Black hole elasticity and gapped transverse phonons in holography

Journal of High Energy Physics, Jan 2018

Lasma Alberte, Martin Ammon, Matteo Baggioli, Amadeo Jiménez, Oriol Pujolàs

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

Black hole elasticity and gapped transverse phonons in holography

HJE Black hole elasticity and gapped transverse phonons in holography Lasma Alberte 0 1 2 5 6 7 Martin Ammon 0 1 2 3 6 7 Matteo Baggioli 0 1 2 4 6 7 Amadeo Jimenez 0 1 2 3 6 7 Oriol Pujolas 0 1 2 6 7 0 71003 Heraklion , Greece 1 Department of Physics, University of Crete 2 Strada Costiera 11 , 34151 Trieste , Italy 3 Theoretisch-Physikalisches Institut, Friedrich-Schiller-Universitat Jena 4 Crete Center for Theoretical Physics, Institute for Theoretical and Computational Physics 5 Abdus Salam International Centre for Theoretical Physics , ICTP 6 Campus UAB , 08193 Bellaterra (Barcelona) , Spain 7 The Barcelona Institute of Science and Technology , BIST We study the elastic response of planar black hole (BH) solutions in a simple class of holographic models with broken translational invariance. We compute the transverse quasi-normal mode spectrum and the propagation speed of the lowest energy mode. We nd that the speed of the lowest mode relates to the BH rigidity modulus as dictated by elasticity theory. This allows to identify these modes as transverse phonons | the pseudo Goldstone bosons of spontaneously broken translational invariance. In addition, we show that these modes have a mass gap controlled by an explicit source of the translational symmetry breaking. These results provide a new con rmation that the BHs in these models do exhibit solid properties that become more manifest at low temperatures. Also, by the AdS/CFT correspondence, this allows to extend the standard results from the e ective eld theory for solids to quantum-critical materials. Holography and condensed matter physics (AdS/CMT); AdS-CFT Corre- 1 Introduction Elasticity Holographic solids from massive gravity Perturbations Translational symmetry breaking pattern Transverse quasi-normal modes The quasi-normal mode spectrum Transverse velocity and elasticity Discussion A Solid EFT and elasticity B Numerical details of QNM computation 1 and dynamics | but it is badly broken in condensed matter. Real-world physical systems break this symmetry in various ways. Already a perfectly ordered lattice structure accommodates phonon excitations, the collective vibrational excitations of the lattice. Their inevitable presence stems from the symmetry breaking pattern: in solid state, the phonons can be understood as the Goldstone bosons associated to translational symmetry breaking [ 1, 2 ]. Impurities, defects and other forms of random disorder provide an additional source of translational symmetry breaking. Both e ects have a deep impact on the transport parameters of the material as well as on the phonon properties. An example of the interplay between various sources of translational symmetry breaking appears in (pinned) charge-density waves and in Wigner crystals, see e.g. [3, 4]. { 1 { The main goal of this work is to show that the black holes in rather simple gravitational theories exhibit completely analogous behaviour, displaying transverse phonon1 excitations whose properties are very sensitive to how the translational symmetry is broken. We shall focus on background geometries with Anti-de-Sitter (AdS) asymptotics, in order to have a simple Conformal Field Theory (CFT) interpretation of the physics via the standard holographic AdS/CFT dictionary [ 5 ]. The simplest class of holographic models that allow to include the e ects of breaking the translational invariance and of momentum relaxation are of Holographic Massive Gravity type [6{9]. The motivation to consider these models is that they are computationally more tractable since the background geometries are homogeneous. The analysis of the elastic properties of these models was initiated in [9{11]. be identi ed as a bona de transverse pseudo phonon. In this work we will show that the latter property is indeed realized. This con rms that the Holographic Massive Gravity models do indeed incorporate the elastic response. These models can therefore be interpreted as CFTs with elastic properties or, equivalently, as elastic materials in a critical scale-invariant regime. Along the way we will uncover some interesting properties of these modes, such as their ability to acquire a mass-gap and what determines it. 1.1 Transverse phonons Let us now explain what we mean by bona de phonons. From the low-frequency point of view, the hallmark of phonons is in their propagation sound speeds. Indeed, it is well known that in solids with negligible damping and mass-gap generating e ects the phonon dispersion relations are linear, and the longitudinal and transverse sound speeds, cL; T , are given by (in 3 + 1 dimensions): ! = cL; T k ; c 2L = + 43 " + p ; c 2T = " + p (1.1) (1.2) direction of propagation. invariance. is familiar from relativistic hydrodynamics.2 where and are the bulk (or compression) and shear (or rigidity) moduli, and ", p stand for the energy density and the pressure. Most textbooks [1{3, 12] derive these relations in the non-relativistic limit where the denominator simpli es to just ". One can see in various ways that the extension to the relativistic case proceeds by the replacement " ! " + p, as We give another derivation of (1.2) from the E ective Field Theory (EFT) for Solids [1, 2, 14{16] in appendix A. Let us emphasize here that an illuminating aspect of the EFT of 1We recall that solid materials support both longitudinal as well as transverse (or shear) sound waves. The transverse sound waves are vibrational modes with material displacements that are orthogonal to the 2See also [13] for a recent discussion in the case of spontaneous and explicit breaking of translational { 2 { Solids is that it makes it very clear that the connection between the phonon velocities and the elastic moduli stems from symmetry principles. In the present work we focus on transverse phonons. Solids (i.e. materials with a non-zero rigidity modulus ) stand out as the materials that support propagating shear or transverse waves. Therefore the black hole geometries with a natural notion of the rigidity modulus [10] o er a possibility to realize transverse phonons in the Goldstone-boson sense that can be tested by the relation (1.2). The main result of this work is to show in what limit of the Holographic Massive Gravity (HMG) models the eq. (1.2) gives a very good approximation for the transverse (pseudo) phonon QNMs speed, with the natural notion of elastic modulus introduced in [10]. This justi es that these QNMs can be identi ed as transverse pseudo phonons. In turn, this con rms that the black brane backgrounds that we consider do have elastic properties akin to the standard solids. In the CFT interpretation of the same events, what we shall describe is a transverse phonon in a critical solid material. 1.2 Gapping The only obstruction in realizing holographically transverse phonons is that the physics of these modes is actually much richer than simply eq. (1.2). Being Goldstone bosons, they are also sensitive to any explicit source of the translational symmetry breaking. This is analogous to the spontaneous breaking of the chiral symmetry by the quark hq qi condensate in QCD and the associated appearance of pions as the Goldstone bosons. In this case the quark masses give rise to an additional explicit breaking of the chiral symmetry and this gives the would-be massless pions a mass. For translational symmetry, what an \explicit breaking" means is a bit more subtle and perhaps confusing to some readers. The physically relevant thing is the presence of another source of translational symmetry breaking, the consequence of which is that the transverse phonons acquire a mass gap. In the real world, systems with a clear notion of multiple breakings of translational symmetry exist, perhaps the sharpest examples being the pinned charge-density waves and Wigner crystals. In the models presented here there will be a clear notion of distinct sources of translational symmetry breaking | in the CFT parlance, the breakdown will take place di erently at di erent energy scales. In the present work we shall step aside from the discussion of what is the corresponding condensed matter phenomenon that ts best as a CFT interpretation of how the transverse phonons become gapped in our models. We prefer to stick to a more agnostic discussion in terms of symmetries only. The net e ect, in any case, is that the degrees of freedom that we will nd approximately do obey (1.2) and are gapped in the simplest models analyzed here. By analogy with pions, it is fair enough to say that this is due to an additional explicit source of the breaking of translational invariance which appears to be inevitable in our setup. Going back to the matter: in what limit should we expect to nd modes that can be identi ed as transverse phonons? The answer is obvious by again looking at the EFT describing QCD at low energies. The (almost tautological) observation is that the properties of the Goldstone bosons are protected provided that the explicit symmetry breaking source is small enough. Hence, a comparison between two scales is required (which is what makes the observation not tautological): the explicit breaking scale vs. the dynamical scale in the { 3 { problem. For QCD this is mq= QCD. Provided mq= QCD 1, the pions still keep their pseudo-Goldstone boson properties. This suggests that in order to possibly realize bona de transverse phonons we need to insure that the explicit breaking source is weak (small), so that the phonon mass gap is small compared to the typical energy scale of the problem.3 In the rst examples of HMG models [6, 8] the explicit breaking is large (comparable to the spontaneous) and this translates into the absence of clearly identi able propagating phonon excitations in the shear channel. As we show below, the situation is reversed when considering HMG models with a metric potential V (X) of the form [9] and N .4 We shall exhibit that this is indeed the case for the representative benchmark model with N = 5.5 QNM with the following properties: As shown below, for 1 this model exhibits a gapped (but light ) transverse phonon i) the mass gap !0 (or, equivalently, the pinning frequency ) is parametrically smaller than the typical scale of the system, !0 . Let us emphasize that the `typical' scale should be an intrinsic scale in the problem. This excludes the temperature T from playing that role, as it should be since we expect that the solid behaviour melts down at too high T . Having again QCD in mind, a natural reference value for is the typical separation in the frequency plane of the QNMs (at low temperature). As a proxy for this we shall take the absolute value of the second QNM, that is, Thus, a good measure of how close to the pure Goldstone limit we are is the ratio !0 : j!2j.6 (1.4) In our benchmark models, this ratio is controlled by the parameter and can be made arbitrarily small. Consequently, we obtain a parametrically light transverse phonon QNM. ii) the propagation speed of this mode complies with (1.2). The agreement between the two quantities on the left and right hand sides of (1.2) is, of course, not perfect because of the explicit breaking | the mass gap. Our phonons are pseudo-Goldstone bosons, and therefore their properties have to coincide with the pure gapless Goldstone limit (1.2) only within a certain precision. The same logic suggests that a good 3In the opposite limit with large explicit breaking, there is no light scalar mode, and the sound speed of the lightest mode does not need to obey (1.2). 4The model in ref. [8] corresponds to too small. = 0, and those of [6] to N = 1=2. These values of N and are 5Let us emphasize here that we expect our results to hold for potentials V (X) more general than (1.3), 6We reserve !1 for the lowest QNM | the light phonon pole. Let us also add that the scale of the QNM mass separation, j!2j, is typically comparable to the energy scale obtained from the entropy density. { 4 { u. The dashed line represents the mass pro le in the models [8]. The colored line is the setup which we analyze in this work and where we would expect a propagating gapped phonon to appear in the spectrum with a small mass gap !0 controlled by the UV graviton mass. In practice, one can achieve this hierarchy using the HMG models of the form (1.3) with large enough N and [9]. indicator of the precision with which relations like (1.2) should hold is precisely the measure of how close to purely spontaneous case we are. This reduces to positive powers of (1.4). Our numerical results are completely in line with this picture. Let us sketch why a HMG model like the one given in (1.3) manages to generate light transverse phonons as seen from the gravity dual picture. As explained in [9], the only gauge invariant notion of an order parameter for the translational symmetry breaking in the presence of dynamical gravity is the graviton mass term, mg. For generic potentials V (X), one easily nds that mg / V 0(X). The important point is that on the background solution this graviton mass term depends on the holographic direction. Therefore, in general, there is an evolution of the amount of the symmetry breaking as we move along the radial direction. In the CFT language, this translates into the fact that along the RG ow generated by these deformations the amount of the breaking of translations can vary. This, however, only happens provided that V 0(X) is non-constant. It is clear then that by an appropriate choice of the function V (X) one can either break the translations more in the UV or in the IR | there are de nitely distinct sources of symmetry breaking. The benchmark models (1.3), with large enough N and , are designed to break the translations most strongly in the IR region of the ow, with a separable and smaller breaking taking place in the UV region. This corresponds to a graviton mass pro le that peaks in the IR region, as depicted in gure 1. By analogy with the standard terminology in AdS/CFT, we shall refer to the breaking associated to the UV region as explicit, and the one in the IR as spontaneous. See section 3.2 for more details on this point. { 5 { An inescapable consequence of the smallness of the phonon mass gap is that they are easily excitable and therefore also very sensitive to thermal and other dissipative e ects. Indeed, at temperatures as small as the mass gap thermal e ects are expected to be noticeable in the phonon dispersion relation, !2 = c2T k2 + !2, with both !0 and cT possibly having 0 temperature corrections. Since this can jeopardize the correct identi cation of the phonon sound speed from the dispersion relation, it is important to understand and isolate these dissipative e ects. Fortunately, there are two facts that will help in this task: { First, the analytic structure of the retarded Green's function relevant for the transport properties constrains quite a lot the way the dissipative e ects enter in the nal dispersion relation for the physical phonons (that are identi ed as QNMs). To a very good approximation, we nd that the poles in the complex frequency plane ! for the lightest QNMs follow simply from the standard phonon wave equation corrected by a mass term and a friction term: ! 2 0 (T ) = 0: The parameters ( , cT , !0) are real-valued and depend on temperature and other control parameters ( , etc.). In Fourier space this leads to the simple quadratic equation ! 2 + i ! = ! 02 + c2T k2 : It is clear that is the only parameter encoding the dissipative e ects and that it can be understood as an absorption or attenuation rate. We also note that whenever both !0 and are small compared to the scale of the system, , the and !0 terms in the wave equation are small perturbations of the phonon degree of freedom introduced above. Indeed, for !0 ; 1 there is a regime in momentum, !0; cT k , for which the mode is propagating with an almost constant sound speed d!=dk = cT . In this situation the inhomogeneous elastic transverse response is dominated by a sharp transverse phonon collective mode.7 At k = 0, however, the location of the phonon poles is very sensitive to the comparison between and 2!0. Indeed, the two roots of (1.5) read ! = r !2 0 1 4 2 i 2 (1.5) (1.6) HJEP01(28)9 and exhibit a characteristic crossover between the under-damped and over-damped regimes (the latter corresponding to the eigenfrequencies having no real part). The crossover takes place whenever = 2!0 and is visualized as a characteristic `collision' [ 9, 17 ] of two purely imaginary poles which after the collision go into a pair of (propagating) poles with a non-zero real part.8 7The presence of a sharp collective mode approaching the real axes is also evident in the optical conductivity properties of the model once a nite charge density is added [9]. 8Note that when the real part of the QNM frequency does not vanish the absolute value j!j coincides with !0, which is what we identify as the gap. { 6 { { Second, the dissipative e ects (in our notation, ) must reduce at low temperatures. Since controls whether this mode is over/under damped, a good enough criterion to ensure that the dissipative e ects are small is to require that !0. The black hole solutions that we shall consider here do have a well-de ned extremal zerotemperature regime in which the phonon modes are still guaranteed to exist by the symmetry breaking pattern. We nd by direct numerical computations that in our black hole solutions we can lower enough T , so that !0 is satis ed for the lowest QNM. The sound speed cT is then easily extracted by looking, e.g., at the real part of its frequency. From this exercise we can test whether the relation (1.2) holds, and a good agreement is found. Note that this behaviour of the quasi-normal shear modes is conceptually very di erent from the one seen in hydrodynamics (or just the thermodynamics) of viscous uids with momentum relaxation.9 This stark contrast is of course completely expected, because the black hole solutions that we study here have all the properties of solid materials, the dynamics of which di er from that of uids. This di erence is seen only at low temperatures, as expected, since typically solids do melt down into a uid phase at a high enough temperature. For the black hole solutions studied here, the interplay between the solid behaviour (in the phonon dispersion relation) and the uid behaviour is actually continuous: the parameters and !0 vary continuously with the external control parameters, going through and out of the over-damped regime. Therefore these BH solutions behave like solids that `melt' at high temperatures in a continuous transition.10 We shall leave for the future a thorough investigation of this transition. Here we limit ourselves to presenting the regime in which the solid behaviour, with propagating transverse phonons complying with (1.2), is exhibited. Let us emphasize that our interest lies in a completely opposite region with respect to the so-called incoherent limit [17{31], where the momentum relaxation rate is parametrically smaller than the typical energy scale rel 1 T . Indeed, in that regime momentum dissipates rapidly and no propagating degree of freedom survives; physics is just dominated by di usion. On the contrary, for our setup having a small relaxation rate is crucial in order to qualify the pseudo-phonon mode as a proper propagating excitation. Context. There have been a large number of results in the context of holographic models that include momentum relaxation [6{9, 32{43] in the direction of mimicking and understanding the physics of the explicit breaking of translational invariance. As a counterpart, the spontaneous breaking of translations and, in particular, the study of the phonon modes in holography is still lacking a concrete playground and robust results. We hope that this work contributes to clarify some aspects of this debate. Within the gauge-gravity duality framework the study of the Goldstone modes has been so far mainly focused on the spontaneous breaking of the U(1) symmetry in rela9Nevertheless, by using a gradient expansion in small frequency ! and momentum k, one can infer the form of the hydrodynamic mode even in the presence of an explicit breaking. See [13] for a recent work on 10This is in line with our previous studies [11] that suggest that these black hole solutions can be interthe topic. preted as viscoelastic materials. { 7 { tion to the onset of super uidity and on the e ects of Lorentz violation on the Goldstone theorem [44{46]. The rst study of the physcics of the pseudo-Goldstones in the context of strongly coupled theories has been initiated in [47] and [48]. Just recently, the Ward identities for phonons and pseudo-phonons have been analyzed in [49], but without an explicit and computable model. One possible way of breaking spontaneously the translational invariance in the context of holography relies on the existence of a possible nite momentum instability of the Reissner-Nordstrom AdS black hole [50{58]. In those situations, new ground states characterized by spontaneous lattice structures appear and the dual ed theory is usually related to the physics of striped materials and charge density waves. Despite all the existing models available, a precise identi cation of the collective phonon mode has not been accomplished yet. In addition to our fundamental question of obtaining phonon's physics and elastic theory within the holographic framework, we can mention another two orthogonal and very interesting motivations in the context of holography and strongly coupled materials. i) Recently in [59], it has been proposed that the transport properties, in particular the electric conductivity, of the bad metals exhibiting linear in T resistivity could be explained by the interplay of explicit and spontaneous breaking of translational symmetry. The analysis has been performed within the framework of hydrodynamics and lacks a concrete model able to check the expectations and provide microscopic information on the various transport coe cients appearing in the hydrodynamic expansion. Our model is certainly a promising candidate for such a test. ii) E ective eld theories for the spontaneous breaking of Poincare symmetry have been developed in the context of weak coupling and at space in [ 14, 16, 60 ] using a language very similar to ours. It is a very valuable question and direction to understand how to embed them into the gauge-gravity duality. The discussions initiated in [10] and our computations provide a rst step towards that task. The manuscript is organized as follows. In section 2 we introduce the theory of elasticity from the eld theory and the holographic sides and we show how to extract the shear modulus in holography. In section 3 we present our holographic model for solids. In section 4 we analyze the spectrum of the transverse excitations of the system, we identify the gapped and damped transverse phonon, and we study in detail the speed of shear sound matching the results obtained from the shear modulus. Finally, in section 5 we discuss our results and conclude. In appendix A we review the EFT for phonons from a modern perspective and in B we provide more details on the numerical analysis and our computations. Note added. While this work was being nished, refs. [61] and [62] discussing similar issues in di erent holographic models were also being completed. 2 Elasticity In the standard mechanical linear response theory [3, 12] the deformation of a solid due to applied stress is described in terms of the displacement vector ui = x0i xi. The coordinates { 8 { xi and x0i correspond to the coordinates of material points in the body before and after the deformation respectively. Under applied forces the distance dl between di erent points changes to dl0 parametrized as dl02 ij dx0idx0j = ( ij + 2uij )dxidxj ; where uij is the strain tensor de ned as 1 2 uij = This deformation brings the material out of its original state of equilibrium and creates internal stress trying to bring the body back to it. Due to the short range of the molecular forces creating the internal stress, the total force acting on any part of the solid body can be written as a surface integral: H dSj ij , where ij is the stress tensor. 2.1 Elastic sound waves The quantity of interest for this work is the speed of propagation of elastic deformations in solids. The sound speed in an elastic medium is determined by its elastic properties. For homogeneous and isotropic solids (that is, upon coarse-graining on the microscopic structure) the linear elastic response parameters are the shear elastic modulus, , and the compression modulus, . These relate the stress tensor ik to the strain tensor uik as ik uik 1 j d uj ik ; where F is the free energy of the medium. The bulk modulus describes the change in the volume of the body while the shear modulus describes the change in the shape of the body. For a pure shear deformation, = 0, and only the traceless part of the stress tensor is non-zero, so that The Newtonian equations of motion for the propagation of small deformations inside the elastic medium are simply given by (T) = 2 ui(kT) : ik ui = deformation (transverse). to be: where is the mass density of the body. For a shear deformation in one particular direction the above equations reduce to plane wave equations for elastic waves propagating either in the direction of deformation (longitudinal) or in a direction perpendicular to the The speeds of propagation of the two types of waves in relativistic systems are known c2T = " + p ; c2L = + 43 : " + p For uids, the modulus of rigidity is equal to zero so that c2T = 0 and only longitudinal sound waves propagate. The derivation of these expressions can be obtained via standard hydrodynamic methods [63]. We present another derivation by the standard EFT methodology for describing solid and uid materials [ 1, 2, 14, 16, 60 ] in appendix A. We will make an explicit use of the expression (2.6) in the following. (2.1) (2.2) (2.3) (2.4) (2.5) (2.6) Another way to think about the displacement ui is to consider it as an in nitesimal diffeomorphism under which the coordinates transform to xi ! x0i = xi + ui. This leads to a transformation of spatial metric perturbations as hij ! h0ij = hij R Re GTij Tij : c2T = 3 1 2L " + p Re i(j0) : (3) ij The equation (2.8) can be written in the form of a linear response as with the retarded Green's function given by R hTij i = GTij Tij ij (0) R GTij Tij = (3) 23L i(j0) : ij On the other hand, the expectation value of the stress tensor due to a deformation uik on the boundary is given by h ij i = F uij = 2 Sren = ij hTij i = 2GTRij Tij uij : Restricting to the transverse traceless components and comparing to (2.4) we nd that the shear modulus can be expressed through the real part of the retarded Green's function as: From (2.6) we thus nd that the transverse sound speed in relativistic dual solids can be expressed in terms of the coe cients in the near boundary expansion of metric perturbations as Assuming that the coordinate system x0i was such that the metric perturbations were vanishing, i.e. h0ij = 0, allows to make the identi cation hij = 2uij and thus associate the strain tensor to metric perturbations. This identi cation is particularly useful in the context of holography. According to the standard prescription, the stress-energy tensor in the dual theory can be obtained by the variation of the renormalized Euclidean boundary action [64, 65]: In the following sections we shall determine the sound speed of the lowest quasi-normal modes in the transverse sector of the metric perturbations and compare to the expression above. We note that the shear modulus is contained in the real part of the Green's function, in contrast to the shear viscosity which lies in its imaginary part. The shear viscosity has been computed in holographic massive gravity models in [11, 66, 67] and shows a violation of the Kovtun-Son-Starinets bound [68]. In order to understand the transport properties in the shear sector dealing with viscoelasticity and in particular their de nitions in terms of Kubo formulas in full detail one would need to perform a proper and generic hydrodynamic description of these models in the spirit of [13, 67, 69]. 3 Holographic solids from massive gravity We consider a 3 + 1 dimensional gravity theory describing holographic solids, as introduced in [10, 11] (see also [9]): where l is the AdS radius, m is a mass parameter of dimension [m] = L 1, and X and the indices I; J = fx; yg are contracted with IJ . The last term in (3.1) is the Gibbons-Hawking boundary term where is the induced metric on the AdS boundary, and K = r n is the extrinsic curvature with n vector to the boundary. On the scalar elds background ^I = iI xi the metric admits the | an outward pointing unit normal black brane solution ds2 = l 2 u2 du2 f (u) f (u)dt2 + dx2 + dy2 ; so that the background value of X = u2=l2. Note that X is dimensionless. The emblackening factor is given in terms of the background value of the mass potential as: f (u) = 1 + u3 du Z m2l2 u4 V u 2 l 2 M~ u3 ; where M~ is a dimensionful integration constant given by the condition that f (uh) = 0. Henceforth we shall work in terms of the dimensionless length z = u=uh. The potential (1.3) with N = 5 on the background can be expressed as 1 2 1 2 (3.1) (3.2) (3.3) (3.4) (3.5) (3.6) The black brane temperature for the emblackening factor (3.4) with the potential (3.5) is then T l = uh 4 1 4 l uh 3 m2u2 h m2u2 h uh 8 l : The energy density of the black brane in the dual eld theory is given in terms of the integration constant M~ appearing in the emblackening factor (3.4) as " = M~ l2. This de nition coincides with the usual AdS/CFT prescription where " = T tt . In our theory, " = 2 l u Moreover, we de ne the hydrodynamic pressure P to be P hT xxi = "=2. Note that P is not identical to the thermodynamic pressure p de ned in terms of the free energy density as p = . It is the hydrodynamic pressure which will enter the relation (2.14) between HJEP01(28)9 the velocity of the transverse (gapped) phonons and the rigiditiy. (3.7) (3.10) (3.11) (3.12) u uh The rigidity of the holographic solid is encoded in transverse vector perturbations which g = g^ + h ; hij = 1 I = ^I + 'I (3.8) this forms a set of four vector elds. Given the di eomorphism invariance of the theory, one of the components is redundant and can be removed by a coordinate transformation leaving us with three independent vector elds. We can write these in a gauge invariant form as [9]: Ti u 2 hti l 2 l 2 Ui f (u) hui Bi b i l2'i : (3.9) The equations of motion allow one to express the eld Ti in terms of the other two elds leaving us with only two independent equations of motion for the elds Ui and Bi. In terms of the dimensionless quantities z = ; !~ = !uh ; k~ = kuh ; m~ = muh ; B~i = ; U~i = zUi Bi u 3 h the equations of motion for the Fourier components of the elds then read: 1 " f z2 VX U~i !# f f ~2 k ~2 k 2m~ 2VX ~ Ui = z fz0k~22 B~i ; 2m~ 2VX B~i = 2 X z U~i : V 0 VX In these expressions f 0(z) = @zf (z) denotes a derivative with respect to z while the subscript X denotes the derivative with respect to X. In particular, VX0 = @z(@X V (X(z))). We also note that according to the de nitions (3.8), (3.9) the eld Bi has a dimension [Bi] = L3 whereas the eld Ui is already dimensionless. We have introduced U~i since in the cases considered below this will be the eld with the correct scaling dimension. Solving these equations of motion with appropriate in-falling horizon conditions will give the spectrum of the quasi-normal modes within massive gravity or, alternatively, the poles of retarded Greens functions in the dual eld theory. This will be done in section 4. In order to compute the rigidity, it is su cient to consider the homogeneous perturbations, i.e. to take the limit k~ ! 0 in the equations (3.11) and (3.12). In particular, we allow for non-normalizable modes for B~i but not for U~i: Hence, the equation (3.11) admits the trivial solution Ui = 0, while the equation (3.12) reduces to f + !~2 f 2 m~2VX B~i = 0 ; (3.13) An exact result for the Green's function can be obtained by numerically integrating (3.13). HJEP01(28)9 In the limit of small graviton mass m~ 1 it is possible to obtain an analytical perturbative result for both the real and the imaginary parts of such correlator (see [11]). Translational symmetry breaking pattern Let us now comment on how the translational invariance is broken in this model. Speci cally, whether it is possible to have a clear notion of either spontaneous or explicit breaking. As advanced in the Introduction, the only gauge invariant measure of how much the translations are broken is proportional to the graviton mass parameter.11 In the absence of gravity the order parameter is given by the energy density of the material (or the elastic moduli for solids [3]). In the presence of gravity the order parameter gives rise to the plasma mass terms for the graviton, see [10]. From the equations of motion (3.11), (3.12), and (3.13) we see that the mass of the dynamical elds is controlled by the quantity M 2(z) m~ 2VX = m2u2 h 1 + 5 l that is a function, dependent on the radial coordinate. In particular, one can de ne the ultraviolet (UV) and infrared (IR) masses as the values of M 2(z) at z = 0 and z = 1 respectively: Their ratio, m2UV = m2u2h ; mI2R = m2u2 is clearly controlled solely by the parameter in the bulk Lagrangian. In the CFT language, the fact that the natural order parameter for the translational symmetry breaking in the bulk is a function of the radial coordinate translates as the fact that the translations are broken by di erent amounts at di erent energy scales. The e ect 11As shown in [9, 10, 31] the model (3.1) and background solutions (3.3), (3.4) can be reformulated in an equivalent way by working in the unitary gauge, which makes no reference to scalar elds getting an expectation value. Therefore the nave identi cation of the scalar elds gradient as the order parameter seems a gauge-dependent statement. In any gauge, however, the net physical e ect is to provide a mass term for the physical spin 2 and shear modes in the gravity side. could be similar to a chain of di erent `super-lattices' on top of underlying `sub-lattices', for instance. For the practical purposes of this work we will not enter into whether this is a correct interpretation. Instead we simply note that for fairly general choices of V (X) one can arrange a simpler situation where the dynamics are controlled only by two relevant scales. For V (X) such that M (z) is approximately constant during most of the ow and then suddenly grows towards the IR, the relevant scales are the asymptotic values of M (z) | the IR and UV masses mIR and mUV . This identi cation is of course imperfect, but by taking large enough and N in (1.3) we can get su ciently close to it. The bottom line of this discussion is that for these models it is natural to expect that there are two di erent sources of the translational symmetry breaking, controlled by two di erent mass scales. When this is the case, we are under the conditions of having light gapped phonons provided that the ratio mUV =mIR is small enough. does indeed scale as pmUV =mIR. Let us argue now how large should the phonon mass gap be in relation to the parameter controlling the amount of the explicit versus spontaneous breaking, mUV =mIR. We recall that in the case of the chiral symmetry breaking in QCD the Goldstone boson mass squared is proportional to the quark masses (the source of the explicit breaking) m2 mq. By analogy we expect that the phonon mass, !0, should be proportional to the square-root of the explicit breaking source. The numerical results presented in section 4 provide a strong indication that this intuition is correct. On the right panel of gure 3 we see that the ratio !0= j!1j=j!2j quantifying the separation between the two lowest quasi-normal modes 4 Transverse quasi-normal modes In this section we compute the quasi-normal modes (QNMs) of the transverse sector (3.11) and (3.12) to show the e ect of on the spectrum, to identify the phonon-like excitations and to check the expected relationship (2.6) between the velocity of the transverse phonon and the elasticity. A detailed discussion on how to obtain these modes and on the numerics can be found in appendix B. In the presence of the model has two independent dimensionless parameters. We have used the scaling symetries of the equations to set l = uh = 1 and therefore consider T =m and as the independent parameters describing our system.12 4.1 The quasi-normal mode spectrum We rst discuss the QNMs at zero momentum and compare them to the known behaviour for = 0 (see [ 17 ]). The spectrum of the lowest QNMs at zero momentum as a function of T =m for = 0 and = 6013 is shown in gure 2. In both cases we see that at high enough temperatures T =m (red) there is a single purely imaginary mode whose absolute value 12The scaling symmetries of the equations show that in terms of the CFT quantities the parameters m=T and u8h are scaling invariants. For convenience, due to the small temperatures considered, we plot against in the gures. In all our computations we have set uh = 1, therefore when we refer to 1 it should be understood as 8 uh 1 or =s4 13The results at this particular value of appearing at other values are discussed in the text. are rather generic for any nite value of . Possible di erences -5 -10 -15 -20 -5 -10 -15 -20 -20 -10 10 20 30 and T =m 2 [0:10; 1:07] (blue-red). Right: = 60 and T =m 2 [0:16; 68:97] (blue-red). is small compared to that of the next lowest mode. By decreasing T =m below a critical value the two lowest purely imaginary modes collide developing a non-zero real part in the frequency (henceforth we refer to this as the collision point ), in agreement with [ 17 ] for = 0.14 Let us examine the e ects of a non-vanishing . The evolution of the QNMs at zero momentum with increasing at xed temperature T =m = 0:001 is depicted in the left panel of gure 3. We see that by increasing the value of the real part of the lowest mode increases while the absolute value of its imaginary part decreases. This implies that for increasing values of the lowest QNM becomes less and less damped thus indicating the presence of a long lived excitation. This is in fact already evident by comparing the two panels in gure 2. In particular, for = 0 the imaginary part at the collision point is !=T 10i. Hence, at the moment when the frequency of the lowest quasinormal mode develops a non-zero real part its decay rate T , so that there is no long-lived excitation in the spectrum. On the other hand, for = 60 we nd !=T 0:1i corresponding to a parametrically much smaller decay rate. Another important e ect due to increasing of is that the rest of the QNM spectrum moves away from the origin, thus generating a stronger hierarchy between the modes. Hence, we see that for large values of the momentum transport in the transverse sector is governed by two long lived gapped modes, well-separated from the rest of the QNM tower. We identify these with the pseudo-Goldstone bosons due to the spontaneous breaking of translational invariance. To better illustrate this point, we can speci cally look at the separation of scales between the rst and second QNMs. In the right panel of gure 3 we plot the ratio between the absolute values of the frequencies of the rst and second QNMs, as well as the ratio of the graviton mass terms in the UV and in the IR, as de ned in (3.15). Even if the two ratios are not in a perfect proportion, we can see a clear correlation in that they both decrease mono14We note that our convention for m2 di ers from that of [ 17 ] as follows: mt2here = 2m2here. Im[ω]/T -10000 -5000 5000 10000 Re[ω]/T -2000 -4000 -6000 -8000 similar behaviour is found also for other values of T =m. Right: comparison between the ratio of the absolute values of the frequencies of the rst and the second lowest QNMs shown on the left panel (blue) and the mass ratio pmUV =mIR (yellow). tonically in . This justi es that we can associate the origin of the phonon mass gap with the presence of a small but nonzero mUV (as compared to mIR), as argued in section 3.2 Finally, we study the e ect of on the lowest QNMs at nite momentum. In gure 4 we show the real and imaginary parts of the frequencies of the two lowest QNMs as a function of momentum at = 20. The di erent colors in gure 4 correspond to di erent values of T =m, changing from blue to red as the temperature increases. Let us rst discuss the behaviour of the real part distinguishing three di erent energy regimes. At high enough momentum with k=T 1, all the di erent temperature modes converge to the same slope, as one would expect in this energy regime. On the other hand, at very low energies, for k=T 1, we nd three qualitatively di erent behaviors, depending on the temperature. First, there exists a temperature T =m at which the mode shows no gap in the real part; this is the collision point (orange line). At this temperature the dispersion relation of the lowest mode looks like Re[!] c k. Let us remark that the temperature of the collision point increases with . Above this temperature the mode is purely di usive (red line) while below it the mode is gapped. In the latter regime the dispersion relation takes the form Re[!] a + v k2 (green and blue lines). These two regimes | the far UV and the deep IR | show a behaviour qualitatively similar to the case = 0, shown in gure 5, see [70{72] for further discussions on this behavior in holography and [73] in kinetic theory. The e ect of in these regimes is only quantitative: it alters the speci c values of k=T at which the qualitative behavior of the modes change, as well as the speci c values of !=T . It is in the intermediate energy regime where 6= 0 induces a qualitatively new behavior, as can be seen in gure 4. In particular, for some nite values of the real part of the modes decreases with increasing nite momenta, showing a dispersion relation that resembles that = 20 for T =m 2 [2:89; 68:27] (blue-red). 5 10 15 k/T -5 -10 -15 -20 0 -5 -10 -15 Im[ω]/T 5 10 15 k/T 2 4 6 8 10 12k/T 2 4 6 8 10 12k/T HJEP01(28)9 = 0 for T =m 2 [0:14; 0:16] (blue-red). of rotons (green line). Eventually, for high enough temperatures the two modes recollapse and become purely imaginary again at (yellow, orange and red lines in gure 4). This phenomenon is absent for = 0 as shown in gure 5. The imaginary part of the frequency behaves \complementarily" to the real part in the following sense. When the modes are propagating, i.e. the real parts are non-zero, the imaginary parts are identical and both modes have the same decay rate. On the contrary, when the modes become purely imaginary, they decouple and show di erent decay rates. Similar behavior has been previously found in holography [ 74, 75 ]. 4.2 As explained in previous sections, we argue that at least in some regime of the parameter space the lowest collective excitations behave as transverse gapped phonons. In order to support this claim we compare the velocity of the lowest QNM to the standard expression for the transverse sound speed in solid materials given in (2.6). The shear modulus in (2.6) can be obtained [10, 11] as the real part of the retarded Green's function of the transverse traceless metric perturbations at zero momentum and frequency (3.13). This leads to the following expression for the sound speed of the transverse excitations: c2T = 3 1 2L " + P z;!~!0 lim Re 1 B~i000(z) 6 B~i(z) l uh 3 ; (4.1) HJEP01(28)9 where the energy density " is given by the black brane energy density (3.7) and P = "=2 is the hydrodynamic pressure appearing in the stress-energy tensor at the boundary. We compute this by solving (3.13) numerically for zero momentum. An analytic expression valid in the m=T ! 0 limit was previously obtained in [11]. On the other hand, the velocity of the collective excitations can be obtained from the real part of the frequency of the lowest QNM at nite, low momentum k=T mentioned in the previous section, at temperatures below the collision point and for small momenta, the real part of the lowest mode can be tted to ! = p a + v2 k2 ; (Re[!])2 = a1 + a2 ka3 ; with a = a(T =m; ) vanishing at the collision point and giving rise to a linear behavior. To be more precise, we have tted the data to (4.2) (4.3) with a3 = 2 10 4 for the parameters investigated in this paper. In gure 6 we show the comparison between the sound speed determined from the quasinormal mode spectrum by the t (4.3) and the expression (4.1). On the left panel we show the comparison for xed = 20 as a function of temperature T =m. We see that the agreement between the speed of propagation extracted from the spectrum of QNMs and the velocity computed by the standard EFT expressions (2.6) and (4.1) is not exact and seemingly improves at low temperatures. This is however a coincidence and is not the case for other values of . The actual reason behind the disagreement is the presence of the small but nite explicit breaking | it is indeed particularly pronounced at small values of where the explicit breaking is large. This is con rmed by what we see in the right panel of gure 6. We further observe that as T =m ! 0 the sound speed approaches some maximal value and continuously decreases with increasing temperature. In other words, it shows a melting-like behaviour. At high temperatures the system is a uid in which there are no transverse sound waves and thus c2T = 0. At intermediate temperatures it is a viscoelastic material with nite viscosity and a non-zero shear modulus . In distinction from melting that takes place in ordinary solids, the transition to the uid phase is continuous in these materials. c 0.6 0.5 given in eq. (4.1) (blue line). Left: the velocity as a function of temperature for = 20. Right: the velocity as a function of for T =m = 0:05. (As a cross-check let us mention that the velocities evaluated at T =m = 0:001 coincide with the results shown here.) On the right panel of gure 6 we plot the velocity as a function of at xed T =m = 0:05. From here we see that the maximal speed of sound reached at low temperatures attens out in the large limit. Despite the general similarity, we do see a slight disagreement between the velocity obtained from the QNMs and the EFT expectations, persistent even in the regime of low temperatures and high values of . We believe that this is a combined e ect due to a non-zero mass gap and a non-zero decay rate of our collective excitations. Neither of these e ects is taken into account in (2.6). It would be interesting to gain a quantitative understanding of this deviation. We leave it for future work. Another interesting feature that gure 6 seems to suggest is that the sound speed might saturate in the large limit to some constant value that is close to c2T = 1=3. That this might be a universal feature for the transverse sound speed in conformal solids is also suggested in [76] and deserves further investigation. 5 Discussion In this work we have shown how the elastic properties of black holes are encoded in certain gravity theories that break translational invariance and therefore allow for momentum dissipation. We have shown that the standard relations from the elasticity theory are realized in these solutions to a very good accuracy, and that a clear identi cation of gapped but light transverse phonons is built in the framework. By the AdS/CFT correspondence, these black holes can be interpreted as strongly coupled theories that describe solid materials in a critical regime, with the phonon-like excitations being part of the dynamical spectrum of the theory. These elastic phenomena are realized in BHs within the class of holographic massive gravity theories introduced in [9]. Our main result is the following: in a certain (expected) parameter regime of these theories, the speed of propagation of the transverse phonons qualitatively coincides with the speed computed from the elastic rigidity modulus (as introduced in [10]), see gure 6. We also discuss how to understand the small mass gap of the transverse phonons from the symmetry breaking pattern, see the right panel of gure 3. These results give explicit con rmation that the lowest modes in the shar channel can be identi ed as transverse phonons and therefore that the planar black brane solutions of these massive gravity theories do behave as solids, as previously advocated in [9{11]. The clear relation between the phonon mass gap and the source of explicit breaking also backs up the proposal of [31] that these gravity duals can be used to model some aspects of disorder. Our results open up a number of questions. By construction our models do not allow to switch o completely the phonon mass gap in the low temperature regime. In particular, it would be interesting to nd a model realizing exactly gapless phonons. Our results suggest that the limit that could generate this is to take mUV ! 0 while keeping mIR nite. In this limit, the shape of the noncanonical kinetic terms for the scalar becomes apparently singular (i.e. V 0(X) ! 0 near the AdS boundary). We leave for future investigation whether this kind of limit is actually well de ned and whether it does give rise to gapless phonons. A natural continuation of our work would be to analyze the longitudinal sector of the excitations. This provides a further check of the emerging picture since the longitudinal sound speed also has to relate to the bulk and shear elastic moduli according to (1.2). Another interesting point for the future is to see whether there is a BH transition at high temperature corresponding to its `melting' into a uid. We already see some evidence for that due to the transverse phonon becoming eventually over-damped and showing a di usive behaviour at high enough temperatures. Moreover, by increasing T =m both the transverse speed of sound and the elastic modulus continuously decrease and tend to the uid limit (cT = = 0). This may be linked to the fact that these BHs have simultaneously elasticity and viscosity (i.e. viscoelasticity) [11]. See also [77, 78] for other related work in holography. One could also add a non-zero charge density to our setup and study in more detail the proposal of [59] in connection with the transport properties of bad metals. The optical conductivity for these models has already been computed in [9]; it exhibits a pinned response showing the presence of both explicit and spontaneous breaking. As a further con rmation of the present picture, the DC conductivity shows a metal-insulator transition which is typical of pinned systems (e.g. the pinned charge density waves). Additionally, it would be important to check for signatures of the presence of the phonons in the DC transport properties and thermodynamical observables, as attempted in [79]. A proper and complete hydrodynamic analysis of these holographic systems with translational symmetry breaking is still lacking. It would be extremely useful to build such a framework in order to study in a more rigorous and systematic way the elastic moduli and the viscosities in terms of Kubo formulas. This sheds light on the several still present puzzles related to those physical observables. Let us end by emphasizing one important aspect of our results. E ective Field Theory description of Solids [ 2, 14, 16 ] is a powerful way to describe phonons in solid materials that seems to be completely orthogonal to the holographic models used here. In fact, we have used quite a lot of the intuition obtained from this EFT. A natural question then arises: is there any solid EFT that describes the properties of our elastic black holes? Our results are very encouraging although this is not obvious a priori. The reason is simple: in order for the EFT description to be a good approximation, there needs to be a separation of scales. In the standard (weakly coupled) solids a large separation of scales does indeed exist between the lattice spacing and the low energy phonon modes. In the holographic solids (the CFT duals of the AdS BHs studied here), however, there is no separation of scales because by construction the eld theory is scale-invariant, i.e. there is a continuum of CFT modes at T = 0. Our results imply that it is su cient to have a separation of scales in the QNM spectum | i.e. that there is one QNM much lighter than the remaining ones. Hence, EFT-like relations may still hold even in conformal eld theories. We nd this an interesting output of our work. We plan to return to these issues in the near future. While this article was under the publication review process, a new work by the same authors [ 80 ] appeared. This work con rms the picture proposed and concludes that it is indeed possible to realize exactly gapless transverse phonons as suggested above. Acknowledgments We thank Andrea Amoretti, Tomas Andrade, Aron Beekman, Angelo Esposito, Daniel Fernandez, Blaise Gouteraux, Sean Hartnoll, Matti Jarvinen, Keun-Young Kim, Elias Kiritsis, Alexander Krikun, Julian Leiber, Weijia Li, Yi Ling, Matthew Lippert, Andy Lucas, Nicodemo Magnoli, Rene Meyer, Daniele Musso, Philip Philips, Napat Poovuttikul, SangJin Sin, Saso Grozdanov and Jan Zaanen for useful discussions and comments about this work and the topics considered. We are particularly grateful to Matthew Lippert, Niko Jokela, Marti Jarvinen, Tomas Andrade, Alexander Krikun and Napat Poovuttikul for sharing with us informations about their unpublished works. OP acknowledges support by the Spanish Ministry MEC under grant FPA2014-55613-P and the Severo Ochoa excellence program of MINECO (grant SO-2012-0234, SEV-20160588), as well as by the Generalitat de Catalunya under grant 2014-SGR-1450. MB is supported in part by the Advanced ERC grant SM-grav, No 669288. MB would like to thank the Asia Paci c Center for Theoretical Physics, Hanyang University and Gwangju Institute of Science and Technology for the hospitality during the completion of this work. MB would also like to thank the Asia Paci c Center for Theoretical Physics for the large quantity of co ee needed to complete this manuscript. AJ acknowledges nancial support by Deutsche Forschungsgemeinschaft (DFG) GRK 1523/2. Solid EFT and elasticity In this section we derive the expressions (2.6) for the transverse and longitudinal sound speeds in an elastic medium with non-negligible pressure. We use the standard EFT methodology for describing solid materials [ 1, 2, 14, 16, 60 ] in at space-time. In 3 + 1 dimensions, the low-energy behavior of solids can be described by using three scalar elds I (xi; t), corresponding to its comoving coordinates. The e ective eld theory of the long wavelength excitations in a solid is then given by the scalar eld action Z S = d3+1x F (X; Y; Z) (A.1) I J : (A.2) The indices I; J = fx; y; zg are contracted with IJ . In this section the spacetime metric is set to the at Minkowski metric = diag( 1; 1; 1; 1). Let us consider an isotropic and homogeneous background eld con guration ^I = xi iI ; 2 3 corresponding to the equilibrium con guration of the solid. On this background, the full stress tensor of the scalars takes a perfect uid form T = diag( "; p; p; p) with energy density and pressure identi ed as " = F ; p = F + (XFX + 2Y FY + 3ZFZ ) ; all quantities evaluated on the background (A.3). It is then easy to see how the functional form of F (X; Y; Z) relates to the linear elastic response parameters and de ned in (2.3). These coe cients are a measure of the strength with which the solid opposes a certain deformation, described in terms of the displacement vector ui. In particular, the stress tensor due to a traceless shear deformation is given in terms of the shear modulus in eq. (2.4). In the EFT setting, the equilibrium con guration of the solid is the one given in (A.3). We can introduce an analogue of the mechanical deformations by taking I = ^I + I : equilibrium value I^IJ = scalar eld perturbations I as Under this deformation the matrix IIJ transforms as IIJ ! I^IJ + @I J + @J I with the IJ . This provides a natural identi cation of the displacement tensor uij , given in (2.2), in terms of the symmetric combination of the derivatives of the It is clear now that this deformation will a ect the energy density and the full stress tensor of the system. The notion of the shear modulus is therefore easy to identify in the EFT: a I IJ I ^IJ : small transverse traceless shear deformation uITJ induces a small non-zero transverse traceless component of the I stress tensor. This is the internal stress generated by such a strain, and is the one that has to be applied by an external action in order to sustain the given shear deformation constant in time. One can then obtain easily the elastic response by calculating from the full stress-energy tensor of the scalar elds. The resulting expression can be put T Tij = T ij T T^ij T T Tij = I ij ^ij I T = 2 u ij T 2 3 = (X FX + 4Y FY ) : (A.7) (A.8) (A.9) (A.10) (A.11) A non-trivial cross-check that this is an appropriate description of a solid is to reproduce the linear dispersion relations for the phonons, ! = cL; T k, with the known relations between the elastic moduli and the propagation speed of the transverse and longitudinal sound waves [3, 12]: c 2L = + 43 In a relativistic theory, one can expect that the denominator in (A.10) is not just given by the energy density ". We can easily derive the relativistic version of these expressions by perturbing the background con guration (A.3) with a spacetime dependent small perturbations I de ned in (A.5). They trivially split into transverse and longitudinal phonon sectors, the transverse one obeying @i i = 0. A straightforward calculation gives for the transverse sound speed c2T = X FX + 4 Y FY X FX + 2 Y FY + 3 Z FZ in the form15 with The expression for the longitudinal sound speed reproduces c2L = ( + 43 )=(" + p). These relations match with the well-known results for solids given in (2.6) and thus represents a zeroth order cross-check that the actions of the type (A.1) do really describe solids. In addition, we see that in the denominators the factor " + p emerges, as expected for relativistic systems. Let us also emphasize that the fact that we were able to reproduce the correct formulas for the speeds from the simple solid e ective Lagrangian means that these relations formally hold only at zero temperature. In other words, they are valid for the single-particle phonon excitations around the QFT vacuum state of the I scalar elds. This is, of course, expected because the linear elastic response is reversible and non-dissipative, and therefore it must be derivable from a zero-temperature Lagrangian. Thermal and other dissipative e ects that a ect the phonon dispersion relations can be incorporated in the EFT description, but this is not the topic of the present paper. Let us keep in mind, though, that in the holographic realizations presented in the main text, the dissipative e ects are indeed present. 15At the level of perturbations one can ignore the di erence between I and i indices. 10-3 10-6 nχ10-9 10-12 10-5 10-8 n χ10-11 10-14 10-5 10-11 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 n n n Numerical details of QNM computation In this section we present some details regarding the numerics of the quasi-normal mode computation presented in the main text. To determine the quasi-normal modes which correspond to poles of the corresponding retarded Green's function in the dual eld theory, we have to solve (3.11) and (3.12) subject to in-falling boundary conditions at the horizon. Moreover, by de nition, the non-normalizable modes of the elds Bi and Ui have to vanish. Due to numerical e ciency we use instead directly the uctuations hij and 'I . We perform a Fourier transformation for the uctuation elds with respect to time and the spatial eld theory directions. Moreover, along the radial direction the uctuation elds are represented by Chebyshev polynomials. Without loss of generality, the x-component of the momentum k can be set to zero due to the SO(2) rotational symmetry of the background in the (x; y)-plane. The relevant transverse coupled uctuations are hxu; hxy; htx and '1: In the following we choose a gauge such that hxu = 0, hence giving rise to a constraint equation. The in-falling horizon conditions are automatically implemented by using in-going Eddington-Finkelstein coordinates. Moreover, by rede ning the elds hxy; htx and '1 by hxy(z) = z h~xy(z); htx(z) = z h~tx(z) and '1(z) = z3'~1(z); respectively, and solving for h~xy(z); h~tx(z) and '~1(z), we ensure that the non-normalizable modes vanish since only regular solutions can be represented on a Chebyshev grid. The problem of nding the quasi-normal modes and their associated solutions can be reformulated in terms of a generalized eigenvalue / eigenvector problem.16 In particular, we solve for the three second-order equation of motion of h~xy; h~tx and '~1 and check afterwards whether the constraint equation is satis ed. In particular, all QNM data in this project are calculated with at least 50 or 70 grid points and 50 digit precision. We checked the convergence of the quasi-normal modes as well as how good the equations of motion and the constraint are satis ed. 16Note that despite using Eddington-Finkelstein coordinates, there are still !2 terms appearing in the equations of motion for the uctuations. However, by introducing auxiliary elds we reduce the generalised eigenvalue problem to a linear one. For more details see [81]. In gure 7 we present convergence plots for the lowest QNM for the following parameters: = 60; T =m = 0:16 and k=T = 0:14. For this computation we used 70 grid points. Moreover, for this case we evaluated the equations of motion and the constraint on the solution using 100 regularly spaced points, nding the square norm to be 10 7 or less for each of the equations of motions as well as for the constraint. Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. [hep-ph/9311264] [INSPIRE]. University Press, Cambridge U.K., (1995). [6] D. Vegh, Holography without translational symmetry, arXiv:1301.0537 [INSPIRE]. [7] M. Blake and D. Tong, Universal resistivity from holographic massive gravity, Phys. Rev. D 88 (2013) 106004 [arXiv:1308.4970] [INSPIRE]. (2014) 101 [arXiv:1311.5157] [INSPIRE]. [8] T. Andrade and B. Withers, A simple holographic model of momentum relaxation, JHEP 05 [9] M. Baggioli and O. Pujolas, Electron-phonon interactions, metal-insulator transitions and holographic massive gravity, Phys. Rev. Lett. 114 (2015) 251602 [arXiv:1411.1003] [10] L. Alberte, M. Baggioli, A. Khmelnitsky and O. Pujolas, Solid holography and massive gravity, JHEP 02 (2016) 114 [arXiv:1510.09089] [INSPIRE]. [11] L. Alberte, M. Baggioli and O. Pujolas, Viscosity bound violation in holographic solids and the viscoelastic response, JHEP 07 (2016) 074 [arXiv:1601.03384] [INSPIRE]. [12] L.D. Landau and E.M. Lifshitz, Course of theoretical physics, volume 7, theory of elasticity, Pergamon Press, Oxford U.K., (1970). [13] L.V. Delacretaz, B. Gouteraux, S.A. Hartnoll and A. Karlsson, Theory of hydrodynamic transport in uctuating electronic charge density wave states, Phys. Rev. B 96 (2017) 195128 [arXiv:1702.05104] [INSPIRE]. [14] S. Dubovsky, L. Hui, A. Nicolis and D.T. Son, E ective eld theory for hydrodynamics: thermodynamics and the derivative expansion, Phys. Rev. D 85 (2012) 085029 [arXiv:1107.0731] [INSPIRE]. [15] S. Endlich, A. Nicolis and J. Wang, Solid in ation, JCAP 10 (2013) 011 [arXiv:1210.0569] uids, super uids, solids and supersolids and incoherent transport, JHEP 01 (2015) 039 [arXiv:1411.1062] [INSPIRE]. JHEP 07 (2017) 055 [arXiv:1705.01766] [INSPIRE]. HJEP01(28)9 [arXiv:1405.3651] [INSPIRE]. critical metals, JHEP 10 (2015) 112 [arXiv:1507.07137] [INSPIRE]. [23] S. Grozdanov, A. Lucas and K. Schalm, Incoherent thermal transport from dirty black holes, Phys. Rev. D 93 (2016) 061901 [arXiv:1511.05970] [INSPIRE]. [arXiv:1604.01754] [INSPIRE]. [INSPIRE]. [arXiv:1704.00947] [INSPIRE]. [25] M. Baggioli, B. Gouteraux, E. Kiritsis and W.-J. Li, Higher derivative corrections to incoherent metallic transport in holography, JHEP 03 (2017) 170 [arXiv:1612.05500] [26] K.-Y. Kim and C. Niu, Di usion and butter y velocity at nite density, JHEP 06 (2017) 030 (2015) 071 [arXiv:1501.05656] [INSPIRE]. [arXiv:1505.05092] [INSPIRE]. [27] M. Blake, R.A. Davison and S. Sachdev, Thermal di usivity and chaos in metals without quasiparticles, Phys. Rev. D 96 (2017) 106008 [arXiv:1705.07896] [INSPIRE]. [28] A. Lucas, Conductivity of a strange metal: from holography to memory functions, JHEP 03 [29] R.A. Davison and B. Gouteraux, Dissecting holographic conductivities, JHEP 09 (2015) 090 [30] A. Lucas, Hydrodynamic transport in strongly coupled disordered quantum eld theories, New J. Phys. 17 (2015) 113007 [arXiv:1506.02662] [INSPIRE]. [31] M. Baggioli and O. Pujolas, On holographic disorder-driven metal-insulator transitions, JHEP 01 (2017) 040 [arXiv:1601.07897] [INSPIRE]. 086003 [arXiv:1306.5792] [INSPIRE]. [32] R.A. Davison, Momentum relaxation in holographic massive gravity, Phys. Rev. D 88 (2013) [33] M. Blake, D. Tong and D. Vegh, Holographic lattices give the graviton an e ective mass, Phys. Rev. Lett. 112 (2014) 071602 [arXiv:1310.3832] [INSPIRE]. [34] R.A. Davison, K. Schalm and J. Zaanen, Holographic duality and the resistivity of strange metals, Phys. Rev. B 89 (2014) 245116 [arXiv:1311.2451] [INSPIRE]. [arXiv:1311.3292] [INSPIRE]. 181 [arXiv:1401.5436] [INSPIRE]. 007 [arXiv:1401.5077] [INSPIRE]. [arXiv:1406.4134] [INSPIRE]. [39] A. Donos, B. Gouteraux and E. Kiritsis, Holographic metals and insulators with helical HJEP01(28)9 symmetry, JHEP 09 (2014) 038 [arXiv:1406.6351] [INSPIRE]. JHEP 07 (2012) 168 [arXiv:1204.0519] [INSPIRE]. [40] G.T. Horowitz, J.E. Santos and D. Tong, Optical conductivity with holographic lattices, [41] M. Baggioli, Gravity, holography and applications to condensed matter, arXiv:1610.02681 [INSPIRE]. [arXiv:1604.08915] [INSPIRE]. dyons, JHEP 06 (2016) 113 [arXiv:1603.03029] [INSPIRE]. Holographic type II Goldstone bosons, JHEP 07 (2013) 108 [arXiv:1302.5641] [INSPIRE]. [45] R. Argurio, A. Marzolla, A. Mezzalira and D. Naegels, Note on holographic nonrelativistic Goldstone bosons, Phys. Rev. D 92 (2015) 066009 [arXiv:1507.00211] [INSPIRE]. [46] A. Esposito, S. Garcia-Saenz and R. Penco, First sound in holographic super uids at zero temperature, JHEP 12 (2016) 136 [arXiv:1606.03104] [INSPIRE]. [47] R. Argurio, A. Marzolla, A. Mezzalira and D. Musso, Analytic pseudo-Goldstone bosons, JHEP 03 (2016) 012 [arXiv:1512.03750] [INSPIRE]. [48] R. Argurio, G. Giribet, A. Marzolla, D. Naegels and J.A. Sierra-Garcia, Holographic Ward identities for symmetry breaking in two dimensions, JHEP 04 (2017) 007 [arXiv:1612.00771] [INSPIRE]. [49] A. Amoretti, D. Arean, R. Argurio, D. Musso and L.A. Pando Zayas, A holographic perspective on phonons and pseudo-phonons, JHEP 05 (2017) 051 [arXiv:1611.09344] [INSPIRE]. [INSPIRE]. [50] S. Nakamura, H. Ooguri and C.-S. Park, Gravity dual of spatially modulated phase, Phys. Rev. D 81 (2010) 044018 [arXiv:0911.0679] [INSPIRE]. Phys. Rev. D 82 (2010) 126001 [arXiv:1007.3737] [INSPIRE]. [51] H. Ooguri and C.-S. Park, Holographic end-point of spatially modulated phase transition, [52] A. Aperis, P. Kotetes, E. Papantonopoulos, G. Siopsis, P. Skamagoulis and G. Varelogiannis, Holographic charge density waves, Phys. Lett. B 702 (2011) 181 [arXiv:1009.6179] [53] H. Ooguri and C.-S. Park, Spatially modulated phase in holographic quark-gluon plasma, Phys. Rev. Lett. 106 (2011) 061601 [arXiv:1011.4144] [INSPIRE]. eld theories with chiral HJEP01(28)9 anomaly from holography, JHEP 03 (2016) 164 [arXiv:1601.02125] [INSPIRE]. [59] L.V. Delacretaz, B. Gouteraux, S.A. Hartnoll and A. Karlsson, Bad metals from uctuating density waves, SciPost Phys. 3 (2017) 025 [arXiv:1612.04381] [INSPIRE]. ordinary stu , extra-ordinary stu , JHEP 06 (2015) 155 [arXiv:1501.03845] [INSPIRE]. 96 (2017) 106017 [arXiv:1708.07837] [INSPIRE]. holographic spontaneous helices, arXiv:1708.08306 [INSPIRE]. [63] L.P. Kadano and P.C. Martin, Hydrodynamic equations and correlation functions, Ann. Phys. 24 (1963) 419. [hep-th/0002230] [INSPIRE]. [64] S. de Haro, S.N. Solodukhin and K. Skenderis, Holographic reconstruction of space-time and renormalization in the AdS/CFT correspondence, Commun. Math. Phys. 217 (2001) 595 [65] S.A. Hartnoll, Lectures on holographic methods for condensed matter physics, Class. Quant. Grav. 26 (2009) 224002 [arXiv:0903.3246] [INSPIRE]. [66] S.A. Hartnoll, D.M. Ramirez and J.E. Santos, Entropy production, viscosity bounds and bumpy black holes, JHEP 03 (2016) 170 [arXiv:1601.02757] [INSPIRE]. [67] P. Burikham and N. Poovuttikul, Shear viscosity in holography and e ective theory of transport without translational symmetry, Phys. Rev. D 94 (2016) 106001 [arXiv:1601.04624] [INSPIRE]. [68] G. Policastro, D.T. Son and A.O. Starinets, The shear viscosity of strongly coupled N = 4 supersymmetric Yang-Mills plasma, Phys. Rev. Lett. 87 (2001) 081601 [hep-th/0104066] [69] M. Blake, Momentum relaxation from the uid/gravity correspondence, JHEP 09 (2015) 010 [INSPIRE]. [INSPIRE]. [arXiv:1505.06992] [INSPIRE]. arXiv:1707.08577 [INSPIRE]. [70] A. Jimenez-Alba, K. Landsteiner and L. Melgar, Anomalous magnetoresponse and the Stuckelberg axion in holography, Phys. Rev. D 90 (2014) 126004 [arXiv:1407.8162] [71] S. Grozdanov, N. Kaplis and A.O. Starinets, From strong to weak coupling in holographic models of thermalization, JHEP 07 (2016) 151 [arXiv:1605.02173] [INSPIRE]. [72] D.M. Hofman and N. Iqbal, Generalized global symmetries and holography, of strongly coupled matter, JHEP 06 (2016) 047 [arXiv:1603.05950] [INSPIRE]. JHEP 12 (2017) 113 [arXiv:1708.09391] [INSPIRE]. HJEP01(28)9 027 [arXiv:1309.0146] [INSPIRE]. arXiv:1711.03100 [INSPIRE]. magnetic black branes & chiral magnetic transport, JHEP 04 (2017) 067 [1] H. Leutwyler , Nonrelativistic e ective Lagrangians, Phys. Rev. D 49 ( 1994 ) 3033 [2] H. Leutwyler , Phonons as goldstone bosons , Helv. Phys. Acta 70 ( 1997 ) 275 [3] P.M. Chaikin and T.C. Lubensky , Principles of condensed matter physics, Cambridge [4] G. Gruner, The dynamics of spin-density waves , Rev. Mod. Phys . 66 ( 1994 ) 1 [INSPIRE]. [5] J.M. Maldacena , The large-N limit of superconformal eld theories and supergravity , Int. J. [16] A. Nicolis , R. Penco and R.A. Rosen , Relativistic from a coset construction , Phys. Rev. D 89 ( 2014 ) 045002 [arXiv: 1307 .0517] [INSPIRE]. [17] R.A. Davison and B. Gouteraux , Momentum dissipation and e ective theories of coherent [18] K.-Y. Kim , K.K. Kim , Y. Seo and S.-J. Sin , Coherent/incoherent metal transition in a holographic model , JHEP 12 ( 2014 ) 170 [arXiv: 1409 .8346] [INSPIRE]. [19] M. Baggioli and W.-J. Li , Di usivities bounds and chaos in holographic Horndeski theories , [20] S.A. Hartnoll , Theory of universal incoherent metallic transport , Nature Phys . 11 ( 2015 ) 54 [21] S. Grozdanov , A. Lucas , S. Sachdev and K. Schalm , Absence of disorder-driven metal-insulator transitions in simple holographic models , Phys. Rev. Lett . 115 ( 2015 ) 221601 [22] R.A. Davison , B. Gouteraux and S.A. Hartnoll , Incoherent transport in clean quantum [24] M. Blake , Universal di usion in incoherent black holes , Phys. Rev. D 94 ( 2016 ) 086014 [35] A. Donos and J.P. Gauntlett , Holographic Q-lattices , JHEP 04 ( 2014 ) 040 [36] B. Gouteraux , Charge transport in holography with momentum dissipation , JHEP 04 ( 2014 ) [37] A. Donos and J.P. Gauntlett , Novel metals and insulators from holography , JHEP 06 ( 2014 ) [38] A. Amoretti , A. Braggio , N. Maggiore , N. Magnoli and D. Musso , Thermo-electric transport in gauge/gravity models with momentum dissipation , JHEP 09 ( 2014 ) 160 [42] M. Baggioli and O. Pujolas , On e ective holographic Mott insulators , JHEP 12 ( 2016 ) 107 [43] A. Amoretti , M. Baggioli , N. Magnoli and D. Musso , Chasing the cuprates with dilatonic [44] I . Amado , D. Arean , A. Jimenez-Alba , K. Landsteiner , L. Melgar and I.S. Landea , [54] A. Donos and J.P. Gauntlett , Holographic striped phases , JHEP 08 ( 2011 ) 140 [55] O. Bergman , N. Jokela , G. Lifschytz and M. Lippert , Striped instability of a holographic Fermi-like liquid , JHEP 10 ( 2011 ) 034 [arXiv: 1106 .3883] [INSPIRE]. [56] A. Donos , J.P. Gauntlett and C. Pantelidou , Spatially modulated instabilities of magnetic black branes , JHEP 01 ( 2012 ) 061 [arXiv: 1109 .0471] [INSPIRE]. [57] A. Donos and J.P. Gauntlett , Black holes dual to helical current phases , Phys. Rev. D 86 [58] M. Ammon , J. Leiber and R.P. Macedo , Phase diagram of 4D [60] A. Nicolis , R. Penco , F. Piazza and R. Rattazzi , Zoology of condensed matter: framids, [61] N. Jokela , M. Jarvinen and M. Lippert , Pinning of holographic sliding stripes , Phys. Rev . D [62] T. Andrade , M. Baggioli , A. Krikun and N. Poovuttikul , Pinning of longitudinal phonons in [73] M. Stephanov , H.-U. Yee and Y. Yin , Collective modes of chiral kinetic theory in a magnetic eld , Phys. Rev. D 91 ( 2015 ) 125014 [arXiv: 1501 .00222] [INSPIRE]. [74] U. Gursoy, S. Lin and E. Shuryak , Instabilities near the QCD phase transition in the holographic models , Phys. Rev. D 88 ( 2013 ) 105021 [arXiv: 1309 .0789] [INSPIRE]. [75] R.A. Janik , J. Jankowski and H. Soltanpanahi , Quasinormal modes and the phase structure [76] A. Esposito , S. Garcia-Saenz , A. Nicolis and R. Penco , Conformal solids and holography, [77] S. Kachru , A. Karch and S. Yaida , Holographic lattices, dimers and glasses , Phys. Rev. D 81 [78] D. Anninos , T. Anous , F. Denef and L. Peeters , Holographic vitri cation , JHEP 04 ( 2015 ) [79] M. Baggioli and D.K. Brattan , Drag phenomena from holographic massive gravity , Class. Quant . Grav. 34 ( 2017 ) 015008 [arXiv: 1504 .07635] [INSPIRE]. [80] L. Alberte , M. Ammon , M. Baggioli , A. Jimenez-Alba and O. Pujolas , Holographic phonons, [81] M. Ammon , M. Kaminski , R. Koirala , J. Leiber and J. Wu , Quasinormal modes of charged

This is a preview of a remote PDF:

Lasma Alberte, Martin Ammon, Matteo Baggioli, Amadeo Jiménez, Oriol Pujolàs. Black hole elasticity and gapped transverse phonons in holography, Journal of High Energy Physics, 2018, 129, DOI: 10.1007/JHEP01(2018)129