YITP-20-111

Multi-field dark energy: cosmic acceleration on a steep potential

Yashar Akrami,1, 2, ∗ Misao Sasaki,3, 4, 5, † Adam R. Solomon,6, ‡ and Valeri Vardanyan3, § 1Laboratoire de Physique de l’Ecole´ Normale Sup´erieure, ENS, Universit´ePSL, CNRS, Sorbonne Universit´e,Universit´ede Paris, F-75005 Paris, France 2Observatoire de Paris, Universit´ePSL, Sorbonne Universit´e,LERMA, 75014 Paris, France 3Kavli Institute for the Physics and Mathematics of the Universe (WPI), UTIAS, The University of Tokyo, Chiba 277-8583, Japan 4CGP, Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan 5LeCosPA, National Taiwan University, Taipei 10617, Taiwan 6Department of Physics & McWilliams Center for Cosmology, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA We argue that dark energy with multiple fields is theoretically well-motivated and predicts distinct observational signatures, in particular when cosmic acceleration takes place along a trajectory that is highly non-geodesic in field space. Such models provide novel physics compared to ΛCDM and quintessence by allowing cosmic acceleration on steep potentials. From the theoretical point of view, these theories can easily satisfy the conjectured swampland constraints and may in certain cases be technically natural, potential problems which are endemic to standard single-field dark energy. Observationally, we argue that while such multi-field models are likely to be largely indistinguishable from the concordance cosmology at the background level, dark energy perturbations can cluster, leading to an enhanced growth of large-scale structure that may be testable as early as the next generation of cosmological surveys. Keywords: quintessence, multi-field dark energy, clustering dark energy, swampland, large-scale structure

Introduction. — Dark energy beyond the cosmologi- We will argue that there is novel and interesting cal standard model is usually studied in the context of physics in multi-field dark energy models which follow theories with a single scalar field, such as quintessence non-geodesic or curved trajectories in field space. As is [1] or scalar-tensor gravity [2]. While this is primarily well-known in the context of inflation, such “turning” tra- motivated by simplicity, physically-realistic models often jectories make accelerated expansion possible in regions include additional scalar degrees of freedom, especially if where the potential is too steep to otherwise support accel- viewed as low-energy effective theories arising from some erated expansion [25, 33]. This is in contrast to standard underlying ultraviolet (UV) completion. For example, single-field dynamics, which trivially follows a geodesic compactifications in string theory are characterized by in the one-dimensional “field space,” and hence has the multiple moduli fields, many of which are not necessarily usual slow-roll requirements. Allowing for this type of fully stabilized and may therefore play important roles in strongly multi-field behavior severs the link between a cosmic evolution at low energies [3, 4]. Further theoreti- flat potential and an equation of state near −1. In addi- cal motivation for considering dark energy with multiple tion to opening up an avenue to evade the (conjectured) fields comes from the recently-proposed swampland con- swampland constraints (see also Refs. [34, 35]), many jectures [5–8], parameter constraints which, it is claimed,1 of which place lower bounds on the slope of the poten- must be satisfied by any low-energy model which possesses tial, non-geodesic multi-field behavior is a novel physical a UV completion in string theory (or sometimes quan- mechanism for dark energy that leads to observable sig- tum gravity more generally). The swampland bounds natures, predominantly by suppressing the sound speed on single-field quintessence have been shown to be in of fluctuations. strong tension with existing cosmological data [10, 14]. These considerations strongly motivate phenomenological In this Letter, we propose curved trajectories in multi- arXiv:2008.13660v3 [astro-ph.CO] 10 Jun 2021 attention to multi-field theories, which are common in the field theories as a framework for building novel, theoret- inflationary literature,2 as theoretically-compelling dark ically well-motivated dark energy models with distinct energy candidates. phenomenological consequences. As a concrete (though non-exhaustive) example, we focus on “spinning” mod- els, in which the scalars rotate in field space with a nearly-constant speed. While the resultant cosmologi- 1 It is important to note that the status of these conjectures is cal background evolution is practically indistinguishable unresolved: see, e.g., Refs. [9–13] for counter-arguments. from ΛCDM (that is, the dark energy equation of state 2 The analogy with inflation will prove instructive throughout, is very close to −1), observable features in the evolution particularly because multi-field dynamics is much better studied in the inflationary context [15–25] than in dark energy, although of large-scale structure have the potential to distinguish see, e.g., Refs. [26–32] for notable examples of multi-field dark multi-field spinning dark energy from both ΛCDM and energy. single-field (or single-field-like) quintessence. 2

a From the observational perspective, these models are where Va ≡ ∂V/∂φ , H ≡ a/a˙ is the Hubble rate, over- an essential part of theory space, viewed in the context dots denote derivatives with respect to cosmic time t, and of interpreting existing and future cosmological data. For the field-space covariant time derivative Dt is defined by single-field models of dark energy, as well as multi-field a ˙ a a b ˙c models with shallow potentials, observing a constant dark DtA ≡ A + ΓbcA φ , (3) energy equation of state wDE would require the fields to a with Γbc the field-space Christoffel symbols. The Fried- be effectively non-dynamical. As we show in this Letter, mann equation is for multi-field models with non-geodesic trajectories it is possible to have w arbitrarily close to −1 even if the 1 DE 3M 2 H2 = φ˙2 + V + ρ + ρ , (4) fields are highly dynamical. This means that a Λ-like Pl 2 M R equation of state, if supported by the next generation of where we have defined cosmological surveys, would not necessarily imply that the ˙2 ˙a ˙b late-time cosmic acceleration is driven by a non-dynamical φ ≡ Gabφ φ , (5) cosmological constant. The novel physical effects of multi-field dark energy are which characterizes the speed along the background trajec- more pronounced at the level of perturbations. In order tory in field space, and denoted the matter and radiation for a single field to drive cosmic acceleration, the relevant energy densities by ρM and ρR, respectively. We will find it convenient to introduce the normalized mass scale typically must be of order H0, the present- day expansion rate. The associated Compton wavelength tangent and normal vectors to the field-space trajectory, is therefore around the size of the horizon, preventing φ˙a dark energy from clustering on observable (sub-horizon) T a ≡ , (6) ˙ scales. We will show that the models we consider here φ are a type of clustering dark energy [36–47]: the sound a 1 a N ≡ − DtT , (7) speed of dark energy fluctuations in these models is much Ω smaller than unity for a wide range of parameters, so the with sound horizon can be much smaller than the cosmological horizon, leading to clustering at observable (sub-horizon) Ω ≡ |DtT |. (8) scales.3 We therefore expect significant enhancements in clustering of large-scale structure at low redshifts. This Projecting the scalar equation of motion (2) along these feature provides a powerful method of testing this impor- directions, we find tant class of dark energy models against a cosmological ¨ ˙ φ + 3Hφ + VT = 0 , (9) constant, as well as more orthodox, slowly-rolling dark ˙ energy models. VN = φΩ , (10) Multi-field dark energy. — We are interested in dark where we have defined energy models with multiple scalar fields minimally cou- a a pled to gravity. At leading order in derivatives, we con- VT ≡ VaT ,VN ≡ VaN . (11) sider a standard σ-model setup, The novelty of multi-field dark energy hinges on the M 2 1 fact that the fields need not follow geodesic trajectories L = Pl R − G (φ)∂ φa∂µφb − V (φ) + L , (1) 2 2 ab µ m in field space. The degree of departure from a geodesic trajectory, or turning, is characterized by Ω ≡ |DtT |, as where Gab is the field-space metric, which is allowed to the geodesic equation is D T a = 0. In order to compute a t depend on the fields φ , V (φ) is the potential, and Lm is Ω along cosmological trajectories, we will need to express the matter Lagrangian. it in terms of the fields φa. For two-field systems, as we Restricting ourselves to cosmological solutions with a will consider in this Letter,√ we can write the normal as Friedmann-Lemaˆıtre-Robertson-Walker (FLRW) metric, a ab N = ε Tb, with ε12 = det G. It follows from eq. (10) and adopting the framework developed in Refs. [15, 16, 19], and the definition of VN that the scalar field equations of motion are 2  ˙ ˙  D φ˙a + 3Hφ˙a + V a = 0 , (2) 1 φ1V2 − φ2V1 t Ω2 = , (12) det G φ˙4

˙ ˙b where φa = Gabφ . For an arbitrary number of scalars, 3 By contrast, in order for a single field to drive cosmic acceleration, this generalizes to the relevant mass scale typically must be of order H0, the present- day expansion rate. The associated Compton wavelength is F F ab therefore around the size of the horizon, preventing single-field Ω2 = ab , (13) dark energy from clustering at smaller scales. 2φ˙4 3

2 2 where where ΩDE ≡ ρφ/ρtot ≈ V/(3MPlH ) is the usual dark energy density parameter. We see that the presence of F ≡ φ˙ V − φ˙ V . (14) ab a b b a additional matter fields can only suppress DE further. We conclude that multi-field dark energy, much like To see this, note that F = 2φ˙T V = 2V T N , where ab [a b] N [a b] inflation, can drive accelerated expansion on arbitrarily we have decomposed V = V T + V N .4 Squaring F a T a N a ab steep potentials as long as φa follows a highly non-geodesic and using eq. (10), the result follows. path in field space, such as a spinning trajectory. Severing To explain our mechanism, it is convenient to first the link between the slope of the potential and cosmic ignore matter and focus on dark energy domination, in acceleration allows these theories to potentially evade which case the physical picture is similar to multi-field problems endemic to single-field theories. In addition to inflation. Cosmic acceleration occurs when H is nearly the aforementioned swampland conjectures, the flatness of constant, i.e.,   1 with  ≡ −H/H˙ 2 the Hubble slow-roll the potential is typically controlled by a small parameter parameter. While a single canonical scalar requires a flat which is not stable against radiative corrections [50], a potential in order to drive a period of acceleration, in the problem made particularly acute if the swampland bounds presence of multiple fields there can also be acceleration are imposed [51]. Observationally, we will see that, while due to large turning, even in regions where the potential multi-field dark energy is practically indistinguishable is steep. Defining the potential slow-roll parameter  as V from a cosmological constant at the background level, it M 2 GabV V M 2 V 2 + V 2 can provide a distinct, rich, and novel phenomenology for  ≡ Pl a b = Pl N T , (15) V 2 V 2 2 V 2 structure formation. and considering the slow-roll r´egime   1 (but not A concrete example. — In order to illustrate the multi- field dark energy mechanism as simply as possible, we necessarily V  1), it is straightforward to show that [25, 48] restrict ourselves to two fields with a polar parametriza- tion, φa = (r, θ), and impose U(1) invariance through  Ω2 −1 the shift symmetry θ → θ + c. The most general U(1)-  =  1 + . (16) V 9H2 invariant field-space metric is

The essential insight for the models we consider is that, Gab = diag(1, f(r)), (19) for a sufficiently non-geodesic trajectory, Ω  H, the scalars can drive accelerated expansion even when they with Ricci curvature are in a steep region of the potential. 1  1 f 02  R = − f 00 − , (20) This phenomenon underlies many of the novel obser- f 2 f vational signatures of multi-field inflation, as well as its avoidance of the swampland bounds that plague single- where primes denote r derivatives. We will leave f(r) field theories (see also Ref. [49]). Our aim is to investigate general, though for numerical illustrations we will choose its utility for dark energy model-building. The main dif- a flat field space, f(r) = r2.5 ference with the inflationary case is the presence of other In the potential, U(1) invariance (V = V (r)) turns matter fields with significant energy densities. Instead of out to be incompatible with our proposed mechanism demanding   1 (which will not hold during radiation and must therefore be broken. To see this, note that the and matter domination), we want to find the conditions symmetry implies a conserved charge, under which the scalars’ energy density changes slowly, 3 ˙ i.e., DE  1, with Q = a f(r)θ. (21)

3 3 φ˙2 In flat space, a = 1 and there exist stable circular orbits, DE = (wDE + 1) = , (17) 2 2 1 φ˙2 + V but in an expanding universe this is not possible: circular 2 orbits decay as θ˙ ∼ a−3 goes to zero and r falls to the 6 where wDE is the dark energy equation of state. When minimum of its potential. To avoid this problem, the dark energy dominates, DE approaches . Repeating the potential must depend on θ. While we keep V (r, θ) general steps that led to eq. (16), we find, assuming DE  1, when possible, when we need a concrete model we will

 Ω2 −1  =  Ω 1 + , (18) DE V DE 9H2 5 This choice is used in “spintessence” [26], where the two fields form a complex scalar Φ = reiθ with a canonical kinetic term |∂Φ|2. 6 This holds for an arbitrary number N of scalars when Gab and V 4 That Va can be decomposed this way is obvious for two fields. In depend on a single field: there are N − 1 conserved charges, each general it follows from eq. (2). decaying as a−3. 4 borrow from the inflationary literature [22] a potential This behavior is demonstrated, for representative pa- which breaks U(1) as softly as possible, rameters, in fig.1. In the upper panel we plot the gradient of the potential over time, showing that, as promised, the 1 V (r, θ) = V − αθ + m2(r − r )2 , (22) dark energy lives in a steep region of the potential, while 0 2 0 in the lower panel we plot the evolution of wDE over cos- with V0, α, m, and r0 free parameters. mic history, finding that it is extremely close to −1. The As an aside, we note that α is radiatively stable, as combination of these is a unique signature of multi-field U(1) invariance is restored in the α → 0 limit. If the dark energy. We have also checked that the swampland field space is flat and r0 = 0, then the effective r mass condition MPl|∇V |/V & O(1) is satisfied over the entirety 2 2 2 9 mr ≡ m + (∂θ) on a particular background may also of field space, as required by the swampland conjectures be protected from large quantum corrections. While (which are more restrictive than just being true along the the “old” cosmological constant problem, i.e., the radia- cosmological trajectory). tive instability of V0, remains, and requires additional physics to address [52–54], as is typically the case even 2 2 for technically-natural dark energy theories (e.g., [55]), 10 m[H0] V0[H0 MPl] the mechanism presented here provides a promising route 30.0 2.137 towards the construction of dark energy with enhanced 8

V 40.0 2.148 / naturalness properties. | On a cosmological background, the equations of motion V 6 50.0 2.158 | 7 l are P 60.0 2.167

M 4 70.0 2.177 2 2 1  2 ˙2 3MPlH = r˙ + fθ + V + ρM + ρR , (23) 2 2 80.0 2.186 1 0 ˙2 r¨ + 3Hr˙ + Vr − f θ = 0 , (24) 2 0 0 ¨ ˙ 1 f ˙ θ + 3Hθ + Vθ + r˙θ = 0 , (25) wDE at N = 0 f f 0.97 -0.993 where primes denote r derivatives. The r equation of -0.991 motion (24) has the usual forcing term Vr, which pulls r

E 0.98 -0.989

down towards the minimum of its potential (as in standard D

1 0 ˙2 w quintessence), as well as a − 2 f θ term which drives r -0.986 up the potential. Our mechanism relies on balancing -0.984 these competing forces by having the fields spin with 0.99 ˙2 0 2 -0.982 θ ≈ 2Vr/f . For the field-space metric (19) with f = r and potential (22), this amounts to a solution with r 1.00 approximately constant and 10 8 6 4 2 0 2  r0  N θ˙2 ≈ m2 1 − . (26) r It is easy to show that the combination of eqs. (12) FIG. 1. Time evolution of the slope of the potential M |∇V |/V along the trajectory (upper panel) and of the and (26) implies Ω2 = θ˙2 on this trajectory. In the in- Pl dark energy equation of state wDE (lower panel) for the two- flationary context, such circularly spinning solutions are field model with a flat field-space metric and the potential (22). 2 2 −3 −4 cosmological attractors [22]. In order to check whether We have chosen α/H0 MPl = 2 × 10 , r0/MPl = 7 × 10 and this mechanism is also viable for dark energy, we include have varied m (specified in the upper panel). For each choice matter and radiation and solve for the resultant cosmolo- of m we have picked V0 such that the spatial curvature of the gies numerically.8 We generically find solutions which universe vanishes. There is no strong dependence on initial realize the proposed mechanism: despite being signifi- conditions. Our time variable is the number of e-foldings N ≡ ln(a), with N = 0 corresponding to the present. cantly displaced from the minimum of its potential at r = r , cosmic history is quantitatively very close to 0 While we expect w ≈ −1 during dark energy dom- ΛCDM. DE ination, as the physics is similar to multi-field inflation,

7 9 For a dynamical-systems analysis of these equations for restricted Strictly speaking V eventually becomes small at very large r, choices of V = V (r), see Refs. [34, 35]. but the potential (22) should be properly viewed as part of an 8 The codes used in this paper are publicly available at https: effective field theory and so cannot be trusted at arbitrarily large //github.com/valerivardanyan/Multifield-Dark-Energy. field values. 5

we see from fig.1 that this equation of state also holds We emphasize that the model discussed in this section during the matter- and radiation-dominated eras. During serves as a minimal working example. Our results do not these epochs, Hubble friction dominates the forcing terms depend strongly on the details of the model, and we expect in eqs. (24) and (25), causing r and θ to freeze. As matter them to be qualitatively robust for any potential that and radiation dilute away, the Hubble friction becomes supports strongly non-geodesic motion in field space.10 smaller than the forcing terms and the fields start to Clustering dark energy. — While the dark energy roll. The r field falls slightly down the potential before mechanism proposed here is likely to be observationally stabilizing as θ spins up, transitioning into the spinning indistinguishable from ΛCDM at the background level, r´egimeand the onset of dark energy domination. Dur- the story changes dramatically when we consider per- ing this period, the dark energy equation of state starts turbations. In this section we briefly discuss why one evolving from its frozen value of −1, but only slightly: in should expect these theories to produce novel signatures the spinning r´egime,the Hubble slow-roll parameter  DE in structure formation, while saving a full analysis of per- is suppressed by the turning rate, cf. eq. (18), so that turbations and the comparison to observations for future w remains close to −1. DE work. Observationally, this model is to some extent a victim For a wide range of parameters, the sound speed of of its own success: it mimics ΛCDM so efficiently that it fluctuations is heavily suppressed, leading to clustering is unlikely to be distinguishable from the concordance cos- dark energy. To see this, we expand the fields around their mology at the background level, even with qualitatively background values as φa = φ¯a + δφa. It is convenient to rather different physics than ΛCDM or standard slow-roll work with the field fluctuations parallel and perpendicular quintessence. Forecast analyses from the forthcoming to the background trajectory, Stage IV cosmological surveys predict percent-level con- straints on parameters like the dark energy equation of T a N a δφ ≡ Taδφ , δφ ≡ Naδφ . (27) state. The Euclid space mission [56], an important rep- resentative of these surveys, is expected to measure the Working with the Newtonian gauge for scalar metric per- present value of wDE, commonly denoted as w0, with at turbations,

best a 1σ uncertainty of σw0 ≈ 0.025 [57]. As seen in the 2 2
lower panel of fig.1, Euclid will not be able to distinguish gµν = diag −a (1 − 2Φ), δija (1 + 2Φ) , (28) wDE in the dark energy model proposed here from the ΛCDM value of −1, although in principle there may be where Φ is the gravitational potential, and including regions of parameter space in which wDE + 1 is just large dark matter fluctuations δDM, the full linearized Einstein enough to be observable. equations become

T 00 T 0 N 0  2 2 2  T N 2 0 0 δφ + 2Hδφ + 2aΩδφ + k + a DT V δφ + 4aHΩδφ − 2a ΦDT V + 4Φ φ = 0, (29) N 00 N 0 T 0  2 2  N 0 δφ + 2Hδφ − 2aΩδφ + k + Meff δφ − 2aΦφ Ω = 0, (30) 2 00 0 2 0
2 a T N
Φ + 6HΦ + 4H + 2H Φ + k Φ = 2 ρDMδDM + 2VT δφ + 2VN δφ , (31) 2MPl

where primes denote conformal time derivatives, H is the effective δφN mass is conformal time Hubble rate, ρ is the background dark DM φ02 matter density, V ≡ D T aV and V ≡ D N aV , where M 2 ≡ a2V − a2Ω2 + R . (32) T a N a eff NN 2 Da is the covariant derivative associated to Gab, and the a b Here VNN ≡ N N DaDbV and R is the Ricci scalar for Gab. For concreteness, we focus on scalar field perturba- tions on scales smaller than the sound horizon and ignore 10 Minor quantitative details, such as the behavior of the fields when dark matter fluctuations and gravitational backreaction, they are subdominant, can change from model to model. For instance, by appropriately changing the θ potential, the system which suffices to illustrate the important physical effects. may enter a scaling r´egimeduring the matter-dominated era, in In this limit, the scalar equations of motion eq. (29) and analogy with the single-field model in Ref. [58]. This allows the eq. (30) are transition from frozen to spinning behavior to occur more quickly, as the θ field is already dynamical when dark energy becomes δφT 00 + k2δφT = −2aΩδφN 0, (33) dominant. While illustrative, this scenario is somewhat contrived, N 00 2 2
N T 0 and the θ shift symmetry would no longer be broken softly. δφ + k + Meff δφ = 2aΩδφ . (34) 6

Note that we have neglected a small mass term in eq. (33), Recall that Meff is this mode’s mass in the geodesic which is necessarily suppressed in the spinning r´egime limit Ω → 0. Spinning increases the mass by a fac- ¨ ˙ −2 where φ  Hφ. tor of cs . This suggests a wide range of intermediate 2 2 2 2 −2 We see from eqs. (33) and (34) that a non-zero Ω intro- scales, H  k /a  Meff cs , where the heavy mode duces a coupling between δφT and δφN . To identify the can be integrated out, leading to a simpler single-field propagating degrees of freedom, we look for solutions with effective theory, as in inflation [21, 22], while remaining time dependence eiωτ to obtain the dispersion relation, in the sub-horizon r´egime.We note speculatively that a

4 2 2 2 2 2
2 2 2
condensate of this heavy field could potentially be a dark ω −ω 2k + Meff + 4a Ω +k k + Meff = 0. (35) matter candidate; we leave a more detailed analysis of On geodesic trajectories, Ω = 0 and the dispersion relation this possibility to future work. 2 2 2 2 2 In regions of parameter space where the sound speed is factorizes, (ω − k )(ω − k − Meff ) = 0, from which we suppressed, we expect enhanced structure formation in can identify a light mode and a heavy mode of mass Meff , each propagating at the speed of light. Including Ω, the the late universe, as there is a well-known correspondence full dispersion relations are between a small sound speed and dark energy clustering [40, 42–44]. The physical explanation is that a reduced M 2 + 4a2Ω2 ω2 = eff + k2 speed of sound pushes the Jeans instability to sub-horizon ± 2 scales. Modes of the light field with larger wavelengths s 11 M 2 + 4a2Ω2 2 will therefore cluster on observationally accessible scales. ± eff + 4a2Ω2k2. (36) This is in contrast to canonical, single-field dark energy, 2 where the Jeans scale is super-horizon, so the Jeans in-

The light mode corresponds to ω−, since ω− → 0 as k → 0. stability is not observable. Considering scales larger than the Compton wavelength In brief we mention two other reasons to expect clus- 2 2 2 2 of the heavy mode, k  Meff + 4a Ω , and expanding tering or other interesting features in structure formation the light-mode dispersion relation to leading order, we in theories of the type discussed here, depending on the see that it propagates with a modified sound speed, details of the model: 2 2 2 4 1. While in the inflationary context the heavy mode is ω− = cs k + O(k ), (37) suppressed in amplitude [22], it could in principle where (depending on initial conditions) be non-negligible 4a2Ω2 in the late universe. This situation would be similar c−2 ≡ 1 + . (38) s M 2 to a canonical massive scalar field, which clusters eff on sub-horizon scales when its mass is larger than 2 2 2 The sound speed is suppressed when a Ω  Meff , which the Hubble scale (cf., e.g., Refs. [59, 60]). per eq. (32) requires Ω2 ≈ V + Rφ˙2/2. To illustrate NN 2. The field space curvature R contributes to M , quantitatively the typical scales involved, we take as an eff and a sufficiently negative curvature can render the example m = 30H (cf. fig.1), for which we have, in the 0 heavy mode tachyonic. During inflation this phe- present day, a2Ω2/H2 ≈ 750 and M 2 /H2 ≈ 150, with 0 eff 0 nomenon is known as geometrical destabilization c2 ≈ 0.047. s and is considered problematic, spoiling otherwise While this r´egime may seem highly tuned, it is in successful models [61]. In context of the late uni- fact supported in the model discussed above, as we have verse, however, a mild tachyonic instability might confirmed numerically for the parameter choice in fig.1. imprint unique features on the dark matter distri- This c2  1 attractor is well-known in the inflationary s bution, opening up a new window for probing the context, and exists as long as the potential parameters curvature of field space. satisfy [22] r m V α Conclusions. — In this Letter we have proposed a novel 1/4 0 √ 1  r0V0  2 2  . (39) class of multi-field dark energy models where the fields αMPl m M mMPl V0 Pl do not follow geodesic trajectories in field space, allowing During dark energy domination, the same equations of steep potentials to lead to cosmic acceleration. We have motion apply and so the same attractor is present. argued why these models are theoretically well-motivated For completeness we present the heavy mode’s disper- when new developments in high energy physics and quan- sion relation to quadratic order in k, tum gravity are considered. By focusing on a concrete 2 2 2 2 2 2 2 Meff + 8a Ω 2 4 ω+ = Meff + 4a Ω + 2 2 2 k + O(k ) Meff + 4a Ω 11 2 While eqs. (33) and (34) hold only below the sound horizon, the Meff 2 2 4 2 2 2 −2 sound speed remains small at all scales k /a  Meff cs , where = 2 + (2 − cs )k + O(k ). (40) cs the heavy mode can be integrated out [22]. 7 and representative example, we have studied the cosmo- Nemanja Kaloper, and John March-Russell, “String Axi- logical background evolution in these models and shown verse,” Phys. Rev. D 81, 123530 (2010), arXiv:0905.4720 that they are practically indistinguishable from the stan- [hep-th]. dard ΛCDM model. This means that constraining the [5] Georges Obied, Hirosi Ooguri, Lev Spodyneiko, and Cumrun Vafa, “De Sitter Space and the Swampland,” equation of state of dark energy by next-generation cos- (2018), arXiv:1806.08362 [hep-th]. mological surveys to values arbitrarily close to −1 will [6] Prateek Agrawal, Georges Obied, Paul J. Steinhardt, and not exclude the possibility of dark energy being highly Cumrun Vafa, “On the Cosmological Implications of the dynamical. We have argued, however, that our models String Swampland,” Phys. Lett. B784, 271–276 (2018), do result in features in formation and evolution of the arXiv:1806.09718 [hep-th]. cosmic large-scale structure, which can potentially dis- [7] Sumit K. Garg and Chethan Krishnan, “Bounds on Slow tinguish the models from ΛCDM and single-field dark Roll and the de Sitter Swampland,” JHEP 11, 075 (2019), arXiv:1807.05193 [hep-th]. energy. We have derived and presented the full linearized [8] Hirosi Ooguri, Eran Palti, Gary Shiu, and Cumrun Vafa, Einstein equations where gravitational effects and dark “Distance and de Sitter Conjectures on the Swampland,” matter fluctuations are included. By restricting ourselves Phys. Lett. B788, 180–184 (2019), arXiv:1810.05506 [hep- to subhorizon scales and only considering dynamics of the th]. scalar field fluctuations in the absence of gravitational [9] Shamit Kachru, Renata Kallosh, Andrei D. Linde, and backreaction, we have demonstrated that in theories of Sandip P. Trivedi, “De Sitter vacua in string theory,” the type discussed in this Letter dark energy is expected Phys. Rev. D 68, 046005 (2003), arXiv:hep-th/0301240. [10] Yashar Akrami, Renata Kallosh, Andrei Linde, and Valeri to cluster during cosmic structure formation. We have Vardanyan, “The Landscape, the Swampland and the identified three reasons for this clustering: sound speed Era of Precision Cosmology,” Fortsch. Phys. 67, 1800075 suppression for the light mode of dark energy fluctuations, (2019), arXiv:1808.09440 [hep-th]. non-negligibility of the heavy mode, and tachyonic insta- [11] Michele Cicoli, Senarath De Alwis, Anshuman Maha- bility of the heavy mode resulting from a negative field rana, Francesco Muia, and Fernando Quevedo, “De Sit- space curvature. Each of these effects on structure forma- ter vs Quintessence in String Theory,” Fortsch. Phys. 67, tion requires a more detailed numerical analysis to derive 1800079 (2019), arXiv:1808.08967 [hep-th]. [12] Shamit Kachru and Sandip P. Trivedi, “A comment on quantitative predictions for cosmological observables, in effective field theories of flux vacua,” Fortsch. Phys. 67, particular in the context of clustering dark energy [36–47]. 1800086 (2019), arXiv:1808.08971 [hep-th]. This is beyond the scope of this Letter and we leave it for [13] Renata Kallosh, Andrei Linde, Evan McDonough, and future work. Marco Scalisi, “dS Vacua and the Swampland,” JHEP We thank Martin Bucher, Edmund J. Copeland, Nick 03, 134 (2019), arXiv:1901.02022 [hep-th]. [14] Marco Raveri, Wayne Hu, and Savdeep Sethi, “Swamp- Kaiser, S´ebastienRenaux-Petel, and Benjamin D. Wan- land Conjectures and Late-Time Cosmology,” Phys. Rev. delt for helpful discussions. Y.A. is supported by LabEx D99, 083518 (2019), arXiv:1812.10448 [hep-th]. ENS-ICFP: ANR-10-LABX-0010/ANR-10-IDEX-0001-02 [15] Misao Sasaki and Ewan D. Stewart, “A General analytic PSL*. M.S. is supported in part by JSPS KAKENHI formula for the spectral index of the density perturbations No. 20H04727. A.R.S. is supported by DOE HEP grants produced during inflation,” Prog. Theor. Phys. 95, 71–78 DOE DE-FG02-04ER41338 and FG02-06ER41449 and (1996), arXiv:astro-ph/9507001. by the McWilliams Center for Cosmology, Carnegie Mel- [16] S. Groot Nibbelink and B.J.W. van Tent, “Scalar pertur- bations during multiple field slow-roll inflation,” Class. lon University. V.V. is supported by the WPI Research Quant. Grav. 19, 613–640 (2002), arXiv:hep-ph/0107272. Center Initiative, MEXT, Japan. [17] Z. Lalak, D. Langlois, S. Pokorski, and K. Turzynski, “Curvature and isocurvature perturbations in two-field inflation,” JCAP 07, 014 (2007), arXiv:0704.0212 [hep- th]. [18] Amjad Ashoorioon, Hassan Firouzjahi, and M.M. Sheikh- ∗ [email protected] Jabbari, “M-flation: Inflation From Matrix Valued Scalar † [email protected] Fields,” JCAP 06, 018 (2009), arXiv:0903.1481 [hep-th]. ‡ [email protected] [19] Ana Achucarro, Jinn-Ouk Gong, Sjoerd Hardeman, Gon- § [email protected] zalo A. Palma, and Subodh P. Patil, “Mass hierarchies [1] Edmund J. Copeland, M. Sami, and Shinji Tsujikawa, and non-decoupling in multi-scalar field dynamics,” Phys. “Dynamics of dark energy,” Int. J. Mod. Phys. D 15, 1753– Rev. D 84, 043502 (2011), arXiv:1005.3848 [hep-th]. 1936 (2006), arXiv:hep-th/0603057. [20] Ana Achucarro, Jinn-Ouk Gong, Sjoerd Hardeman, Gon- [2] Timothy Clifton, Pedro G. Ferreira, Antonio Padilla, and zalo A. Palma, and Subodh P. Patil, “Features of heavy Constantinos Skordis, “Modified Gravity and Cosmology,” physics in the CMB power spectrum,” JCAP 01, 030 Phys. Rept. 513, 1–189 (2012), arXiv:1106.2476 [astro- (2011), arXiv:1010.3693 [hep-ph]. ph.CO]. [21] Ana Achucarro, Jinn-Ouk Gong, Sjoerd Hardeman, Gon- [3] Michael R. Douglas and Shamit Kachru, “Flux compacti- zalo A. Palma, and Subodh P. Patil, “Effective theories fication,” Rev. Mod. Phys. 79, 733–796 (2007), arXiv:hep- of single field inflation when heavy fields matter,” JHEP th/0610102. 05, 066 (2012), arXiv:1201.6342 [hep-th]. [4] Asimina Arvanitaki, Savas Dimopoulos, Sergei Dubovsky, [22] Ana Achucarro, Vicente Atal, Sebastian Cespedes, Jinn- 8

Ouk Gong, Gonzalo A. Palma, and Subodh P. Patil, [40] Wayne Hu and Ryan Scranton, “Measuring dark energy “Heavy fields, reduced speeds of sound and decoupling clustering with CMB-galaxy correlations,” Phys. Rev. D during inflation,” Phys. Rev. D 86, 121301 (2012), 70, 123002 (2004), arXiv:astro-ph/0408456. arXiv:1205.0710 [hep-th]. [41] Wayne Hu, “Crossing the phantom divide: Dark energy [23] Shi Pi and Misao Sasaki, “Curvature Perturbation Spec- internal degrees of freedom,” Phys. Rev. D 71, 047301 trum in Two-field Inflation with a Turning Trajectory,” (2005), arXiv:astro-ph/0410680. JCAP 10, 051 (2012), arXiv:1205.0161 [hep-th]. [42] Masahiro Takada, “Can A Galaxy Redshift Survey Mea- [24] Ana Ach´ucarro,Renata Kallosh, Andrei Linde, Dong- sure Dark Energy Clustering?” Phys. Rev. D 74, 043505 Gang Wang, and Yvette Welling, “Universality (2006), arXiv:astro-ph/0606533. of multi-field α-attractors,” JCAP 04, 028 (2018), [43] Paolo Creminelli, Guido D’Amico, Jorge Norena, and Fil- arXiv:1711.09478 [hep-th]. ippo Vernizzi, “The Effective Theory of Quintessence: [25] Ana Ach´ucarroand Gonzalo A. Palma, “The string the w < −1 Side Unveiled,” JCAP 02, 018 (2009), swampland constraints require multi-field inflation,” arXiv:0811.0827 [astro-ph]. JCAP 02, 041 (2019), arXiv:1807.04390 [hep-th]. [44] Paolo Creminelli, Guido D’Amico, Jorge Norena, [26] Latham A. Boyle, Robert R. Caldwell, and Marc Leonardo Senatore, and Filippo Vernizzi, “Spherical Kamionkowski, “Spintessence! New models for dark mat- collapse in quintessence models with zero speed of sound,” ter and dark energy,” Phys. Lett. B 545, 17–22 (2002), JCAP 03, 027 (2010), arXiv:0911.2701 [astro-ph.CO]. arXiv:astro-ph/0105318. [45] Luca Amendola et al., “Cosmology and fundamental [27] Soo A. Kim, Andrew R. Liddle, and Shinji Tsujikawa, physics with the Euclid satellite,” Living Rev. Rel. 21, 2 “Dynamics of assisted quintessence,” Phys. Rev. D 72, (2018), arXiv:1606.00180 [astro-ph.CO]. 043506 (2005), arXiv:astro-ph/0506076. [46] Caroline Heneka, David Rapetti, Matteo Cataneo, [28] Carsten van de Bruck and Joel M. Weller, “Quintessence Adam B. Mantz, Steven W. Allen, and Anja von der dynamics with two scalar fields and mixed kinetic terms,” Linden, “Cold dark energy constraints from the abun- Phys. Rev. D 80, 123014 (2009), arXiv:0910.1934 [astro- dance of galaxy clusters,” Mon. Not. Roy. Astron. Soc. ph.CO]. 473, 3882–3894 (2018), arXiv:1701.07319 [astro-ph.CO]. [29] Jose Beltran Jimenez, Paulo Santos, and David F. Mota, [47] Farbod Hassani, Julian Adamek, and Martin Kunz, “Clus- “Cosmology with a Continuous Tower of Scalar Fields,” tering dark energy imprints on cosmological observables Phys. Lett. B723, 7–14 (2013), arXiv:1212.5266 [astro- of the gravitational field,” Mon. Not. Roy. Astron. Soc. ph.CO]. 500, 4514–4529 (2020), arXiv:2007.04968 [astro-ph.CO]. [30] Valeri Vardanyan and Luca Amendola, “How can we tell [48] Alexander Hetz and Gonzalo A. Palma, “Sound Speed of whether dark energy is composed of multiple fields?” Phys. Primordial Fluctuations in Supergravity Inflation,” Phys. Rev. D 92, 024009 (2015), arXiv:1502.05922 [gr-qc]. Rev. Lett. 117, 101301 (2016), arXiv:1601.05457 [hep-th]. [31] Alexander Leithes, Karim A. Malik, David J. Mulryne, [49] Adam R. Solomon and Mark Trodden, “Non-canonical ki- and Nelson J. Nunes, “Linear Density Perturbations netic structures in the swampland,” JCAP 09, 049 (2020), in Multifield Coupled Quintessence,” Phys. Rev. D 95, arXiv:2004.09526 [hep-th]. 123519 (2017), arXiv:1608.00908 [astro-ph.CO]. [50] Christopher F. Kolda and David H. Lyth, “Quintessential [32] Yashar Akrami, Renata Kallosh, Andrei Linde, and difficulties,” Phys. Lett. B 458, 197–201 (1999), arXiv:hep- Valeri Vardanyan, “Dark energy, α-attractors, and ph/9811375. large-scale structure surveys,” JCAP 1806, 041 (2018), [51] Mark P. Hertzberg, McCullen Sandora, and Mark Trod- arXiv:1712.09693 [hep-th]. den, “Quantum Fine-Tuning in Stringy Quintessence Mod- [33] Adam R. Brown, “Hyperbolic Inflation,” Phys. Rev. Lett. els,” Phys. Lett. B 797, 134878 (2019), arXiv:1812.03184 121, 251601 (2018), arXiv:1705.03023 [hep-th]. [hep-th]. [34] Michele Cicoli, Giuseppe Dibitetto, and Francisco G. [52] Steven Weinberg, “The Cosmological Constant Problem,” Pedro, “New accelerating solutions in late-time cosmol- Rev. Mod. Phys. 61, 1–23 (1989), [,569(1988)]. ogy,” Phys. Rev. D 101, 103524 (2020), arXiv:2002.02695 [53] Gia Dvali, Stefan Hofmann, and Justin Khoury, “Degrav- [gr-qc]. itation of the cosmological constant and graviton width,” [35] Michele Cicoli, Giuseppe Dibitetto, and Fran- Phys. Rev. D 76, 084006 (2007), arXiv:hep-th/0703027. cisco G. Pedro, “Out of the Swampland with Multifield [54] Justin Khoury, Jeremy Sakstein, and Adam R. Solomon, Quintessence?” JHEP 10, 035 (2020), arXiv:2007.11011 “Superfluids and the Cosmological Constant Problem,” [hep-th]. JCAP 08, 024 (2018), arXiv:1805.05937 [hep-th]. [36] Rachel Bean and Olivier Dore, “Probing dark energy [55] Guido D’Amico, Nemanja Kaloper, and Albion Lawrence, perturbations: The Dark energy equation of state and “Strongly Coupled Quintessence,” Phys. Rev. D 100, speed of sound as measured by WMAP,” Phys. Rev. D 103504 (2019), arXiv:1809.05109 [hep-th]. 69, 083503 (2004), arXiv:astro-ph/0307100. [56] R. Laureijs et al. (EUCLID), “Euclid Definition Study [37] Jochen Weller and A.M. Lewis, “Large scale cosmic mi- Report,” (2011), arXiv:1110.3193 [astro-ph.CO]. crowave background anisotropies and dark energy,” Mon. [57] A. Blanchard et al. (Euclid), “Euclid preparation: VII. Not. Roy. Astron. Soc. 346, 987–993 (2003), arXiv:astro- Forecast validation for Euclid cosmological probes,” As- ph/0307104. tron. Astrophys. 642, A191 (2020), arXiv:1910.09273 [38] D.F. Mota and C. van de Bruck, “On the Spherical col- [astro-ph.CO]. lapse model in dark energy cosmologies,” Astron. Astro- [58] Edmund J. Copeland, Andrew R Liddle, and David phys. 421, 71–81 (2004), arXiv:astro-ph/0401504. Wands, “Exponential potentials and cosmological scaling [39] Nelson J. Nunes and D.F. Mota, “Structure formation in solutions,” Phys. Rev. D 57, 4686–4690 (1998), arXiv:gr- inhomogeneous dark energy models,” Mon. Not. Roy. As- qc/9711068. tron. Soc. 368, 751–758 (2006), arXiv:astro-ph/0409481. [59] Yasusada Nambu and Misao Sasaki, “Quantum Treatment 9

of Cosmological Axion Perturbations,” Phys. Rev. D 42, [61] S´ebastienRenaux-Petel and Krzysztof Turzy´nski,“Ge- 3918–3924 (1990). ometrical Destabilization of Inflation,” Phys. Rev. Lett. [60] Luca Amendola and Shinji Tsujikawa, Dark Energy: The- 117, 141301 (2016), arXiv:1510.01281 [astro-ph.CO]. ory and Observations (Cambridge University Press, 2015).