Using dark energy to suppress power at small scales
Abstract
The latest Planck results reconfirm the existence of a slight but chronic tension between the bestfit Cosmic Microwave Background (CMB) and lowredshift observables: power seems to be consistently lacking in the late universe across a range of observables (e.g. weak lensing, cluster counts). We propose a twoparameter model for dark energy where the dark energy is sufficiently like dark matter at large scales to keep the CMB unchanged but where it does not cluster at small scales, preventing concordance collapse and erasing power. We thus exploit the generic scaledependence of dark energy instead of the more usual timedependence to address the tension in the data. The combination of CMB, distance and weak lensing data somewhat prefer our model to CDM, at . Moreover, this improved solution has , consistent with the value implied by cluster counts.
I Introduction
The recent Planck cosmology results Ade:2013zuv (); Ade:2015xua () provide stunning support for the CDM “standard model” of cosmology. One of the few results that are not in quite as excellent agreement with the parameter constraints from the measurements of the anisotropies in the cosmic microwave background (CMB) is the determination of the amplitude of cosmological perturbations at late times on small scales. This slight tension is most apparent when comparing to weak lensing measurements as provided by CFHTLenS, which prefers lower values of (Ref. (Ade:2015xua, , Fig. 18)) and also the lowerthanexpected cluster abundances (Ref. (Ade:2013lmv, , Fig. 10) and Ade:2015fva ()). In addition, dark matter growth rates, proportional to , tend to fall on the low side of the concordance values, even if they are not inconsistent (Ade:2015xua, , Fig. 16). Although it is well possible that these discrepancies come from systematic effects in the analysis of the data in the lowredshift universe, it is also the case that lateuniverse measurements of consistently imply a smaller value than that obtained by processing the initial amplitude of perturbations observed in the CMB through concordance gravitational collapse.
One possible resolution of this conflict is to suppress the clustering at low redshift through a new physical mechanism active only at late times. For example, heavy neutrinos could play such a role for clusters Wyman:2013lza (); Battye:2014qga (). However, the latest data do not seem to support this Ade:2015xua ().
Somewhat more exotically, dark energy could be dynamical, changing the evolution of the universe at late times. However, the analysis in (Ade:2015rim, , Fig. 4) shows that distance data from baryon acoustic oscillations (BAO) constrains the usual quintessencelike models too much to significantly improve the possible fits to weak lensing data. More general models of modified gravity in the quasistatic limit that do not contain ghosts typically serve to increase growth rates Piazza:2013pua (); Nesseris:2008mq (); Nesseris:2009jf (); Basilakos:2013nfa (); Nesseris:2013jea (); Nesseris:2011pc (); Nesseris:2013fca (); Nesseris:2014mea (), although a temporary suppression around redshift is also typical Perenon:2015sla () and could possibly be exploited with some tuning.
There is another possibility: dynamical dark energy could affect darkmatter clustering in a scaledependent manner. The CMB is mostly a largescale observable, while galaxy weak lensing has only been measured on small scales; clusters are also smallscale phenomena. A model where DE causes the dark matter to cluster slower at small scales than at large without significantly modifying the background expansion history could produce the desired phenomenology.
We propose a mechanism to achieve this: we exploit the dark degeneracy Kunz:2007rk () to trade some of the dark matter for dark energy. Provided the two are similar enough, observables remain unchanged with respect to CDM. We investigate a very minimal extension of the concordance cosmological model: we allow the dark energy to be a perfect fluid with constant pressure and a constant nonzero sound speed. This keeps the expansion history exactly the same as CDM. Such a choice allows us to concentrate on investigating the effect of changing the properties of the perturbations of dark energy while keeping the background, which is already very well constrained, fixed.
If the dark energy has exactly zero sound speed, then the dark degeneracy also ensures that the linear perturbations in the metric are unchanged with respect to CDM. Effectively, the dark energy component that replaces part of the dark matter also clusters like that missing dark matter. However, if the sound speed is nonzero then the dark energy clusters less than the missing dark matter on scales inside its sound horizon. We show that this model, as a result of this change in the behavior of DE at its sound horizon, indeed goes some way to relieve the tension between the early and late universe and is favored over CDM when the combination of CMB and weak lensing data are used. Moreover, the bestfit model has a lower value of which should help with the tension with cluster number counts.
Ii DarkEnergy Model
All measurements of the cosmological background expansion history, such as baryon acoustic oscillations (BAO) or supernovae (SNe), or even cosmic chronometers, only measure relative distances or times, which are integrated quantities depending only on the dimensionless Hubble parameter , where is the Hubble constant. Thus, provided that the equation of state of the total dark sector be kept constant (i.e. the sum of CDM and DE), distance measurements are completely degenerate to any changes in relative composition between the two species. This fact was named the dark degeneracy in Ref. Kunz:2007rk () and it implies that background measurements are incapable of measuring the cold dark matter (CDM) fraction absent a choice of parametrization for . Measurements of largescale structure, such as the CMB anisotropies, can break this degeneracy, but require a model for the behavior of darkenergy perturbations to be specified Amendola:2012ky (); Motta:2013cwa (). One should really think of the density fraction of DM as a perturbation variable and not a background parameter Bellini:2014fua (). While is frequently parametrized as constant and the bestfit cosmology prefers values close to with a good determination of Ade:2015xua (), the indubitable fact is that our universe seems to be close to one which has a constant pressure at late times, making a DE component with a constant energy density just one of the possibilities. As we will show here, this is not just an academic discussion.
Our only modification to the concordance model is to use a perfectfluid dark energy with a constant (restframe) sound speed and a particular designer choice of equation of state that ensures that the background expansion history exactly mimics CDM (i.e. our DE has constant pressure).
ii.1 Background
In order to maintain a fixed expansion history, we pick the density fraction of dark energy such that its sum with the CDM density fraction remains the same as the sum of the density fraction of the cosmological constant and some reference CDM density in CDM. We will refit the data for the values of all the parameters in the paradigm of our extended model, but for the moment, one should think of it as the standard Planck value for ,
(1) 
where is the scale factor. This relation implies that the equation of state of the DE satisfies
(2) 
where and where the subscript denotes the value today. is the only new parameter introduced at the level of the background and it replaces the more typical constant parametrization for .
Generically, a violation of the null energy condition leads to instabilities. These are inescapable for NECviolating perfect fluids (either ghost or gradient instabilities) Dubovsky:2005xd (), but may even exist in more general cases Sawicki:2012pz (). We thus only consider , i.e. .
We note here that such choices for as Eq. (2) are badly captured by the prevailing parametrizations in use in the community, such as a constant or the CPL parametrization Chevallier:2000qy (); Linder:2002et (), which is one of the origins of our results. It is much more similar to models with a transition in the equation of state (e.g. Ref. Bassett:2002qu ()) or Early Dark Energy (EDE) parametrizations (e.g. Ref. Pettorino:2013ia ()), since the DE component tracks the CDM before the acceleration era. Other models with similar include quintessence coupled to neutrinos Amendola:2007yx () or the generalized Chaplygin gas (GCG) Kamenshchik:2001cp ().
The evolution of the background is essentially missing all information about the microphysical content of the DE model, which is defined through the behavior of perturbations. For example, modeling the GCG perturbations in various ways gives rise to completely different constraints Sandvik:2002jz (); Wang:2013qy (). We focus on defining the evolution of perturbations next.
ii.2 Linear properties
We start off with a FriedmannRobertsonWalker (FRW) metric with small scalar perturbations, with the gauge fixed to be Newtonian:
(3) 
keeping only the scalar perturbations.
By construction, the scalar sector of the linear perturbation equations for the dark energy takes the standard form for a perfect fluid which we do not give here (see e.g. Hu:1998kj (); dePutter:2010vy (); Sapone:2009mb ()). Since the background expansion history mimics CDM, the pressure of the dark energy is constant and therefore the adiabatic sound speed exactly. Therefore the only new parameter that we need to specify is the sound speed which we take to be constant and to lie in the range . This is the second and final new parameter of our model. We can define the (physical) Jeans scale as
(4) 
which we will demonstrate separates two different regimes of evolution for perturbations.
Provided that the dark energy consists of a single degree of freedom, and therefore has no additional internal freedom, the dynamics of the perturbed cosmology can be rewritten in a manner much more illustrative than the usual presentation with fluid variables (see (Motta:2013cwa, , Eq. (21))): one combines the various constraints given by the Einstein equations to eliminate the DE fluid variables and to form a single equation of motion for the Newtonian potential which is coupled to the other matter species,
(5)  
(6) 
where the fluid variables with subscript “m" describe the total perturbation of the matter sector, including radiation, baryons, CDM and neutrinos, but not the DE. This system is closed by supplying the evolution equations for the total matter energymomentum tensor, whether by assuming the matter be a collection of fluids, or by solving some sort of Boltzmann hierarchy.
Equation (5) should be contrasted with its version in CDM:
(7) 
There are three important differences between Eqs. (5) and (7):

the presence of a soundhorizon (Jeans) term ;

the sourcing of the potential by the matter perturbations with a coupling ;

the potential in our model is in fact an independent degree of freedom with its own initial conditions.^{1}^{1}1Note that in tracker models of DE, the isocurvature modes decay away Malquarti:2002iu (). In CDM, it is completely constrained and not independent.
Note that the cosmological horizon does not appear as a relevant scale here at all; its sound horizon is the only scale relevant for the DE.
At scales larger than the DE sound horizon, , the quasistatic approximation for the DE cannot be used Sawicki:2015zya (). The Jeans term on the lefthand side of Eq. (5) is negligible and the difference with respect to the CDM case is the new source term coupled with on the righthand side. It affects the evolution of the potentials whenever it is comparable to the pressure perturbation term coming from radiation, essentially providing a new early integrated SachsWolfe contribution. For example, around the time of decoupling, the CDM and radiation density fractions are comparable, so the new term is negligible when . For such DE sound speeds, the evolution of perturbations in the metric is the same as in CDM, irrespective of the value of , i.e. how much CDM has been removed and replaced with the DE.
This means that, outside of its sound horizon, the DE clusters just like CDM since it cannot react causally to create a pressure to arrest the dustlike collapse along geodesics. The fact that quintessence tracking models behave in this way at large scales was shown in Ref. Malquarti:2002iu () and this is also one of the implications of the separateuniverse approach Bertschinger:2006aw (). Essentially, the part of the DE energy density that is tracking contributes in the same way as the CDM dust at these scales and therefore the model looks like CDM there.
At scales inside the Jeans scale, , the Jeans term makes the evolution equation (5) stiff, leading to the solution usually known as the quasistatic solution,
(8) 
When the pressure is irrelevant, this recovers the standard Poisson equation of CDM, with the gravitational potential driven purely by the comoving density fluctuations of matter. At these scales, the DE reacts causally and sets up a pressure profile to prevent its clustering. The gravitational potential is sourced only by the standard species. Thus picking a to replace a part of CDM will reduce the depth of the potentials and therefore also the growth rate on scales inside the DE Jeans length.
When the total pressure perturbation is not negligible compared to the total density perturbation, e.g. at decoupling, the final term in Eq. (8) could in principle contribute for small DE sound speeds. On the other hand, we have to bear in mind that the angular scale of the DE Jeans scale in the CMB is proportional to the DE sound speed. We can use the Limber approximation to estimate that the multipoles at which the approximation (8) is valid is
(9) 
where is the multipole at which the first CMB peak is centered and therefore the solution (8) is only possibly relevant to the observed CMB anisotropies for .
We illustrate the effect of our model on the CMB TT powerspectrum in Fig. 1, where it can be seen that already for , the modification of the CDM curve from the terms discussed above is negligible when the effects from the late universe are not included. This can be verified by comparing a perfect Planck forecast with no latetime effects included for CDM with our model.
Turning to the late universe, as the mode crosses the DE sound horizon, the standard CDMlike solution for the potential valid superJeans is modified to (8) as the DE perturbations are erased. This then modifies the growth function for that mode at all subsequent times, with the matter perturbation evolving according to
(10) 
This is essentially the same equation as in the CDM case, apart from the fact that even during matter domination , which changes the growing mode evolution during matter domination from to
(11) 
for . This means that the matter growth functions at some redshift during matter domination receives a scaledependent correction:
(12) 
where is the redshift of observation. Therefore, even for a fixed amount of dark energy , a larger sound speed means that the pivot of the suppression effect lies at smaller and therefore the reduction in power is higher for any particular mode inside the sound horizon (see Fig. 3 for the effect on the CDM power spectrum or Fig. 3 for an equivalent presentation of how the CDM power spectrum amplitude today depends on the amount and sound speed of our DE).
We remind the reader that the CDM and gravitational potential power spectra are not affected for , since the dark energy is clustering and the potential has the full depth as in CDM.^{2}^{2}2At very late times, the comoving horizon shrinks as a result of the acceleration. Thus modes can in principle exit the DE sound horizon and have their growth increase back to CDM rates.
In principle, the amplitude of is closely related to the rate of formation and mass function of halos. However, we should note here that two different ’s can be defined: one given by the autocorrelation of the CDM density contrast, while the other defined from the autocorrelation of the gravitational potential. In the CDM, limit, they are identical. But whenever the DE Jeans length lies at scales larger than 8 comoving Mpc, the two are different, with the former larger. On one hand, the forming clusters only know about the gravitational potential, since the CDM does not interact in any way but through gravity. On the other hand, the clusters form in configuration space and locally only the mass in the CDM perturbations is available to form them. Which of the definitions is more appropriate requires further study outside of the scope of this work and we will use the more usual CDM definition henceforth.
The scaledependent reduction of the amplitude of the gravitational potential affects lensing. We plot the lensing potential in Fig. 4, showing that the suppression is very strong for large sound speeds and for large . The effect of lensing on the CMB was detected by Planck by two methods: the smoothing of the power spectrum at small scales and using the trispectrum; and it matches that predicted by CDM closely (although the powerspectrum method sees a 2 excess of lensing in the context of CDM) Ade:2015xua (). As we have already mentioned and show in Fig. 5, the CMB at decoupling is only mildly sensitive to DE sound speeds. It is the reduction of the CMB lensing effect in the late universe that breaks the degeneracy between the DE and CDM when .
Reference Bean:2003fb () proposed that the integrated SachsWolfe effect could be used to constrain the sound speed of dark energy and shows that WMAP data imply . Since the evolution of gravitational potentials at scales in our model is purely determined by the evolution of the cosmological background, the integrated SachsWolfe (ISW) effect on the CMB anisotropies also matches that of CDM for a lowenough sound speed and this does not provide an interesting constraint. Similarly, crosscorrelations of ISW with galaxy clustering are unlikely to constrain this model beyond some minimal sound speed Hu:2004yd (). For lower sound speeds, the above probes only test the expansion history.^{3}^{3}3As was argued in Ref. Bertschinger:2006aw (), the presence of gravitational slip changes the evolution of the potential at large scales. Thus the latetime ISW effect could still be used to constrain it, even for a fixed expansion history, and therefore test for modifications of gravity Saltas:2014dha (). The novelty of Planck with respect to WMAP in this context is that CMB lensing provides new constraints which push the limits on much further than largescale measurements.
The picture we have therefore built up is that, for small enough DE sound speeds , the dark energy perturbations behave as dust outside of their Jeans scale, making up for the removed CDM and giving essentially the same predictions as CDM. Inside the Jeans scale, the DE reacts and develops a pressure which arrests its collapse, allowing only the CDM to cluster. This reduces the growth rate at small scales when compared to large, introducing a new scale dependence in all the observables. In the limit , the dark energy behaves as dust at all scales, and therefore is completely degenerate with CDM, despite its equation of state (see e.g. the Angel Dust model of Lim:2010yk () or mimetic dark matter of Chamseddine:2013kea (); Chamseddine:2014vna ()).
As a result, the constraints that we obtain from the CMB are always an upper bound on the DE sound speed. Lateuniverse measurements will match the CDM results if they are performed at scales larger than and will differ if they are centered on smaller scales. For very low DE sound speeds, there is total degeneracy. The scaledependent growth rate resulting from the Jeans horizon allows us to alleviate some of the tension of CMB measurements when the sound speed is dialed up.
It is important to stress that at no point is the effective Newton’s constant, describing the response of the gravitational potential to the CDM, smaller than its standard value. Deep inside the Jeans scale, they are exactly equal. Thus we have achieved a reduction of the growth rate without introducing some sort of gravitylike repulsive interaction which would in all likelihood be pathological.
ii.3 Microscopic model
We have purposefully deemphasized the discussion of the precise model of dark energy which might produce such phenomenology as above. Essentially any DE/MG model which comprises a single degree of freedom and has an energymomentum tensor of perfectfluid form gives this phenomenology when the solution has the equation of state (2) and a constant sound speed for the perturbations around it. Specifying the model allows for a calculation of any nonlinear effects and for an estimate of the domain of validity of our linear model.
As an example, the class of kessence models ArmendarizPicon:2000dh () provides for such behavior. The action is given by
(13) 
The equation of state and sound speed by this dark energy are given by
(14) 
and given some initial conditions for a suitable can be found by integrating the above. An appropriate choice could in fact be
(15) 
whenever the sound speed is small and where is some sufficiently slowly varying potential, is a constant mass scale.
Another way of thinking about this model, which is completely equivalent at the level of this paper, is that we are introducing a subcomponent of the dark matter which exists in a condensate with a nonvanishing sound speed Sawicki:2013wja (). In such an interpretation, the acceleration is still being driven by a cosmological constant.
One should, however, be mindful when taking such scalarfield superfluid models literally: the behavior of perturbations when nonlinear deviates from standard fluids, since the scalar field cannot carry vorticity and therefore cannot virialize. The perturbations in the DE can become nonlinear when Batista:2013oca (). At the same time, for sound speeds the pressure support that the DE can provide is never quite enough to arrest the collapse and such a fluid would continue to collapse either until the breakdown of the effective description or would form a black hole.
Iii Results
For our analysis we have slightly modified the CAMB/CosmoMC public codes Lewis:1999bs (); Lewis:2002ah () to properly model the evolution of a perfect fluid with a timedependent equation of state (2) and a constant sound speed . Unless stated otherwise, we only let the parameters and vary, keeping the others fixed to their CDM Planck 2013 bestfit values Ade:2013zuv (). Also, we should mention that we scan the parameter space in terms of instead of simply , in order make it easier for the MCMC code to sample the subtle effects of the sound speed and recover the degeneracy for low . For the final combined analysis, we have freed all parameters.
Whenever possible, we have tried following the relevant parts of the analysis of Ade:2015xua (), although we were only able to use the Planck likelihoods from the 2013 data release Ade:2013zuv (). We first discuss the constraints from the different data sets in turn and then perform a combined analysis in Sec. III.3.
In order to provide a reasonable representation of the degeneracies, we are always including distance data together with each of the perturbationrelated data sets. Therefore, we have included the BAO measurements from CMASS and LOWZ of Ref. Anderson:2013zyy (), the 6DF measurement from Ref. Beutler:2011hx (), the MGS measurement from Ref. Ross:2014qpa () and the Union 2.1 SNe Ia catalog from Ref. Suzuki:2011hu (), all readily available in the CosmoMC code. We do not include any measurements of the Hubble constant , apart from a uniform prior .
The setup most similar to ours was previously investigated in dePutter:2010vy (), albeit the parametrizations for the equations of state of DE used were different. At the time, WMAP did not provide strong constraints on the DE sound speed. As we show here, this is no longer the case in the context of a fixed background expansion history.
iii.1 Constraints from full CMB
The Planck papers on cosmology Ade:2015xua () and on dark energy and modified gravity Ade:2015rim () demonstrated that CDM is a good fit to the data. Given the discussion in Sec. II.2, we expect to recover a perfect degeneracy between and when the sound speed is sufficiently low.
For this fit we are using both the CMB temperature and lowmultipole polarization power spectra from Planck including of course the lensing of the CMB. We have not included the lensing information extracted from the temperature trispectrum.
As expected, Fig. 5 shows that CMB anisotropies constrain only the sum whenever . For small admixtures of DE, much higher sound speeds are also allowed. The CMB does not show any preference for having this extra component, but it cannot rule out its existence either.
As we have previewed already in Sec. II.2, this constraint on actually comes mainly from CMB lensing, which takes place in the late universe. We have generated a perfect Planck forecast without lensing for CDM and our model and have found that the dark degeneracy cannot be broken by the CMB power spectrum for . The CMB at decoupling is not very sensitive to clustering properties of the dark matter.
iii.2 Weak lensing shear
As was noted by Planck Ade:2015xua (); Ade:2015rim (), the data from weak lensing shear are slightly incompatible with the Planck CDM analysis: the preferred amplitude of the lensing potential found by CFHTLeNS at angular scales is lower than that expected when the initial normalization of perturbations implied by the CMB is extended to smaller scales and evolved forward in CDM Heymans:2013fya (). If is kept constant at the Planck bestfit value, and the distance measurements are in the likelihood, the optimal solution for WL has a lower density fraction and a reduced amplitude for the gravitational potential sourced by a matter distribution with . Provided that the sound speed of our DE , which ensures that the data lie at scales outside of the DE Jeans horizon, the WL data cannot break the dark degeneracy.
However, a new, equally good, solution for the lensing data appears in the presence of a small amount of our DE () with a large sound speed, which also allows for a much larger value of the CDM density fraction . In such a scenario, the DE clusters only at largest scales and therefore the growth rate of the CDM is reduced according to Eq. (12). This reduces the amplitude of the lensing potential at late times despite the large . We have verified that the existence of this solution does not strongly depend on the nonlinear completion used for the matter power spectrum: just as Planck, for the fits in this paper we use Halofit with the parameters of Takahashi et al. Takahashi:2012em (). However, we have verified that the constraints are not biased by comparing them with the ones obtained using the original Halofit parameters Smith:2002dz () and even by switching off the nonlinear correction to the matter power spectrum completely. We have also used the correlation functions from the alternative CFHTLeNS analysis of Ref. Kilbinger:2012qz (), verifying that it would not significantly influence our results.
In principle, one could have also varied the initial fluctuation amplitude which would have given a much weaker constraint for , compatible with the CMB fits. However, the CMB measures the initial amplitude very well and therefore WL does not significantly correct the posterior for it. Thus the posteriors presented in Fig. 6 represent better the effective contribution of WL to the posterior of the combined data. We of course free all the parameters for the final combined fit.
iii.3 Combined constraints
For the combined analysis, we include all the CMB data in the Planck anisotropy power spectrum and the CFHTLeNS weak lensing shear. We free all the standard CDM parameters as well as and .
Like in previous cases we find that a dark degeneracy is recovered in the combined fits for , i.e. in the parameter range as determined by the WL measurement. This parameter range is equivalent to the concordance cosmology and offers no improvement in the combined fit.
However, there exists an alternative, superior solution already mentioned in Sec. III.2, which fits the combined data better than CDM with a while containing two extra parameters. In order to discuss the posterior of this solution, we impose a prior to eliminate the part of infinite logarithmic parameter space where our model is completely degenerate with CDM, cutting the chains at . We show this part of the parameter space in Fig. 7.
Given this cut, the combined data prefer together with an admixture of of dark energy with a sound speed of . The combined density fraction is consistent with the Planck 2013 best fit of . At small scales, the DE stops to cluster and recovers the preferred lensing density fraction.
We note that in this alternative solution, the amplitude of the CDM power spectrum is , which lies within 1 of the value as implied by clusters in Ref. Ade:2013lmv (). However, the power spectrum is quite modified at small scales and the predictions for nonlinear structures might not be related to the value of in the same manner as in CDM.
Iv Discussion and Conclusions
There is a weak but chronic tension between the CMB measurements of cosmological inhomogeneities and those obtained from the late universe, which consistently suggest that there is less power at late times than expected in CDM. There are many systematic effects that could be playing a role in biasing our interpretations of the lowredshift universe (baryonic physics, intrinsic alignment, badly modeled nonlinear physics), but one should ask whether such a phenomenology, if it were to persist, could be evidence of some sort of dynamics in the dark energy/gravity sector.
The usual approach to solve this is to change the geometry of the universe at late times (i.e. change ). However, it is difficult to create a large effect on perturbations without significantly altering distances. The constraints from BAOs and SNe are now tight enough that it is very difficult to create a large effect consistent with these background data and the CMB. Moreover, the CMB lensing and the WL are driven by the same physics and the geometrical kernels lie close together at lower redshifts, so that it is difficult to produce a significant discrepancy in the clustering seen by these two probes with the help of a purely timedependent effect.
Taking account of the fact that the data in tension (CMB and WL/clusters) probe different scales, we advocate an alternative approach. We keep the expansion history fixed and use the DE to make the CDM cluster differently at small and large scales. We have used a simple perfect fluid to model this effect. Yet this type of model is relatively poorly explored as a result of the focus on particular classes of parametrization of (CPL). We have instead exploited the dark degeneracy to freely vary the CDM density fraction while keeping fixed.
The sort of parametrization we have employed gives a nonnegligible contribution of dark energy at early times. This does not deform the CMB in an unacceptable manner, because our model strongly violates the quasistatic approximation typically used in the DE/MG calculations. Perfect fluids at scales beyond their sound horizons evolve as dust, irrespective of their equation of state. Thus the growth rate at largest scales is completely determined by and perturbation evolution in our model is degenerate with CDM at large enough scales. This is a known and generic result for any model with no anisotropic stress Bertschinger:2006aw () and exploited in the parametrization of Ref. Hu:2007pj ().
That being said, CMB lensing, detected in the Planck power spectrum at 10, puts a very strong upper limit on the DE sound speed. Since the CMB lensing is consistent with CDM, the DE must cluster as dust on the scales probed by it. Then there needs to be a rapid transition in the clustering properties, so that the CDM growth rate is reduced at the smaller scales probed by WL.
We stress that despite the fact that the CMB is a highredshift observable, it constrains our model mostly at low redshifts. In total, the combination the combination of CMB and WL data fit the model better than CDM with a . This solution has a with the presence of a small admixture of DE . This keeps the CMB anisotropies unchanged while reducing the amplitude of the gravitational potential inside the Jeans horizon and bringing the two data sets together.
Our improved solution has reduced , which is closer to the result reported by Planck in 2013 from clusters () Ade:2013lmv (). This decrement in power continues at small scales, which could in principle help alleviate some of the tension between Nbody simulations of CDM and the observations (e.g. the missingsatellites problem Klypin:1999uc () and Too Big to Fail BoylanKolchin:2011dk ()). It is interesting to ask whether the nonlinear behavior of CDM in this scaledependent model could be mapped from standard CDM Nbody simulation using a method such as that proposed in Mead:2014gia (). Since in our model at small scales gravity is not modified and there are no new screening effects, this might give a simple method for adjusting calibration in e.g. Halofit for this subclass of DE models and therefore to predict smallscale lensing and cluster formation with a better accuracy than we can currently.
We have not used constraints from redshiftspace distortions on the CDM growth rate . On one hand, we find that the uncertainty in the most precise measurements to date Samushia:2013yga (); Beutler:2013yhm (); Chuang:2013hya () is still high enough not to affect the posteriors significantly in the vicinity of the bestfit region of our scenario. More importantly, we are dealing with a model in which an evolving scaledependence of the matter power spectrum is a key feature. Interpreting the growthrate data in such a scenario is somewhat subtle since the growth rates are scaledependent Ballesteros:2008qk (); Nesseris:2015fqa () and we will give this question the consideration it deserves in a separate work.
We should stress that we have exploited a completely generic feature of modified gravity. The response of the gravitational potential to the CDM perturbations depends on scale in all models apart from CDM, whether because such a transition is already explicit in the quasistatic approximation (e.g. gravity where the Compton mass provides a scale Song:2006ej (), or see Ref. Motta:2013cwa ()) or because the quasistatic approximation fails beyond the Jeans scale (in the scenario discussed here; also see Ref. Sawicki:2015zya ()). This implies that a scaledependent modification of the matter power spectrum is completely generic in dynamical dark energy/modified gravity models and typically will take place in a range of modes corresponding to the Jeans/Comptonmass scale today and mode corresponding to the Jeans scale at the time that the DE/MG started contributing to the gravitational potential significantly.
Observations of largescale structure are performed not only around a particular redshift, but inside a particular range of scales probed by the survey. This scaledependence can make the interpretation of measurements complicated, if the measurements are reported assuming a CDMlike scaleindependent behavior, instead of a rawer form closer to the observation.
The phenomenology of such “cold dark energy” Sapone:2010uy () models featuring low sound speeds as proposed here and exploiting a scaledependent phenomenology, but in a more general context of effective field theory of dark energy Gubitosi:2012hu (); Bloomfield:2012ff (); Gleyzes:2013ooa (); Gleyzes:2014qga (); Bellini:2014fua () remains largely unexplored, but is necessary if we are to build a full understanding of the constraints data placed on the properties of the dark sector.
Acknowledgements.
We are grateful to M.A. Amin, K.C. Chan, C.H. Chuang, P.S. Corasaniti, J. Gleyzes, B. Hu, A. Mead, F. Piazza, V. Valkenburg and F. Vernizzi. M.K. and S.N. acknowledge funding by the Swiss National Science Foundation. I.S. is supported by the Maria SkłodowskaCurie IntraEuropean Fellowship Project “DRKFRCS”. The numerical computations for our analysis were performed on the Baobab cluster at the University of Geneva. The development of Planck was supported by: ESA; CNES and CNRS/INSU IN2P3INP (France); ASI, CNR, and INAF (Italy); NASA and DoE (USA); STFC and UKSA (UK); CSIC, MICINN and JA (Spain); Tekes, AoF and CSC (Finland); DLR and MPG (Germany); CSA (Canada); DTU Space (Denmark); SER/SSO (Switzerland); RCN (Norway); SFI (Ireland); FCT/MCTES (Portugal). A description of the Planck Collaboration and a list of its members, including the technical or scientific activities in which they have been involved, can be found at http://www.cosmos.esa.int/web/planck/planckcollaboration.References
 (1) Planck Collaboration, P. Ade et al., “Planck 2013 results. XVI. Cosmological parameters,” Astron.Astrophys. 571 (2014) A16, arXiv:1303.5076 [astroph.CO].
 (2) Planck Collaboration, P. Ade et al., “Planck 2015 results. XIII. Cosmological parameters,” arXiv:1502.01589 [astroph.CO].
 (3) Planck Collaboration, P. Ade et al., “Planck 2013 results. XX. Cosmology from Sunyaev–Zeldovich cluster counts,” Astron.Astrophys. 571 (2014) A20, arXiv:1303.5080 [astroph.CO].
 (4) Planck Collaboration, P. Ade et al., “Planck 2015 results. XXIV. Cosmology from SunyaevZeldovich cluster counts,” arXiv:1502.01597 [astroph.CO].
 (5) M. Wyman, D. H. Rudd, R. A. Vanderveld, and W. Hu, “Neutrinos Help Reconcile Planck Measurements with the Local Universe,” Phys.Rev.Lett. 112 no. 5, (2014) 051302, arXiv:1307.7715 [astroph.CO].
 (6) R. A. Battye, T. Charnock, and A. Moss, “Tension between the power spectrum of density perturbations measured on large and small scales,” arXiv:1409.2769 [astroph.CO].
 (7) Planck Collaboration, P. Ade et al., “Planck 2015 results. XIV. Dark energy and modified gravity,” arXiv:1502.01590 [astroph.CO].
 (8) F. Piazza, H. Steigerwald, and C. Marinoni, “Phenomenology of dark energy: exploring the space of theories with future redshift surveys,” JCAP 1405 (2014) 043, arXiv:1312.6111 [astroph.CO].
 (9) S. Nesseris, “Matter density perturbations in modified gravity models with arbitrary coupling between matter and geometry,” Phys.Rev. D79 (2009) 044015, arXiv:0811.4292 [astroph].
 (10) S. Nesseris and A. Mazumdar, “Newton’s constant in theories of gravity and constraints from BBN,” Phys.Rev. D79 (2009) 104006, arXiv:0902.1185 [astroph.CO].
 (11) S. Basilakos, S. Nesseris, and L. Perivolaropoulos, “Observational constraints on viable f(R) parametrizations with geometrical and dynamical probes,” Phys.Rev. D87 no. 12, (2013) 123529, arXiv:1302.6051 [astroph.CO].
 (12) S. Nesseris, S. Basilakos, E. Saridakis, and L. Perivolaropoulos, “Viable models are practically indistinguishable from CDM,” Phys.Rev. D88 (2013) 103010, arXiv:1308.6142 [astroph.CO].
 (13) S. Nesseris, C. Blake, T. Davis, and D. Parkinson, “The WiggleZ Dark Energy Survey: constraining the evolution of Newton’s constant using the growth rate of structure,” JCAP 1107 (2011) 037, arXiv:1107.3659 [astroph.CO].
 (14) S. Nesseris, “Can the degeneracies in the gravity sector be broken?,” Phys.Rev. D88 (2013) 123003, arXiv:1309.1055 [astroph.CO].
 (15) S. Nesseris and S. Tsujikawa, “Cosmological perturbations and observational constraints on nonlocal massive gravity,” Phys.Rev. D90 no. 2, (2014) 024070, arXiv:1402.4613 [astroph.CO].
 (16) L. Perenon, F. Piazza, C. Marinoni, and L. Hui, “Phenomenology of dark energy: general features of largescale perturbations,” arXiv:1506.03047 [astroph.CO].
 (17) M. Kunz, “The dark degeneracy: On the number and nature of dark components,” Phys.Rev. D80 (2009) 123001, arXiv:astroph/0702615 [astroph].
 (18) L. Amendola, M. Kunz, M. Motta, I. D. Saltas, and I. Sawicki, “Observables and unobservables in dark energy cosmologies,” Phys.Rev. D87 no. 2, (2013) 023501, arXiv:1210.0439 [astroph.CO].
 (19) M. Motta, I. Sawicki, I. D. Saltas, L. Amendola, and M. Kunz, “Probing Dark Energy through Scale Dependence,” Phys.Rev. D88 no. 12, (2013) 124035, arXiv:1305.0008 [astroph.CO].
 (20) E. Bellini and I. Sawicki, “Maximal freedom at minimum cost: linear largescale structure in general modifications of gravity,” JCAP 1407 (2014) 050, arXiv:1404.3713 [astroph.CO].
 (21) S. Dubovsky, T. Gregoire, A. Nicolis, and R. Rattazzi, “Null energy condition and superluminal propagation,” JHEP 0603 (2006) 025, arXiv:hepth/0512260 [hepth].
 (22) I. Sawicki and A. Vikman, “Hidden Negative Energies in Strongly Accelerated Universes,” Phys.Rev. D87 no. 6, (2013) 067301, arXiv:1209.2961 [astroph.CO].
 (23) M. Chevallier and D. Polarski, “Accelerating universes with scaling dark matter,” Int.J.Mod.Phys. D10 (2001) 213–224, arXiv:grqc/0009008 [grqc].
 (24) E. V. Linder, “Exploring the expansion history of the universe,” Phys.Rev.Lett. 90 (2003) 091301, arXiv:astroph/0208512 [astroph].
 (25) B. A. Bassett, M. Kunz, J. Silk, and C. Ungarelli, “A Late time transition in the cosmic dark energy?,” Mon.Not.Roy.Astron.Soc. 336 (2002) 1217–1222, arXiv:astroph/0203383 [astroph].
 (26) V. Pettorino, L. Amendola, and C. Wetterich, “How early is early dark energy?,” Phys.Rev. D87 (2013) 083009, arXiv:1301.5279 [astroph.CO].
 (27) L. Amendola, M. Baldi, and C. Wetterich, “Quintessence cosmologies with a growing matter component,” Phys.Rev. D78 (2008) 023015, arXiv:0706.3064 [astroph].
 (28) A. Y. Kamenshchik, U. Moschella, and V. Pasquier, “An Alternative to quintessence,” Phys.Lett. B511 (2001) 265–268, arXiv:grqc/0103004 [grqc].
 (29) H. Sandvik, M. Tegmark, M. Zaldarriaga, and I. Waga, “The end of unified dark matter?,” Phys.Rev. D69 (2004) 123524, arXiv:astroph/0212114 [astroph].
 (30) Y. Wang, D. Wands, L. Xu, J. DeSantiago, and A. Hojjati, “Cosmological constraints on a decomposed Chaplygin gas,” Phys.Rev. D87 no. 8, (2013) 083503, arXiv:1301.5315 [astroph.CO].
 (31) W. Hu, “Structure formation with generalized dark matter,” Astrophys.J. 506 (1998) 485–494, arXiv:astroph/9801234 [astroph].
 (32) R. de Putter, D. Huterer, and E. V. Linder, “Measuring the Speed of Dark: Detecting Dark Energy Perturbations,” Phys.Rev. D81 (2010) 103513, arXiv:1002.1311 [astroph.CO].
 (33) D. Sapone, M. Kunz, and M. Kunz, “Fingerprinting Dark Energy,” Phys.Rev. D80 (2009) 083519, arXiv:0909.0007 [astroph.CO].
 (34) M. Malquarti and A. R. Liddle, “Evolution of large scale perturbations in quintessence models,” Phys.Rev. D66 (2002) 123506, arXiv:astroph/0208562 [astroph].
 (35) I. Sawicki and E. Bellini, “Limits of QuasiStatic Approximation in ModifiedGravity Cosmologies,” arXiv:1503.06831 [astroph.CO].
 (36) E. Bertschinger, “On the Growth of Perturbations as a Test of Dark Energy,” Astrophys.J. 648 (2006) 797–806, arXiv:astroph/0604485 [astroph].
 (37) R. Bean and O. Dore, “Probing dark energy perturbations: The Dark energy equation of state and speed of sound as measured by WMAP,” Phys.Rev. D69 (2004) 083503, arXiv:astroph/0307100 [astroph].
 (38) W. Hu and R. Scranton, “Measuring dark energy clustering with CMBgalaxy correlations,” Phys.Rev. D70 (2004) 123002, arXiv:astroph/0408456 [astroph].
 (39) I. D. Saltas, I. Sawicki, L. Amendola, and M. Kunz, “Anisotropic Stress as a Signature of Nonstandard Propagation of Gravitational Waves,” Phys.Rev.Lett. 113 no. 19, (2014) 191101, arXiv:1406.7139 [astroph.CO].
 (40) E. A. Lim, I. Sawicki, and A. Vikman, “Dust of Dark Energy,” JCAP 1005 (2010) 012, arXiv:1003.5751 [astroph.CO].
 (41) A. H. Chamseddine and V. Mukhanov, “Mimetic Dark Matter,” JHEP 1311 (2013) 135, arXiv:1308.5410 [astroph.CO].
 (42) A. H. Chamseddine, V. Mukhanov, and A. Vikman, “Cosmology with Mimetic Matter,” JCAP 1406 (2014) 017, arXiv:1403.3961 [astroph.CO].
 (43) C. ArmendarizPicon, V. F. Mukhanov, and P. J. Steinhardt, “A Dynamical solution to the problem of a small cosmological constant and late time cosmic acceleration,” Phys.Rev.Lett. 85 (2000) 4438–4441, arXiv:astroph/0004134 [astroph].
 (44) I. Sawicki, V. Marra, and W. Valkenburg, “Seeding supermassive black holes with a nonvortical darkmatter subcomponent,” Phys.Rev. D88 (2013) 083520, arXiv:1307.6150 [astroph.CO].
 (45) R. Batista and F. Pace, “Structure formation in inhomogeneous Early Dark Energy models,” JCAP 1306 (2013) 044, arXiv:1303.0414 [astroph.CO].
 (46) A. Lewis, A. Challinor, and A. Lasenby, “Efficient Computation of CMB anisotropies in closed FRW models,” Astrophys. J. 538 (2000) 473–476, astroph/9911177.
 (47) A. Lewis and S. Bridle, “Cosmological parameters from CMB and other data: a Monte Carlo approach,” Phys. Rev. D66 (2002) 103511, astroph/0205436.
 (48) L. Anderson et al., “The clustering of galaxies in the SDSSIII Baryon Oscillation Spectroscopic Survey: Baryon Acoustic Oscillations in the Data Release 10 and 11 galaxy samples,” Mon.Not.Roy.Astron.Soc. 441 (Jun, 2014) 24–62, arXiv:1312.4877 [astroph.CO].
 (49) F. Beutler, C. Blake, M. Colless, D. H. Jones, L. StaveleySmith, et al., “The 6dF Galaxy Survey: Baryon Acoustic Oscillations and the Local Hubble Constant,” Mon.Not.Roy.Astron.Soc. 416 (2011) 3017–3032, arXiv:1106.3366 [astroph.CO].
 (50) A. J. Ross, L. Samushia, C. Howlett, W. J. Percival, A. Burden, et al., “The Clustering of the SDSS DR7 Main Galaxy Sample I: A 4 per cent Distance Measure at z=0.15,” Mon.Not.Roy.Astron.Soc. 449 no. 1, (2015) 835–847, arXiv:1409.3242 [astroph.CO].
 (51) N. Suzuki, D. Rubin, C. Lidman, G. Aldering, R. Amanullah, et al., “The Hubble Space Telescope Cluster Supernova Survey: V. Improving the Dark Energy Constraints Above z>1 and Building an EarlyTypeHosted Supernova Sample,” Astrophys.J. 746 (2012) 85, arXiv:1105.3470 [astroph.CO].
 (52) C. Heymans, E. Grocutt, A. Heavens, M. Kilbinger, T. D. Kitching, et al., “CFHTLenS tomographic weak lensing cosmological parameter constraints: Mitigating the impact of intrinsic galaxy alignments,” Mon.Not.Roy.Astron.Soc. 432 (2013) 2433, arXiv:1303.1808 [astroph.CO].
 (53) R. Takahashi, M. Sato, T. Nishimichi, A. Taruya, and M. Oguri, “Revising the Halofit Model for the Nonlinear Matter Power Spectrum,” Astrophys.J. 761 (2012) 152, arXiv:1208.2701 [astroph.CO].
 (54) VIRGO Consortium Collaboration, R. Smith et al., “Stable clustering, the halo model and nonlinear cosmological power spectra,” Mon.Not.Roy.Astron.Soc. 341 (2003) 1311, arXiv:astroph/0207664 [astroph].
 (55) M. Kilbinger, L. Fu, C. Heymans, F. Simpson, J. Benjamin, et al., “CFHTLenS: Combined probe cosmological model comparison using 2D weak gravitational lensing,” Monthly Notices of the Royal Astronomical Society 430 no. 3, (2013) 2200–2220, arXiv:1212.3338 [astroph.CO].
 (56) W. Hu and I. Sawicki, “A Parameterized PostFriedmann Framework for Modified Gravity,” Phys.Rev. D76 (2007) 104043, arXiv:0708.1190 [astroph].
 (57) A. A. Klypin, A. V. Kravtsov, O. Valenzuela, and F. Prada, “Where are the missing Galactic satellites?,” Astrophys.J. 522 (1999) 82–92, arXiv:astroph/9901240 [astroph].
 (58) M. BoylanKolchin, J. S. Bullock, and M. Kaplinghat, “The Milky Way’s bright satellites as an apparent failure of LCDM,” Mon.Not.Roy.Astron.Soc. 422 (2012) 1203–1218, arXiv:1111.2048 [astroph.CO].
 (59) A. Mead, J. Peacock, L. Lombriser, and B. Li, “Rapid simulation rescaling from standard to modified gravity models,” arXiv:1412.5195 [astroph.CO].
 (60) L. Samushia, B. A. Reid, M. White, W. J. Percival, A. J. Cuesta, et al., “The clustering of galaxies in the SDSSIII Baryon Oscillation Spectroscopic Survey: measuring growth rate and geometry with anisotropic clustering,” Mon.Not.Roy.Astron.Soc. 439 no. 4, (2014) 3504–3519, arXiv:1312.4899 [astroph.CO].
 (61) BOSS Collaboration, F. Beutler et al., “The clustering of galaxies in the SDSSIII Baryon Oscillation Spectroscopic Survey: Testing gravity with redshiftspace distortions using the power spectrum multipoles,” Mon.Not.Roy.Astron.Soc. 443 no. 2, (2014) 1065–1089, arXiv:1312.4611 [astroph.CO].
 (62) C.H. Chuang, F. Prada, A. J. Cuesta, D. J. Eisenstein, E. Kazin, et al., “The clustering of galaxies in the SDSSIII Baryon Oscillation Spectroscopic Survey: singleprobe measurements and the strong power of normalized growth rate on constraining dark energy,” Mon.Not.Roy.Astron.Soc. 433 (2013) 3559, arXiv:1303.4486 [astroph.CO].
 (63) G. Ballesteros and A. Riotto, “Parameterizing the Effect of Dark Energy Perturbations on the Growth of Structures,” Phys.Lett. B668 (2008) 171–176, arXiv:0807.3343 [astroph].
 (64) S. Nesseris and D. Sapone, “On the accuracy of the growth index in the presence of dark energy perturbations,” arXiv:1505.06601 [astroph.CO].
 (65) Y.S. Song, W. Hu, and I. Sawicki, “The Large Scale Structure of f(R) Gravity,” Phys.Rev. D75 (2007) 044004, arXiv:astroph/0610532 [astroph].
 (66) D. Sapone, M. Kunz, and L. Amendola, “Fingerprinting Dark Energy II: weak lensing and galaxy clustering tests,” Phys.Rev. D82 (2010) 103535, arXiv:1007.2188 [astroph.CO].
 (67) G. Gubitosi, F. Piazza, and F. Vernizzi, “The Effective Field Theory of Dark Energy,” JCAP 1302 (2013) 032, arXiv:1210.0201 [hepth].
 (68) J. K. Bloomfield, E. E. Flanagan, M. Park, and S. Watson, “Dark energy or modified gravity? An effective field theory approach,” JCAP 1308 (2013) 010, arXiv:1211.7054 [astroph.CO].
 (69) J. Gleyzes, D. Langlois, F. Piazza, and F. Vernizzi, “Essential Building Blocks of Dark Energy,” JCAP 1308 (2013) 025, arXiv:1304.4840 [hepth].
 (70) J. Gleyzes, D. Langlois, F. Piazza, and F. Vernizzi, “Exploring gravitational theories beyond Horndeski,” JCAP 1502 (2015) 018, arXiv:1408.1952 [astroph.CO].