
The recent discovery of a $\gamma$ray counterpart to a gravitational wave
(GW) event has put extremely stringent constraints on the speed of
gravitational waves at the present epoch. In turn, these constraints place
strong theoretical pressure on potential modifications of gravity, essentially
allowing only the conformal sector to be active in the present Universe. In
this paper, we show that direct detection of gravitational waves from optically
identified sources can also measure or constrain the conformal sector of
modified gravity models through the time variation of the Planck mass. As a
first rough estimate, we find that the LISA satellite can measure the
dimensionless time variation of the Planck mass (the socalled parameter
$\alpha_M$) at redshift around 1.5 with an error of about 0.03 to 0.13,
depending on the assumptions concerning future observations. Stronger
constraints can be achieved once reliable distance indicators at $z>2$ are
developed, or with GW detectors that extend the capabilities of LISA, like the
proposed Big Bang Observer. We emphasize that, just like the constraints on the
gravitational speed, the bound on $\alpha_M$ is independent of the cosmological
model.

The detection of an electromagnetic counterpart (GRB 170817A) to the
gravitational wave signal (GW170817) from the merger of two neutron stars opens
a completely new arena for testing theories of gravity. We show that this
measurement allows us to place stringent constraints on general scalartensor
and vectortensor theories, while allowing us to place an independent bound on
the graviton mass in bimetric theories of gravity. These constraints severely
reduce the viable range of cosmological models that have been proposed as
alternatives to general relativistic cosmology.

We present the public version of hi_class (www.hiclasscode.net), an
extension of the Boltzmann code CLASS to a broad ensemble of modifications to
general relativity. In particular, hi_class can calculate predictions for
models based on Horndeski's theory, which is the most general scalartensor
theory described by secondorder equations of motion and encompasses any
perfectfluid dark energy, quintessence, BransDicke, $f(R)$ and covariant
Galileon models. hi_class has been thoroughly tested and can be readily used to
understand the impact of alternative theories of gravity on linear structure
formation as well as for cosmological parameter extraction.

We use largescale cosmological observations to place constraints on the
darkmatter pressure, sound speed and viscosity, and infer a limit on the mass
of warmdarkmatter particles. Measurements of the cosmic microwave background
(CMB) anisotropies constrain the equation of state and sound speed of the dark
matter at last scattering at the per mille level. Since the redshifting of
collisionless particles universally implies that these quantities scale like
$a^{2}$ absent shell crossing, we infer that today $w_{\rm (DM)}< 10^{10.0}$,
$c_{\rm s,(DM)}^2 < 10^{10.7}$ and $c_{\rm vis, (DM)}^{2} < 10^{10.3}$ at the
$99\%$ confidence level. This very general bound can be translated to
modeldependent constraints on darkmatter models: for warm dark matter these
constraints imply $m> 70$ eV, assuming it decoupled while relativistic around
the same time as the neutrinos; for a cold relic, we show that $m>100$ eV. We
separately constrain the properties of the DM fluid on linear scales at late
times, and find upper bounds $c_{\rm s, (DM)}^2<10^{5.9}$, $c_{\rm vis,
(DM)}^{2} < 10^{5.7}$, with no detection of nondust properties for the DM.

Euclid is a European Space Agency medium class mission selected for launch in
2020 within the Cosmic Vision 2015 2025 program. The main goal of Euclid is to
understand the origin of the accelerated expansion of the universe. Euclid will
explore the expansion history of the universe and the evolution of cosmic
structures by measuring shapes and redshifts of galaxies as well as the
distribution of clusters of galaxies over a large fraction of the sky. Although
the main driver for Euclid is the nature of dark energy, Euclid science covers
a vast range of topics, from cosmology to galaxy evolution to planetary
research. In this review we focus on cosmology and fundamental physics, with a
strong emphasis on science beyond the current standard models. We discuss five
broad topics: dark energy and modified gravity, dark matter, initial
conditions, basic assumptions and questions of methodology in the data
analysis. This review has been planned and carried out within Euclid's Theory
Working Group and is meant to provide a guide to the scientific themes that
will underlie the activity of the group during the preparation of the Euclid
mission.

We point out that the observed time delay between the detection of the signal
at the Hanford and Livingston LIGO sites from the gravitational wave event
GW150914 places an upper bound on the speed of propagation of gravitational
waves, $c_{gw}\lesssim 1.7$ in the units of speed of light. Combined with the
lower bound from the absence of gravitational Cherenkov losses by cosmic rays
that rules out most of subluminal velocities, this gives a modelindependent
doublesided constraint $1\lesssim c_{gw}\lesssim 1.7$. We compare this result
to modelspecific constraints from pulsar timing and cosmology.

Despite its continued observational successes, there is a persistent (and
growing) interest in extending cosmology beyond the standard model,
$\Lambda$CDM. This is motivated by a range of apparently serious theoretical
issues, involving such questions as the cosmological constant problem, the
particle nature of dark matter, the validity of general relativity on large
scales, the existence of anomalies in the CMB and on small scales, and the
predictivity and testability of the inflationary paradigm. In this paper, we
summarize the current status of $\Lambda$CDM as a physical theory, and review
investigations into possible alternatives along a number of different lines,
with a particular focus on highlighting the most promising directions. While
the fundamental problems are proving reluctant to yield, the study of
alternative cosmologies has led to considerable progress, with much more to
come if hopes about forthcoming highprecision observations and new theoretical
ideas are fulfilled.

We investigate the limits of applicability of the quasistatic approximation
in cosmologies featuring general models of dark energy or modified gravity. We
show that, at best, the quasistatic approximation breaks down outside of the
sound horizon of the darkenergy, rather than the cosmological horizon as is
frequently assumed. When the sound speed of dark energy is significantly below
that of light, the quasistatic limit is only valid in a limited range of
observable scales and this must be taken into account when computing effects on
observations in such models. As an order of magnitude estimate, in the analysis
of data from today's weaklensing and peculiarvelocity surveys, dark energy
can be modelled as quasistatic only if the sound speed is larger than order 1%
of that of light. In upcoming surveys, such as Euclid, it should only be used
when the sound speed exceeds around 10% of the speed of light. In the analysis
of the cosmic microwave background, the quasistatic limit should never be used
for the integrated SachsWolf effect and for lensing only when the sound speed
exceeds 10% of the speed of light.

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 $\Lambda$CDM,
at $\Delta\chi^2=2.4$. Moreover, this improved solution has $\sigma_8=0.79 \pm
0.02$, consistent with the value implied by cluster counts.

A perfect irrotational fluid with the equation of state of dust, Irrotational
Dark Matter (IDM), is incapable of virializing and instead forms a
cosmoskeleton of filaments with supermassive black holes at the joints. This
stark difference from the standard cold dark matter (CDM) scenario arises
because IDM must exhibit potential flow at all times, preventing shellcrossing
from occurring. This scenario is applicable to general nonoscillating
scalarfield theories with a small sound speed. Our model of combined IDM and
CDM components thereby provides a solution to the problem of forming the
observed billionsolarmass black holes at redshifts of six and higher. In
particular, as a result of the reduced vortical flow, the growth of the black
holes is expected to be more rapid at later times as compared to the standard
scenario.

There is an approximately 9% discrepancy, corresponding to 2.4sigma, between
two independent constraints on the expansion rate of the universe: one
indirectly arising from the cosmic microwave background and baryon acoustic
oscillations, and one more directly obtained from local measurements of the
relation between redshifts and distances to sources. We argue that by taking
into account the local gravitational potential at the position of the observer
this tension  strengthened by the recent Planck results  is partially
relieved and the concordance of the standard model of cosmology increased. We
estimate that measurements of the local Hubble constant are subject to a cosmic
variance of about 2.4% (limiting the local sample to redshifts z>0.010) or 1.3%
(limiting it to z>0.023), a more significant correction than that taken into
account already. Nonetheless, we show that one would need a very rare
fluctuation to fully explain the offset in the Hubble rates. If this tension is
further strengthened, a cosmology beyond the standard model may prove
necessary.

We point out that theories of cosmological acceleration which have equation
of state, w, such that 1+w is small but positive may still secretly violate the
null energy condition. This violation implies the existence of observers for
whom the background has infinitely negative energy densities, despite the fact
that the perturbations are free of ghosts and gradient instabilities.

We present a new prescription for analysing cosmological perturbations in a
moregeneral class of scalarfield darkenergy models where the energymomentum
tensor has an imperfectfluid form. This class includes BransDicke models,
f(R) gravity, theories with kinetic gravity braiding and generalised galileons.
We employ the intuitive language of fluids, allowing us to explicitly maintain
a dependence on physical and potentially measurable properties. We demonstrate
that hydrodynamics is not always a valid description for describing
cosmological perturbations in general scalarfield theories and present a
consistent alternative that nonetheless utilises the fluid language. We apply
this approach explicitly to a worked example: kessence nonminimally coupled
to gravity. This is the simplest case which captures the essential new features
of these imperfectfluid models. We demonstrate the generic existence of a new
scale separating regimes where the fluid is perfect and imperfect. We obtain
the equations for the evolution of darkenergy density perturbations in both
these regimes. The model also features two other known scales: the Compton
scale related to the breaking of shift symmetry and the Jeans scale which we
show is determined by the speed of propagation of small scalarfield
perturbations, i.e. causality, as opposed to the frequently used definition of
the ratio of the pressure and energydensity perturbations.

Theories of gravity other than general relativity (GR) can explain the
observed cosmic acceleration without a cosmological constant. One such class of
theories of gravity is f(R). Metric f(R) theories have been proven to be
equivalent to BransDicke (BD) scalartensor gravity without a kinetic term.
Using this equivalence and a 3+1 decomposition of the theory it has been shown
that metric f(R) gravity admits a wellposed initial value problem. However, it
has not been proven that the 3+1 evolution equations of metric f(R) gravity
preserve the (hamiltonian and momentum) constraints. In this paper we show that
this is indeed the case. In addition, we show that the mathematical form of the
constraint propagation equations in BDequilavent f(R) gravity and in f(R)
gravity in both the Jordan and Einstein frames, is exactly the same as in the
standard ADM 3+1 decomposition of GR. Finally, we point out that current
numerical relativity codes can incorporate the 3+1 evolution equations of
metric f(R) gravity by modifying the stressenergy tensor and adding an
additional scalar field evolution equation. We hope that this work will serve
as a starting point for relativists to develop fully dynamical codes for valid
f(R) models.

We introduce a large class of scalartensor models with interactions
containing the second derivatives of the scalar field but not leading to
additional degrees of freedom. These models exhibit peculiar features, such as
an essential mixing of scalar and tensor kinetic terms, which we have named
kinetic braiding. This braiding causes the scalar stress tensor to deviate from
the perfectfluid form. Cosmology in these models possesses a rich
phenomenology, even in the limit where the scalar is an exact Goldstone boson.
Generically, there are attractor solutions where the scalar monitors the
behaviour of external matter. Because of the kinetic braiding, the position of
the attractor depends both on the form of the Lagrangian and on the external
energy density. The latetime asymptotic of these cosmologies is a de Sitter
state. The scalar can exhibit phantom behaviour and is able to cross the
phantom divide with neither ghosts nor gradient instabilities. These features
provide a new class of models for Dark Energy. As an example, we study in
detail a simple oneparameter model. The possible observational signatures of
this model include a sizeable Early Dark Energy and a specific equation of
state evolving into the final deSitter state from a healthy phantom regime.

We introduce a novel class of field theories where energy always flows along
timelike geodesics, mimicking in that respect dust, yet which possess nonzero
pressure. This theory comprises two scalar fields, one of which is a Lagrange
multiplier enforcing a constraint between the other's field value and
derivative. We show that this system possesses no wavelike modes but retains a
single dynamical degree of freedom. Thus, the sound speed is always identically
zero on all backgrounds. In particular, cosmological perturbations reproduce
the standard behaviour for hydrodynamics with vanishing sound speed. Using all
these properties we propose a model unifying Dark Matter and Dark Energy in a
single degree of freedom. In a certain limit this model exactly reproduces the
evolution history of LambdaCDM, while deviations away from the standard
expansion history produce a potentially measurable difference in the evolution
of structure.

We propose a cosmological scenario based on the assumption that the Standard
Model possesses a large number of copies. It is demonstrated that baryons in
the hidden copies of the standard model can naturally account for the dark
matter. The right abundance of the hiddensector baryons and the correct
spectrum of density perturbations are simultaneously generated during modulated
reheating. We show that for the natural values of inflaton coupling constants,
dictated by unitarity, the darkmatter abundance is predicted to be
proportional to the ratio of observed cosmological parameters: the square of
the amplitude of cosmological perturbations and the baryontophoton number
ratio.

We develop a parameterized postFriedmann (PPF) framework which describes
three regimes of modified gravity models that accelerate the expansion without
dark energy. On large scales, the evolution of scalar metric and density
perturbations must be compatible with the expansion history defined by distance
measures. On intermediate scales in the linear regime, they form a
scalartensor theory with a modified Poisson equation. On small scales in dark
matter halos such as our own galaxy, modifications must be suppressed in order
to satisfy stringent local tests of general relativity. We describe these
regimes with three free functions and two parameters: the relationship between
the two metric fluctuations, the large and intermediate scale relationships to
density fluctuations and the two scales of the transitions between the regimes.
We also clarify the formal equivalence of modified gravity and generalized dark
energy. The PPF description of linear fluctuation in f(R) modified action and
the DvaliGabadadzePorrati braneworld models show excellent agreement with
explicit calculations. Lacking cosmological simulations of these models, our
nonlinear halomodel description remains an ansatz but one that enables
wellmotivated consistency tests of general relativity. The required
suppression of modifications within dark matter halos suggests that the linear
and weakly nonlinear regimes are better suited for making complementary test
of general relativity than the deeply nonlinear regime.

We reconcile seemingly conflicting statements in the literature about the
behavior of cosmological solutions in modified theories of gravity where the
EinsteinHilbert Lagrangian for gravity is modified by the addition of a
function of the Ricci scalar, f(R). Using the example of f(R) = +/\mu^4/R we
show that only such choices of f(R) where d^2f/dR^2>0 have stable
highcurvature limits and wellbehaved cosmological solutions with a proper era
of matter domination. The remaining models enter a phase dominated by both
matter and scalar kinetic energy where the scalar curvature remains low.

We study a class of metricvariation f(R) models that accelerates the
expansion without a cosmological constant and satisfies both cosmological and
solarsystem tests in the smallfield limit of the parameter space.
Solarsystem tests alone place only weak bounds on these models, since the
additional scalar degree of freedom is locked to the highcurvature
generalrelativistic prediction across more than 25 orders of magnitude in
density, out through the solar corona. This agreement requires that the
galactic halo be of sufficient extent to maintain the galaxy at high curvature
in the presence of the lowcurvature cosmological background. If the galactic
halo and local environment in f(R) models do not have substantially deeper
potentials than expected in LCDM, then cosmological field amplitudes f_R >
10^{6} will cause the galactic interior to evolve to low curvature during the
acceleration epoch. Viability of largedeviation models therefore rests on the
structure and evolution of the galactic halo, requiring cosmological
simulations of f(R) models, and not directly on solarsystem tests. Even small
deviations that conservatively satisfy both galactic and solarsystem
constraints can still be tested by future, percentlevel measurements of the
linear power spectrum, while they remain undetectable to cosmologicaldistance
measures. Although we illustrate these effects in a specific class of models,
the requirements on f(R) are phrased in a nearly modelindependent manner.

The selfaccelerating braneworld model (DGP) can be tested from measurements
of the expansion history of the universe and the formation of structure.
Current constraints on the expansion history from supernova luminosity
distances, the CMB, and the Hubble constant exclude the simplest flat DGP model
at about 3sigma. The bestfit open DGP model is, however, only a marginally
poorer fit to the data than flat LCDM. Its substantially different expansion
history raises structure formation challenges for the model. A darkenergy
model with the same expansion history would predict a highly significant
discrepancy with the baryon oscillation measurement due the high Hubble
constant required and a large enhancement of CMB anisotropies at the lowest
multipoles due to the ISW effect. For the DGP model to satisfy these
constraints new gravitational phenomena would have to appear at the nonlinear
and crossover scales respectively. A prediction of the DGP expansion history
in a region where the phenomenology is well understood is that highredshift
galaxies should be substantially correlated with the CMB through the ISW
effect. This correlation should provide a sharp test of the DGP model in the
future.

We develop a scaling ansatz for the master equation in Dvali, Gabadadze,
Porrati cosmologies, which allows us to solve the equations of motion for
perturbations off the brane during periods when the onbrane evolution is
scalefree. This allows us to understand the behavior of the gravitational
potentials outside the horizon at high redshifts and close to the horizon
today. We confirm that the results of Koyama and Maartens are valid at scales
relevant for observations such as galaxyISW correlation. At larger scales,
there is an additional suppression of the potential which reduces the growth
rate even further and would strengthen the ISW effect.

We study the evolution of linear cosmological perturbations in f(R) models of
accelerated expansion in the physical frame where the gravitational dynamics
are fourth order and the matter is minimally coupled. These models predict a
rich and testable set of linear phenomena. For each expansion history, fixed
empirically by cosmological distance measures, there exists two branches of
f(R) solutions that are parameterized by B propto d^2 f/dR^2. For B<0, which
include most of the models previously considered, there is a shorttimescale
instability at high curvature that spoils agreement with high redshift
cosmological observables. For the stable B>0 branch, f(R) models can reduce the
largeangle CMB anisotropy, alter the shape of the linear matter power
spectrum, and qualitatively change the correlations between the CMB and galaxy
surveys. All of these phenomena are accessible with current and future data and
provide stringent tests of general relativity on cosmological scales.

One way to account for the acceleration of the universe is to modify general
relativity, rather than introducing dark energy. Typically, such modifications
introduce new degrees of freedom. It is interesting to consider models with no
new degrees of freedom, but with a modified dependence on the conventional
energymomentum tensor; the Palatini formulation of $f(R)$ theories is one
example. Such theories offer an interesting testing ground for investigations
of cosmological modified gravity. In this paper we study the evolution of
structure in these ``modifiedsource gravity'' theories. In the linear regime,
density perturbations exhibit scale dependent runaway growth at late times and,
in particular, a mode of a given wavenumber goes nonlinear at a higher redshift
than in the standard $\Lambda$CDM model. We discuss the implications of this
behavior and why there are reasons to expect that the growth will be cut off in
the nonlinear regime. Assuming that this holds in a full nonlinear analysis, we
briefly describe how upcoming measurements may probe the differences between
the modified theory and the standard $\Lambda$CDM model.

The braneworld model of Dvali, Gabadadze and Porrati (DGP) provides an
intriguing modification of gravity at large distances and late times. By
embedding a threebrane in an uncompactified extra dimension with separate
EinsteinHilbert terms for both brane and bulk, the DGP model allows for an
accelerating universe at late times even in the absence of an explicit vacuum
energy. We examine the evolution of cosmological perturbations on large scales
in this theory. At late times, perturbations enter a DGP regime in which the
effective value of Newton's constant increases as the background density
diminishes. This leads to a suppression of the integrated SachsWolfe effect,
bringing DGP gravity into slightly better agreement with WMAP data than
conventional LCDM. However, we find that this is not enough to compensate for
the significantly worse fit to supernova data and the distance to the
lastscattering surface in the pure DGP model. LCDM is, therefore, a better
fit.