proposed in the mid-20th century (see [224–226] for a historical account). The Bergmann–Wagoner theory (2.3) can be expressed in a different form through a scalar field redefinition j = j(f) and a conformal transformation of the metric g g A g2 ( ) j =mn mn mn - . In particular, fixing A(j) = f−1/2, the action (2.3)—generally referred to as the Jordan-frame action—transforms into the Einstein-frame action S x g R g V S A g 1 16 d 2 , , 2.44 M 2( )( ) ( ) ( ) ( )⎡⎣ ⎤⎦ ⎡⎣ ⎤⎦   òp j j j j= - - ¶ ¶ - + Ymn m n mn where gåand Rå are the determinant and Ricci scalar of gmn , respectively, and the potential V(j) ≡ A4(j)U(f(j)). The price paid for the minimal coupling of the scalar field in the gravitational sector is the nonminimal coupling in the matter sector of the action: particle masses and fundamental constants depend on the scalar field. We remark that the actions (2.3) and (2.4) are just different representations of the same theory: the outcome of an experiment will not depend on the chosen representation, as long as one takes into account that the units of physical quantities do scale with powers of the conformal factor A [194, 227]. It is then legitimate, when modeling a physical process, to choose the conformal frame in which calculations are simpler: for instance, in vacuum the Einstein-frame action (2.4) formally reduces to the GR action minimally coupled with a scalar field. It may then be necessary to change the conformal frame when extracting physically meaningful statements (since the scalar field is minimally coupled to matter in the Jordan frame, test particles follow geodesics of the Jordan-frame metric, not of the Einstein-frame metric). The relation between Jordan-frame and Einstein-frame quantites is simply f = A−2(j), 3 2 2( ) ( )w f a j+ = - , where Ad ln d( ) ( ( ))a j j jº [2]. Note that the theory is fixed once the function ω(f)—or, equivalently, α(j)—is fixed, and the form of the scalar potential is chosen. Moreover, many phenomenological studies neglect the scalar potential. This Class. Quantum Grav. 32 (2015) 243001 Topical Review 18 approximation corresponds to neglecting the cosmological term, the mass of the scalar field and any possible scalar self-interaction. In an asymptotically flat spacetime the scalar field tends to a constant f0 at spatial infinity, corresponding to a minimum of the potential. Taylor expanding U(f) around f0 yields a cosmological constant and a mass term for the scalar field to the lowest orders [36, 220]. Scalar–tensor theory with a vanishing scalar potential is characterized by a single function α(j). The expansion of this function around the asymptotic value j0 can be written in the form . 2.50 0 0( ) ( ) ( )a j a b j j= + - +  As mentioned above, the choice α(j) = α0 = constant (i.e., ω(f) = constant) corresponds to Brans–Dicke theory. A more general formulation, proposed by Damour and Esposito-Farèse, is parametrized by α0 and β0 [26, 116]. Another simple variant is massive Brans–Dicke theory, in which α(j) is constant, but the potential is nonvanishing and has the form U U 1 2 0 0 2( ) ( )( )f f f f= ¢¢ - , so that the scalar field has a mass m Us 2 0( )f~ ¢¢ . Note that since the scalar field j in the action (2.4) is dimensionless, the function α(j) and the constants α0, β0 are dimensionless as well. The field equations of scalar–tensor theory in the Jordan frame are (see e.g. [228, 229]) G T g g U g a 8 1 2 1 2 , 2.6 2 g( ) ( ) ( ) ( ) ⎜ ⎟⎛ ⎝ ⎞ ⎠  p f w f f f f f f f f f f f = + ¶ ¶ - ¶ ¶ +   - - mn mn m n mn l l m n mn mn T T U U b 1 3 2 8 16 d d d d 2 , 2.6g ( ) ( ) ( ) ⎛ ⎝⎜ ⎞ ⎠⎟ f w f p pf f w f f f f f f= + - ¶ ¶ - ¶ ¶ + -l l where T g S g g2 ,1 2 M( ) ( )d d= - - Ymn mn mn - is the Jordan-frame stress–energy tensor of matter fields, and T = gμνTμν. In the Einstein frame, the field equations are G g g V T a2 1 2 1 2 8 , 2.7( ) ( )⎜ ⎟⎛ ⎝ ⎞ ⎠    j j j j j p= ¶ ¶ - ¶ ¶ - +mn m n mn s s mn mn T V b4 1 4 d d , 2.7g ( ) ( ) j pa j j = - + where T g S A g g2 ,1 2 M 2( ) ( )  d d= - - Ymn mn mn - is the Einstein-frame stress–energy tensor of matter fields and T g T  = mn mn (see e.g. [39]). Equation (2.7b) shows that α(j) couples the scalar fields to matter [230], as does 3 2 1( ( ))w f+ - in the Jordan frame: see equation (2.6b). Astrophysical observations set bounds on the parameter space of scalar–tensor theories. In the case of Brans–Dicke theory, the best observational bound ( 3.5 100 3a < ´ - ) comes from the Cassini measurement of the Shapiro time delay. In the more general case with 00b ¹ , current constraints on (α0, β0) have been obtained by observations of NS–NS and NS–WD binary systems [37], and will be discussed in section 6 (see figure 37). Observations of compact binary systems also constrain massive Brans–Dicke theory, leading to exclusion regions in the (α0, ms) plane [36]. An interesting feature of scalar–tensor gravity is the prediction of certain characteristic physical phenomena which do not occur at all in GR. Even though we know from observations that 10a  and that GR deviations are generally small, these phenomena may Class. Quantum Grav. 32 (2015) 243001 Topical Review 19 lead to observable consequences. There are at least three possible smoking guns of scalar– tensor gravity. The first is the emission of dipolar gravitational radiation from compact binary systems [228, 231], which will be discussed in section 5.1. Dipolar gravitational radiation is ‘pre-Newtonian,’ i.e. it occurs at lower PN order than quadrupole radiation, and it does not exist in GR. The second is the existence of nonperturbative NS solutions in which the scalar field amplitude is finite even for 10a  . This spontaneous scalarization phenomenon [26, 116] will be discussed in detail in section 4.2. Here we only remark that spontaneous scalarization would significantly affect the mass and radius of a NS, and therefore the orbital motion of a compact binary system, even far from coalescence. The third example is also nonperturbative, and it involves massive fields. The coupling of massive scalar fields to matter in orbit around rotating BHs leads to a surprising effect: because of superradiance, matter can hover into ‘floating orbits’ for which the net gravitational energy loss at infinity is entirely provided by the BHʼs rotational energy [232]. The phenomenology of scalar–tensor theory in vacuum spacetimes, such as BH spacetimes, is less interesting. When the matter action SM can be neglected, the Einstein- frame formulation of the theory is equivalent to GR minimally coupled to a scalar field. BHs in Bergmann–Wagoner theories satisfy the same no-hair theorem as in GR, and thus the stationary BH solutions in the two theories coincide [56, 59]. Moreover, dynamical (vacuum) BH spacetimes satisfy a similar generalized no-hair theorem: the dynamics of a BH binary system in Bergmann–Wagoner theory with vanishing potential are the same as in GR [39], up to at least 2.5PN order for generic mass ratios [233] and at any PN order in the extreme mass- ratio limit [234] (see section 5.1.1). These no-hair theorems will be discussed in section 3.2. 2.2.2. Scalar–tensor theories with multiple scalar fields. When gravity is coupled with more than one scalar field, the action (2.3) has the more general form [39] S x g F R g V S g 1 16 d , , 2.8ab a b4 M( )( ) ( ) ( ) ( )⎡⎣ ⎤⎦òp f g f f f f= - - ¶ ¶ - + Ymn m n mn where F, V are functions of the N scalar fields f a (a = 1KN). The scalar fields live on a manifold (the target space) with metric γab(f). The action (2.8) is invariant not only under space–time diffeomorphisms, but also under target-space diffeomorphisms, i.e. scalar field redefinitions. These theories have a richer structure than those with a single scalar field, since the geometry of the target space can affect the dynamics. For instance, the theories with a complex scalar field discussed in section 3.2.2, in which the no-hair theorems can be circumvented, can also be seen as multiscalar–tensor theories with N = 2. 2.2.3. Horndeski gravity. The most general scalar–tensor theory with second-order field equations (and one scalar field) is Horndeski gravity [235]. The action of Horndeski gravity can be written in terms of Galileon interactions (see [236] and section 2.6) as S x g K X G X G X R G X G X G G X d , , , , , , 6 3 2 , 2.9 X X 4 3 4 4, 2 5 5, 3 } ( ) ( ) ( ) { ( ) ( ) ( )( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ⎡⎣ ⎤⎦ ⎡⎣ ⎤⎦     ò f f f f f f f f f f f f f f f f f f = - - + + -     +   - -     +       m n m n mn m n m n m n m n m s n s where K and the Giʼs (i = 3...5) are functions of the scalar field f and of its kinetic term X 1 2 f f= - ¶ ¶m m , and Gi,X are derivatives of Gi with respect to the kinetic term X. For a Class. Quantum Grav. 32 (2015) 243001 Topical Review 20 particular choice of these functions, this theory coincides with Gauss–Bonnet gravity (see section 2.4). As we shall discuss in section 3.2, in Horndeski theory the no-hair theorem can be circumvented, and thus stationary BH solutions can be different from GR. 2.3. Metric f(R) theories The standard paradigm to explain the acceleration of the cosmic expansion is to postulate the existence of a diffuse form of dark energy described by an exotic EOS (P ≈ −ρ) and amounting to roughly 70% of the critical energy density. The cosmological constant is the most natural candidate for this dark ‘fluid,’ although its tiny value (as inferred by cosmo- logical observations) clashes with the value of vacuum energy as inferred from particle physics. As mentioned above, this is one of the main problems in theoretical physics: the cosmological constant problem [5, 237, 238]. As an alternative to the standard ΛCDM (Λ-cold dark matter) model, it has been proposed that infrared modifications of gravity could be the explanation for the cosmic acceleration. In this context, so-called f(R) modified gravities have a long history [239] and have been widely explored as prototypical infrared corrections to GR. The action for f(R) gravity reads S x g f R S g 1 16 d , , 2.104 M( ) ( )⎡⎣ ⎤⎦òp = - + Y mn where Ψ collectively denotes all matter fields and f(R) is a function of the scalar curvature R. It is customary to use a simplified notation where fR ≡ f ′(R), fRR ≡ f ″(R) and so on. We shall focus for the moment on the theory obtained from the action above through a metric variational principle. Palatini f(R) gravity is a completely different theory, that will be discussed in section 2.7 below. Primarily, f(R) theories attracted attention for their potential to describe the cosmological acceleration of the Universe without a fine-tuned cosmological constant [11]. Viable f(R) models are usually chosen by ensuring that the field equations admit a de Sitter solution with curvature radius RdS. We refer the reader to specialized reviews [11, 12, 14] for a more detailed discussion of the theoretical aspects and of current experimental constraints. Viable f(R) theories. If one wishes to modify GR at cosmological scales, while leaving the large curvature behavior essentially unaffected, very stringent constraints are in place. Solar System observations and local tests strongly constrain viable f(R) models and rule out many candidates (see [12] for a review). In general, f(R) models must be described by monotonically growing and convex functions, i.e. fR > 0 and fRR > 0, in order to avoid ghosts (i.e., negative kinetic energy states) and tachyons. Furthermore, f(R) gravity theories are introduced to modify the infrared behavior of GR when R  Rc, Rc being some cosmological curvature scale of the order of RdS. In order to recover Einsteinʼs theory at higher curvature and to pass Solar System tests, viable models usually have, at leading order f R f f R R, 1, 0, . 2.11RRR c ( )    In the following, we shall focus on classes of f(R) theories of gravity that satisfy the above requirements. Different formulations of f(R) theories. It is well known that f(R) theories are dynamically equivalent to a specific class of scalar–tensor theories [240–243] (see [11] for a review), so they propagate an additional scalar degree of freedom. These theories allow for different formulations depending on which quantity is identified as the scalar field. At least three different approaches to the study of f(R) theories have been proposed. While equivalent in principle, each approach has different features and practical drawbacks. A common choice in Class. Quantum Grav. 32 (2015) 243001 Topical Review 21 the literature is to transform the f(R) action (2.10) into a Brans–Dicke theory with ω = 0 in the Jordan frame. If f 0RR ¹ , the action (2.10) is dynamically equivalent to S x g R V S g 1 16 d , , 2.12J 4 KM M[ ]( ) ( )⎡⎣ ⎤⎦òp f f= - - + Y mn where f = fR and V Rf fKM R( )f = - (the reason for the KM subscript will be apparent shortly). It should be stressed that, if fRR = 0 at some point, the equivalence is not guaranteed and must be checked on a case-by-case basis. Scalar–tensor theories with a vanishing kinetic term of the form (2.12) were also studied by O’Hanlon [244] and others (see e.g. [245]). In the context of compact objects, Kobayashi and Maeda [145, 147] integrated the field equations arising from the action (2.12), which read R fg g T 1 2 8 , 2.13( )f f f p- -   + =mn mn m n mn mn T f R R T V8 3 1 3 2 8 3 d d , 2.14KM[ ( ( )) ( )] ( )f p f f f p f = + - º + where the evolution equation for the scalar degree of freedom f is obtained from the trace of equation (2.13) above, and R is now an implicit function of f. It is possible to recast f(R) theory as a scalar–tensor theory in the Einstein frame for a new scalar field logj fµ [240, 246]. By defining f g A g A f 3 2 log , , e , 2.15R 2 2 R 2 3 ( )j º = º =mn mn j- - the action (2.12) becomes S x g R V S A g 1 16 d 2 , , 2.16 E 4 BL M 2[ ]( ) ( ) ⎡⎣ ⎤⎦  ò p j j j= - - ¶ ¶ - + Ya a mn where the new scalar potential reads V Rf f f , 2.17BL R R 2 ( ) ( )j = - Here R = R(j), and we introduced a subscript ‘BL’ because this formulation was used by Babichev and Langlois in their study of compact stars [148, 149]. Besides the standard approaches discussed above, another formulation of the theory was proposed by Jaime et al [151]. In this approach, the Ricci curvature R is considered as an independent scalar degree of freedom. By introducing a new scalar field ψ, the action (2.10) is dynamically equivalent to S x g f R V S g 1 16 d , , 2.18R 4 JPS M[ ]( ) ( ) ( )⎡⎣ ⎤⎦òp y y= - ¢ - + Y mn where V f fJPS ( ) ( )y y yº ¢ - . Variation with respect to ψ leads to ψ = R, if f 0RR ¹ . In fact, this is usually considered as an intermediate step in reducing the action (2.10) to the scalar– tensor theory (2.12): see e.g. [11]. As in the case of Brans–Dicke theory in the Jordan frame, the field equations for the scalar field simply impose ψ = R, but the scalar evolution arises from the trace of Einsteinʼs equation. The field equations read Class. Quantum Grav. 32 (2015) 243001 Topical Review 22 G f f R f R R g Rf f T T R V R f T f f R V R 1 6 16 8 , d d 8 3 d d , 2.19 RR RRR RR RRR RR R R JPS eff 2 JPS ( ) ( )( ) ( ) ( ) ⎡ ⎣⎢ ⎤ ⎦⎥  p p p =   +   - + + + = º -  + mn m n m n mn mn with V R f Rf fd d 2 3 RRJPS R( )º - . As pointed out in [151], in this formulation the potential is as well defined as the function f(R). The field equations of f(R) gravity are of fourth differential order, but the theory admits a well-posed initial value problem by virtue of its equivalence with scalar–tensor grav- ity [40, 41]. 2.4. Quadratic gravity One of the most pressing problems in theoretical physics is to accommodate GR in the framework of quantum field theories. It has long been known that Einsteinʼs theory is not renormalizable in the standard quantum field theory sense, and this is a major obstacle on the route to quantum gravity. The situation changes if the Einstein–Hilbert action is assumed to be only the first term in an expansion containing all possible curvature invariants, as also suggested by low-energy effective string theories. Already in the 1970s, Stelle showed that including quadratic curvature terms in the action makes the theory renormalizable [3]. This comes at the cost of having higher-derivative terms in the field equations, which generically introduce ghosts or other pathologies (but see [247, 248] for recent progress in constructing a class of ghost-free, higher-derivative extensions of GR). At second order in the curvature, the only independent algebraic curvature invariants are R R R RR, , , , 2.202 2 2 ( )*mn mnrs where R R R2 ºmn mn mn , R R R2 ºmnrs mnrs mnrs, RR R R1 2 * º mnrs nmlk rs lk is the Pontryagin scalar, and  mnrs is the Levi-Civita tensor. Of particular interest are the Gauss–Bonnet scalar R R R R4GB 2 2 2 2º - +mn mnrs and the Pontryagin scalar (also referred to as the Chern–Simons scalar) defined above, because these terms can be shown to emerge in low-energy realizations of string theory [215, 249]. The Pontryagin scalar also appears in loop quantum gravity [250]. However, these terms alone do not yield modifications to Einsteinʼs equations in four spacetime dimensions, because their integrals are four-dimensional topological invariants and only account for boundary terms in the action. To circumvent this problem one is thus forced to add extra dynamical fields, i.e., extra propagating degrees of freedom (but see section 2.7 below for a different strategy using nondynamical fields). The simplest way to introduce nontrivial higher-order curvature corrections is via coupling with a scalar field. The most generic class of four-dimensional theories obtained by including all quadratic algebraic curvature invariants coupled to a single scalar field reads [74, 80] S g x R V f R f R R f R R f RR S g 1 16 d 2 , , 2.21 4 1 2 2 3 4 mat ( ) ( ) ( ) ( ) ( ) ( ) ( ) ⎡⎣ ⎤⎦ ⎡⎣ ⎤⎦ * òp f f f f f f f g f = - -   - + + + + + Y m m mn mn mnrs mnrs mn where V(f) is the scalar self-potential, fi(f) (i = 1,...,4) are coupling functions, and in the matter action Smat we have included a nonminimal but universal metric coupling, which thus satisfies the weak (but in general not the strong) equivalence principle. The action (2.21) generically yields higher-order field equations that are prone to the Ostrogradski instability Class. Quantum Grav. 32 (2015) 243001 Topical Review 23 and to the appearance of ghosts, unless the various terms appear in the special combination corresponding to the four-dimensional Gauss–Bonnet invariant (discussed in section 2.4.1 below). To avoid this instability, the theory (2.21) must be considered as an effective action, obtained as the truncation of a more general theory, valid only up to second order in curvature37. In the decoupling limit (where the effective theory is valid, see section 2.8), a perturbative approach is applicable and the field equations remain of second differential order for generic combinations of the curvature invariants. For example, it has been shown that dCS gravity (introduced in section 2.4.2 below) does not exhibit any ghost-like instabilities when treated order-by-order in the perturbation scheme and, in fact, can be cast into a well-posed Cauchy problem in the decoupling limit [44]. We expect a similar argument to hold for EdGB gravity (see section 2.4.1), but a rigorous proof in this case is still missing. The EFT approach is not only motivated by the desire to avoid higher-order derivatives in the field equations, but it arises naturally in some low-energy expansion in string theory, which indeed contains the Gauss–Bonnet and Chern–Simons terms coupled respectively to the dilaton and axion at second order in the curvature. In this approach the Einstein–Hilbert term is considered as the first-order term in a (possibly infinite) series expansion containing all possible curvature corrections. In this sense, GR may be only accurate up to second-order terms in the curvature. In the geometrical units adopted here, the scalar field entering the action (2.21) is dimensionless, whereas the coupling functions fi(f) have the dimensions of a length squared, i.e. of an inverse curvature. Thus, at variance with the scalar–tensor theories previously discussed, quadratic-gravity corrections may require introducing a new fundamental length scale. If the length scale is taken to be the Planck length, quadratic-gravity corrections would be negligible at the scales of compact objects: they would be suppressed by a factor of ℓ ℓ 10 , 2.22Planck 2 BH 2 78 ( )~ - where ℓ 10 mPlanck 35~ - is the Planck length, and ℓBH ∼ 10 km is the typical scale of a compact object. However, as discussed in the introduction, the region close to compact objects has been poorly probed, particularly when the spacetime is highly dynamical. Moreover, assuming that GR is correct all the way down to the Planck scale, with no new gravitational physics along the way, would be a tremendous extrapolation38. When considering quadratic gravity, the standard approach is to assume the existence of this new fundamental scale, unrelated to the Planck scale, and proceed with calculations of observables from compact objects. For experimental observations to differ from GR predictions the ‘new’ length scale must be comparable to astrophysical scales. Here we will adopt this agnostic phenomenological point of view. In this approach, quadratic-curvature terms may be important when dealing with non- linear, relativistic solutions. Clearly, within this perturbative context we can only consider corrections which are small compared to the leading Einstein–Hilbert term. In practice, the 37 Alternatively, one can circumvent the Ostrogradski instability by expanding the phase-space of the (dynamical) variables if the resulting equations of motion constitute a closed system of PDEs that are at most second order [251, 252]. 38 As an illustration, the gravitational potential at the Earthʼs surface, where Newtonian gravity proved to be extremely successful, is only four orders of magnitude smaller than the gravitational potential at the Sunʼs surface, where relativistic effects are relevant, as shown by the classical tests of GR. In a particle physics context, even a very successful theory such as quantum electrodynamics cannot be extrapolated from atomic to nuclear energy scales, where the strong interaction dominates over electromagnetism; and again, these two scales are separated by just six orders of magnitude. Class. Quantum Grav. 32 (2015) 243001 Topical Review 24 coupling functions fi are expanded as f , 2.23i i i 2( )( ) ( )f h a f f= + + where ηi and αi (i = 1, 2, 3, 4) are dimensionful coupling constants. When the coupling functions are constant, i.e. αi = 0, the theories above admit all vacuum GR solutions [74, 253]. However, even in this case the background solutions generically have a different linear response with respect to GR: for example, these theories predict a different GW emission [88, 254, 255]. We will mostly be interested in theories that modify the structure of BHs and NSs, and we will consider the generic case 0ia ¹ . At any rate, it is remarkable that in the weak-coupling limit (and provided that the fiʼs are analytic functions) all viable quadratic theories of gravity boil down to a small number of coupling constants that parametrize strong-curvature deviations from GR. 2.4.1. EdGB gravity. When f f f4 42 1 3= - = - and f4 = 0, the theory (2.21) reduces to EdGB gravity [73], with action S g x R V f R 1 16 d 2 , 2.244 1 GB 2( ) ( ) ( )⎡⎣ ⎤⎦òp f f f f= - -   - +m m where f1(f) is a generic coupling function and the Gauss–Bonnet invariant RGB 2 has been defined below equation (2.20). This is the only quadratic theory of gravity whose field equations are of second differential order for any coupling, and not just in the weak-coupling limit. Indeed, when f e1 GB 2( )f a= f- , the theory reduces to the bosonic sector of heterotic string theory [256]. Gauss–Bonnet gravity can also be seen as a particular case of Horndeski gravity [257], as mentioned in section 2.2.3. For instance, in the case f1(f) = αf, the action (2.24) can be shown to be equivalent to the action (2.9) with K = X/2, G3 = 0, G4 = 1/2, G X2 ln5 a= - [92]. As in all of these theories, the coupling parameter is dimensionful and, specifically, it has dimensions of an inverse curvature. It is thus natural to expect that the strongest constraints on the theory should come from physical systems involving high curvature: BHs, NSs and the early Universe. We postpone a discussion of BHs and NSs to sections 3 and 4, respectively. Here we anticipate the observational bounds that have been derived. Most bounds have been derived in the weak-coupling approximation, where one expects L , 2.25GB ( ) ( )a where L is the typical curvature radius in the system under consideration. Thus, Solar System constraints—such as those derived by measuring the Shapiro time delay of the Cassini probe [35]—give a mild bound 10 cmGB 13a , which is in fact of the order of an astronomical unit. On the other hand, as we shall discuss in section 3, BHs in this theory carry a scalar charge, and observations of BH low-mass x-ray binaries give a constraint which is six orders of magnitude stronger [43]: 5 10 cm 2.26GB 6 ( )a ´ (in the units of equation (2.24)). As expected, this constraint is comparable to the typical size of a stellar-mass BH. On the other hand, the only bound on EdGB gravity as an exact theory is of theoretical nature, because the existence of BH solutions implies that GBa be smaller than the BH horizon size [73]; this bound implies M 0.691GB 2 a [75]. Thus, the observational constraint (2.26) is likely to be a good estimate also for the exact EdGB gravity. Class. Quantum Grav. 32 (2015) 243001 Topical Review 25 As previously mentioned, the bounds listed above are clearly satisfied if one assumes that quadratic curvature corrections become relevant only at the Planck scale. Nonetheless, they represent the best constraints on quadratic gravity to date, and they were obtained without any a priori assumptions on the regime in which deviations from GR should be relevant. 2.4.2. Chern–Simons gravity. While the terms proportional to f1, f2 and f3 in the action (2.21) are all associated with qualitatively similar corrections to GR, the term proportional to f4 is peculiarly different, to the extent that the special case f1 = f2 = f3 = 0 describes a specific theory (Chern–Simons gravity) which has been widely scrutinized in recent years (see [258] for a review). At variance with EdGB gravity, to avoid higher-order derivatives in the field equations Chern–Simons theory must be considered as an EFT. Almost all work so far has focused on the special case f4 = αCSf, working perturbatively in the coupling constant αCS. Then the action reads S g x R V RR 1 16 d 2 , 2.274 CS( ) ( )⎡⎣ ⎤⎦*òp f f f a f= - -   - +m m and most of the literature considered the case of a vanishing scalar potential: V(f) = 0. Like the Gauss–Bonnet term, the Chern–Simons term RR* is also a topological invariant, so that if f4 = const the theory is equivalent to GR. For historical reasons, Chern–Simons gravity comes in two flavors: (i) a nondynamical version in which the scalar kinetic term in (2.27) is absent, and (ii) a theory where the scalar is a true dynamical degree of freedom, that goes under the name of dCS gravity. These two theories are actually very different from each other. Despite some confusion in the literature, only the nondynamical theory is parity breaking, whereas dCS gravity simply has different solutions than GR for spacetimes which are not reflection-invariant, as in the case of spinning objects. Furthermore, the nondynamical version introduces a constraint, RR 0* = , arising from the variation of the CS action with respect to the nondynamical scalar field [259]. This constraint limits the space of solutions of the modified gravitational equations and introduces other problems [258]. For these reasons, the dynamical version of the theory has received much more attention in recent years. It can be shown that any spherically symmetric solution of GR is also a solution of dCS gravity [254], and this makes it challenging to distinguish between the two theories. On the other hand, dCS gravity is almost unique as an extension of GR, as it predicts corrections only in the presence of a parity-odd source such as rotation. Among the most studied predictions of the theory are an amplitude birefringence in GW propagation [258] and modified spinning solutions, including corrections to Kerr BHs and rotating NSs, to be discussed in detail in sections 3 and 4. To lowest order in the rotation rate, the CS modification to GR only affects the gravitomagnetic sector of the metric. Tests of the theory might therefore rely on frame- dragging effects. Using the results of Gravity Probe B [260], [45] derived the bound 10 cm. 2.28CS 13( ) ( )a < As mentioned above, dCS gravity should be interpreted as an EFT, and to have perturbative control requires MCS 2a  1. This requirement is stronger than the bound (2.28) for BHs with masses M108Me. Similar bounds come from the Lense–Thirring effect as measured by the LAGEOS satellites, which have also been used to constrain the nondynamical version of the theory [258]. Note that these bounds are of the order of an astronomical unit, as expected from the previous dimensional analysis. The detection of GWs from an extreme mass-ratio inspiral Class. Quantum Grav. 32 (2015) 243001 Topical Review 26 (EMRI) can potentially yield constraints of the order 10 cmCS 10( )a < or even determine the Chern–Simons parameter with fractional errors below 5% [261]. Since large- curvature environments are expected to put stronger bounds on the theory, the optimal systems to constrain quadratic gravity are compact binaries. Indeed, [262] derived projected bounds that are six orders of magnitude more stringent than the one above by considering future GW detection of the late inspiral of BH binaries. Similar bounds were also recently estimated in [86] by analyzing CS corrections of rapidly spinning Kerr BHs. Such corrections could be constrained from electromagnetic observations of accreting stellar mass BHs such as those in low-mass x-ray binaries, e.g. GRO J1655−40 (see e.g. [263]). 2.5. Lorentz-violating theories While Lorentz invariance has been tested to high precision in the matter sector [201, 264–267], constraints in the gravity sector are much weaker. Constraints on Lorentz invar- iance in gravity beyond those obtainable in the Solar System have received much interest after Hořava [200] pointed out that a power-counting renormalizable theory can be constructed by giving up Lorentz invariance in gravity. We will focus here on Einstein-Æther and khrono- metric gravity, which are the most generic theories violating boost symmetry in gravity at low energies. A very clear review of these theories can be found in [268]39. 2.5.1. Einstein-Æther. To break boost invariance in the most generic way, one can describe the gravitational degrees of freedom by means of a metric and a timelike vector field, u, usually referred to as the ‘æther.’ Up to total divergences, the most general action composed of the metric and two or fewer derivatives of the æther, and that couples the æther minimally to matter (so as to enforce the weak equivalence principle and experimental evidence against the existence of ‘fifth forces’) is given by the Einstein-Æther action [47, 274, 275] S G g R M u u x S g 1 16 d , , 2.29 Æ Æ 4 mat( ) ( ) ⎡⎣ ⎤⎦ò p = - -   + Yab mn a m b n mn where M c g g c c c u u g , 2.301 2 3 4 ( )d d d d= + + -ab mn ab mn m a n b n a m b a b mn ci (i = 1,...,4) are dimensionless couplings, and Ψ denotes the matter degrees of freedom. In this section we do not assume GN = 1; the ‘bare’ GÆ is related to the ‘Newtonian’ gravitational constant GN measured locally by Cavendish-type experiments via [276] G G c c 2 2 . 2.31N Æ 1 4( ) ( )= - + To enforce the timelike character of the æther, one has to impose g u u 1, 2.32( )= -mn m n either implicitly or by adding a Lagrange multiplier ℓ g u u 1( )+mn m n in the variation of the action above. The field equations for Einstein-Æther theory can be derived by varying the action (2.29) with respect to gab and uμ, while imposing the constraint (2.32). This results in the following 39 An alternative parametrized EFT approach to Lorentz violations in both the gravity and matter sectors was developed by Kostelecky et al [264, 269–271]. For binary pulsar constraints in the EFT framework of [264, 269– 271], see [272, 273]. Class. Quantum Grav. 32 (2015) 243001 Topical Review 27 modified Einstein equations: E G T G T8 0, 2.33Æ Æ mat ( )pº - - =ab ab ab ab where Tmat ab is the matter stress–energy tensor T J u J u J u c u u u u u J c u u u c u u M u u g 1 2 , 2.34 Æ 1 4 2 4 ( ) ( ) ( ) ( ) ( ) ( ) ˙ ˙ ˙ ( ) ( ) ( ) ( ) ⎡⎣ ⎤⎦ ⎡⎣ ⎤⎦ =-  - - -   -   +  + + -   ab m a m b a m b ab m m a m b a m b m n m mn a b a b sr mn s m r n ab J M u= a m ab mn b n , and u u u˙ º a b b a. These are completed by the æther equations J c u u g u uÆ 0. 2.354 ( )( )˙ ( )º  +  + =m a an a n a mn m n Strong constraints on the coupling constants ci come from Solar System tests. This is because Einstein-Æther theory predicts that the (dimensionless) preferred-frame parameters α1 and α2 of the PPN expansion will in general be nonzero functions of the ciʼs [46, 47]. Because Solar System experiments constrain 101 4a - and 102 7a - [2], one can expand the theory in α1 and α2, and reduce the parameter space to just two independent couplings c c c1 3º  . The remaining couplings are given by c c c c c22 1 2 1 3 3 2( )= - - + c3 ,1 1 2( ) ( ) a a+ , c c c ,4 3 2 1 1 2( ) a a= - + [46, 47]. Further constraints on the two independent couplings c± come from requiring that the theory should have positive energy (i.e. no ghosts) and that Minkowski space should be linearly stable (i.e. no gradient instabilities) [2]. Einstein-Æther theory predicts the existence of not only spin-2 gravitational perturba- tions (like in GR), but also spin-1 and spin-0 gravitational perturbations. All these propagating modes have speeds that are functions of the ci, and which differ in general from the speed of light [277]. However, if these modes were propagating at speeds lower than the speed of light, photons (or relativistic particles) could Cherenkov radiate into the gravitational field and lose energy to these modes, and this would lead to (unobserved) experimental consequences [278]. Therefore, one has to impose that the speed of the spin-2, spin-1 and spin-0 gravitons is larger than (or equal to) the speed of light. Taking into account these constraints, one obtains the viable region plotted in cyan in figure 38 (left panel) for the two independent couplings c±. As previewed in that figure and discussed in section 6.1, more stringent constraints on c± come from binary pulsar data [48, 49]. 2.5.2. Khronometric theory. If we impose that the æther is always hypersurface-orthogonal, one can express it as u T g T T , 2.36( )= - ¶ - ¶ ¶ a a mn m n where T is the hypersurface-defining scalar and the constraint (2.32) has already been enforced. By assumption, surfaces of constant T foliate the spacetime, and one can re-express the action (2.29) adapted to this ‘preferred time’ T. This yields a different theory, described by the ‘khronometric theory’ action [200, 279, 280] Class. Quantum Grav. 32 (2015) 243001 Topical Review 28 S G T x N h K K K R a a S g 1 16 d d 1 1 1 1 1 , , 2.37 ij ij i i K Æ 3 2 3 mat ( )( ) ⎛ ⎝⎜ ⎞ ⎠⎟ ⎡⎣ ⎤⎦ ò b p l b b a b = - - + - + - + - + Y mn where N gTT 1 2( )= - - is the lapse function, Kij is the extrinsic curvature of T= constant hypersurfaces, hij is the induced spatial metric on those hypersurfaces, R3( ) their three- dimensional Ricci curvature, a Nlni i= ¶ , and the æther is now related to the lapse via u N .Td= -a a We have also redefined the theoryʼs parameters via c c c c c, , . 2.382 3 1 4 1 ( )l b aº º + º + It should be stressed at this stage that the action (2.37) only depends on three couplings, as opposed to four in the action (2.29). This is because the hypersurface-orthogonality constraint (2.36) makes it possible to re-express one of those four couplings in terms of the remaining three without loss of generality. The field equations of khronometric theory are obtained by replacing the hypersurface orthogonality constraint (2.36) in the action (2.29), and then varying the action with respect to gab and T; they are E u2Æ 0, 2.39( )( )+ =ab a b T T Æ 0. 2.40( ) ⎛ ⎝⎜ ⎞ ⎠⎟ -  =m m a a Note that equation (2.40) actually follows from equation (2.39) and from the conservation of the matter stress–energy tensor, i.e. the only independent equations are the modified Einstein equations and the equations of motion of matter [280]. By comparing this set of equations with the Einstein-Æther equations (2.33) and (2.35), it is easy to see that the hypersurface- orthogonal solutions of Einstein-Æther theory will also be solutions of khronometric theory. The converse is true in spherical symmetry [50, 102, 280, 281], but not in more general situations. For instance, slowly rotating BH solutions are different in the two theories [99, 101, 281]. As in the Einstein-Æther case, in Khronometric theory the PPN preferred-frame parameters α1 and α2 are nonzero and functions of the couplings. In light of the bounds 101 4a - and 102 7a - [2], one can expand khronometric theory in α1 and α2. As a result, one is left with two independent parameters (say β and λ), while the third parameter α is related to the first two by40 2 ,1 2( )a b a a= + . From the hypersurface orthogonality constraint (2.36), there are no propagating spin-1 gravitational modes. Requiring positive energies, linear stability of Minkowski space, and the absence of gravitational Cherenkov radiation for the remaining spin-0 and spin-2 degrees of freedom still selects a sizeable region of the parameter space (λ, β) [50, 98, 278], shown in cyan in figure 38 (right panel). Further constraints come from requiring that the theoretically predicted Big Bang nucleosyntesis elemental abundances agree with observations [48, 49, 276]; these constraints are much stronger for Khronometric theory than for Einstein-Æther [47, 276], and are represented by the orange region in figure 38 (right panel). 40 Though it may seem that the conditions 01 2a a= = would reduce the dimensionality of the parameter space to a one-dimensional subspace, both α1 and α2 happen to vanish for α = 2β in Khronometric theory. Thus, the conditions α1 = α2 = 0 still select a two-dimensional subspace. However, this only holds at the origin in α-space, and so saturating the bounds to 101 4a » - and 102 7a » - reduces to a one-dimensional subspace. We refer the reader to [49] for a detailed discussion. Class. Quantum Grav. 32 (2015) 243001 Topical Review 29 As reviewed in that figure and discussed later in section 6.1, even more stringent constraints on λ and β come from binary pulsar observations [48, 49]. 2.5.3. Horǎva gravity. The khronometric theory action (2.37) is particularly interesting because it is the low-energy (or infrared) limit of Hořava gravity [200], a renormalizable quantum field theory which has only spatial diffeomorphism invariance. The complete action of Hořava gravity is [279] S G T x N h L M L M L 1 16 d d , 2.41H H 3 2 2 2 4 4 4 6 ( ) ⎛ ⎝⎜ ⎞ ⎠⎟     òp = + + where L K K K R a a 1 1 1 1 1 2.42ij ij i i 2 2 3 ( )( )l b b a b = - + - + - + - is the Lagrangian density of Khronometric theory (see equation (2.37)), Må is a mass scale, and L4 and L6 are terms of fourth and sixth order in the spatial derivatives, but contain no derivatives with respect to the preferred time T. Complete constraints on Må are somewhat elusive to obtain, and are probably one of the most important open questions in Hořava gravity [282]. The reason is that one would expect Lorentz violations to percolate from gravity into the matter sector, where Lorentz symmetry has been verified to high precision by particle physics and cosmic-ray experiments [201, 264– 267]. However, several mechanisms have been put forward to suppress this percolation. For instance, it has been suggested that the operators that violate Lorentz symmetry in the matter sector might be finely tuned to much smaller values than those in the gravity sector. Also, Lorentz invariance in matter might be an emergent property at low energies [283], as an accidental symmetry [284] or due to renormalization group phenomena [285, 286] . Finally, it has been shown that two sectors with different Lorentz violation degrees can easily coexist if their interaction is suppressed by a high mass scale [287], and this could be the case for the gravity and matter sector. Therefore, taking into account only the gravitational bounds (i.e. assuming that percolation of Lorentz violation into the matter sector is efficiently suppressed), one obtains M 10 eV2   - from sub-millimeter gravitational experiments. Also, perhaps surprisingly, Må has an upper bound (Må  1016 GeV) from the requirement that the theory remains perturbative at all scales [288–290], so that the power-counting renormalizability arguments proposed in [200] apply. Three things are worth stressing about the higher-order derivative terms L4 and L6 in the action. First, the presence of sixth-order spatial derivatives is essential for power-counting renormalizability [200]. Second, the fourth- and sixth-order terms in the spatial derivatives generally lead to nonlinear dispersion relations for the gravitational degrees of freedom of the theory, i.e. the spin-2 and spin-0 gravitons (the latter present in the theory because of the foliation-defining scalar T) satisfy k M k M k , 2.432 2 4 2 4 6 4 6 ( ) ⎛ ⎝⎜ ⎞ ⎠⎟ ⎛ ⎝⎜ ⎞ ⎠⎟     w a aµ + + +  where ω and k are respectively the frequency and the wave-number, while α4 and α6 are dimensionless constants. Because such a dispersion relation allows for infinite propagation speeds in the ultrav