In this paper we consider a specific type of the bimetric theory of gravitation with the two different metrics introduced in the cosmological frame. Both metrics respect all the symmetries of the standard FLRW solution and contain conformally related spatial parts. One of the metric is assumed to describe the causal structure for the matter. Another metric defines the causal structure for the gravitational interactions. A crucial point is that the spatial part of the metric describing gravity is given by the spatial part of the matter metric confromally rescaled by a time-dependent factor $\alpha$ which, as it turns out, can be linked to the effective gravitational constant and the effective speed of light. In the context of such a bimetric framework we examine the strength of some singular cosmological scenarios in the sense of the criteria introduced by Tipler and Kr\'olak. In particular, we show that for the nonsingular scale factor associated with the matter metric, both the vanishing or blowing up of the factor $\alpha$ for some particular moment of the cosmic expansion may lead to a strong singularity with infinite value of the energy density and infinite value of the pressure.

Using the idea of regularisation of singularities due to the variability of the fundamental constants in cosmology we study the cyclic universe models. We find two models of oscillating and non-singular mass density and pressure ("non-singular" bounce) regularised by varying gravitational constant $G$ despite the scale factor evolution is oscillating and having sharp turning points ("singular" bounce). Both violating (big-bang) and non-violating (phantom) null energy condition models appear. Then, we extend this idea onto the multiverse containing cyclic individual universes with either growing or decreasing entropy though leaving the net entropy constant. In order to get an insight into the key idea, we consider the doubleverse with the same geometrical evolution of the two "parallel" universes with their physical evolution (physical coupling constants $c(t)$ and $G(t)$) being different. An interesting point is that there is a possibility to exchange the universes at the point of maximum expansion -- the fact which was already noticed in quantum cosmology. Similar scenario is also possible within the framework of Brans-Dicke theory where varying $G(t)$ is replaced by the dynamical Brans-Dicke field $\phi(t)$ though these theories are slightly different.

In this paper we extend a new method to measure possible variation of the speed of light by using Baryon Acoustic Oscillations and the Hubble function presented in our earlier paper [V. Salzano, M. P. D\c{a}browski, and R. Lazkoz, Phys. Rev. D93, 063521 (2016)] onto an inhomogeneous model of the universe. The method relies on the fact that there is a simple relation between the angular diameter distance $(D_{A})$ maximum and the Hubble function $(H)$ evaluated at the same maximum-condition redshift, which includes speed of light $c$. One limit of such method was the assumption of null spatial curvature (even if we showed that even a non-zero curvature would have negligible effects). Here, we move one step further: we explicitly assume a model with intrinsic non-null curvature, and calculate the exact relation between $D_{A}$ and $H$ in this case. Then, we evaluate if current or future missions such as SKA can be sensitive enough to detect any such kind of spatial variation of $c$ which can perhaps be related to the recently observed spatial variation of the fine structure constant (an effect known as $\alpha$-dipole).

We consider gravity theory with varying speed of light and varying gravitational constant. Both constants are represented by non-minimally coupled scalar fields. We examine the cosmological evolution in the near curvature singularity regime. We find that at the curvature singularity the speed of light goes to infinity while the gravitational constant vanishes. This corresponds to the Newton's Mechanics limit represented by one of the vertex of the Bronshtein-Zelmanov-Okun cube. The cosmological evolution includes both the pre-big-bang and post-big-bang phases separated by the curvature singularity. We also investigate the quantum counterpart of the considered theory and find the probability of transition of the universe from the collapsing pre-big-bang phase to the expanding post-big-bang phase.

We consider spherically symmetric inhomogeneous pressure Stephani universes, the center of symmetry being our location. The main feature of these models is that comoving observers do not follow geodesics. In particular, comoving perfect fluids have necessarily a radially dependent pressure. We consider a subclass of these models characterized by some inhomogeneity parameter $\beta$. We show that also the velocity of sound, like the (effective) equation of state parameter, of comoving perfect fluids acquire away from the origin a time and radial dependent change proportional to $\beta$. In order to produce a realistic universe accelerating at late times without dark energy component one must take $\beta < 0$. The redshift gets a modified dependence on the scale factor $a(t)$ with a relative modification of $-9\%$ peaking at $z\sim 4$ and vanishing at the big-bang and today on our past lightcone. The equation of state parameter and the speed of sound of dustlike matter (corresponding to a vanishing pressure at the center of symmetry $r=0$) behave in a similar way and away from the center of symmetry they become negative -- a property usually encountered for the dark energy component only. In order to mimic the observed late-time accelerated expansion, the matter component must significantly depart from standard dust, presumably ruling this subclass of Stephani models out as a realistic cosmology. The only way to accept these models is to keep all standard matter components of the universe including dark energy and take an inhomogeneity parameter $\beta$ small enough.

We discuss minisuperspace models within the framework of varying physical constants theories including $\Lambda$-term. In particular, we consider the varying speed of light (VSL) theory and varying gravitational constant theory (VG) using the specific ans\"atze for the variability of constants: $c(a) = c_0 a^n$ and $G(a)=G_0 a^q$. We find that most of the varying $c$ and $G$ minisuperspace potentials are of the tunneling type which allows to use WKB approximation of quantum mechanics. Using this method we show that the probability of tunneling of the universe "from nothing" ($a=0)$ to a Friedmann geometry with the scale factor $a_t$ is large for growing $c$ models and is strongly suppressed for diminishing $c$ models. As for $G$ varying, the probability of tunneling is large for $G$ diminishing, while it is small for $G$ increasing. In general, both varying $c$ and $G$ change the probability of tunneling in comparison to the standard matter content (cosmological term, dust, radiation) universe models.

Exact luminosity distance and apparent magnitude formulas are applied to Union2 557 supernovae sample in order to constrain possible position of an observer outside of the center of symmetry in spherically symmetric inhomogeneous pressure Stephani universes which are complementary to inhomogeneous density Lema\^itre-Tolman-Bondi (LTB) void models. Two specific models are investigated. The first which allows a barotropic equation of state at the center of symmetry with no scale factor function being specified (model IIA), and the second which has no barotropic equation of state at the center, but has an explicit dust-like scale factor evolution (model IIB). It is shown that even at $3\sigma$ CL, an off-center observer cannot be further than about 4.4 Gpc away from the center of symmetry which is comparable to the reported size of a void in LTB models with the most likely value of the distance from the center about 341 Mpc for model IIA and 68 Mpc for model IIB. The off-center observer cannot be farther away from the center than about 577 Mpc for model IIB at $3\sigma$ CL. It is evaluated that the best-fit parameters which characterize inhomogeneity are: $\Omega_{inh} = 0.77$ (dimensionless - model IIA) and $\alpha = 7.31 \cdot 10^{-9}$ $(s/km)^{2/3} Mpc^{-4/3}$ (model IIB).

We derive a redshift drift formula within the framework of varying speed of light (VSL) theory using the specific ansatz for the variability of $c(t) = c_0 a^n(t)$. We show that negative values of the parameter $n$, which correspond to diminishing value of the speed of light during the evolution of the universe, effectively rescales dust matter to become little negative pressure matter, and the cosmological constant to became phantom. Positive values of $n$ (growing $c(t)$) make VSL model to become more like Cold Dark Matter (CDM) model. Observationally, there is a distinction between the VSL model and the $\Lambda$CDM model for the admissible values of the parameter $n \sim - 10^{-5}$, though it will be rather difficult to detect by planned extremely large telescopes (E-ELT, TMT, GMT) within their accuracy.

We derive a luminosity distance formula for the varying speed of light (VSL) theory which involves higher order characteristics of expansion such as jerk, snap and lerk which can test the impact of varying $c$ onto the evolution of the universe. We show that the effect of varying $c$ is possible to be isolated due to the relations connecting observational parameters already by measuring the second-order term in redshift $z$ unless there is a redundancy between the curvature and an exotic fluid of cosmic strings scaling the same way as the curvature.

We review the variety of new singularities in homogeneous and isotropic FRW cosmology which differ from standard Big-Bang and Big-Crunch singularities and suggest how the nature of these singularities can be influenced by the varying fundamental constants.

We derive a redshift drift formula for the spherically symmetric inhomogeneous pressure Stephani universes which are complementary to the spherically symmetric inhomogeneous density Lema\^itre-Tolman-Bondi models. We show that there is a clear difference between redshift drift predictions for these two models as well as between the Stephani models and the standard $\Lambda$CDM Friedmann models. The Stephani models have positive drift values at small redshift and behave qualitatively (but not quantitatively) as the $\Lambda$CDM models at large redshift, while the drift for LTB models is always negative. This prediction may perhaps be tested in future telescopes such as European Extremely Large Telescope (EELT), Thirty Meter Telescope (TMT), Giant Magellan Telescope (GMT), and especially, in gravitational wave interferometers DECi-Hertz Interferometer Gravitational Wave Observatory and Big Bang Observer (DECIGO/BBO), which aim at low redshift.

We discuss evolution of density perturbations in cosmological models which admit finite scale factor singularities. After solving the matter perturbations equations we find that there exists a set of the parameters which admit a finite scale factor singularity in future and instantaneously recover matter density evolution history which are indistinguishable from the standard LCDM scenario.

By setting some special boundary conditions in the variational principle we obtain junction conditions for the five-dimensional $f(R)$ gravity which in the Einstein limit $f(R)\rightarrow R$ transform into the standard Randall-Sundrum junction conditions. We apply these junction conditions to a particular model of a Friedmann universe on the brane and show explicitly that the limit gives the standard Randall-Sundrum model Friedmann equation.

By the application of the generalized Israel junction conditions we derive cosmological equations for the fourth-order $f(R)$ brane gravity and study their cosmological solutions. We show that there exists a non-static solution which describes a four-dimensional de-Sitter $(dS_4)$ brane embedded in a five-dimensional anti-de-Sitter $(AdS_5)$ bulk for a vanishing Weyl tensor contribution. On the other hand, for the case of a non-vanishing Weyl tensor contribution, there exists a static brane only. We claim that in order to get some more general non-static $f(R)$ brane configurations, one needs to admit a dynamical matter energy-momentum tensor in the bulk rather than just a bulk cosmological constant.

We discuss a very general theory of gravity, of which Lagrangian is an arbitrary function of the curvature invariants, on the brane. In general, the formulation of the junction conditions (except for Euler characteristics such as Gauss-Bonnet term) leads to the powers of the delta function and requires regularization. We suggest the way to avoid such a problem by imposing the metric and its first derivative to be regular at the brane, the second derivative to have a kink, the third derivative of the metric to have a step function discontinuity, and no sooner as the fourth derivative of the metric to give the delta function contribution to the field equations. Alternatively, we discuss the reduction of the fourth-order gravity to the second order theory by introducing extra scalar and tensor fields: the scalaron and the tensoron. In order to obtain junction conditions we apply two methods: the application of the Gauss-Codazzi formalism and the application of the generalized Gibbons-Hawking boundary terms which are appended to the appropriate actions. In the most general case we derive junction conditions without assuming the continuity of the scalaron and the tensoron on the brane. The derived junction conditions can serve studying the cosmological implications of the higher-order brane gravity models.

We derive the most general junction conditions for the fourth-order brane gravity constructed of arbitrary functions of curvature invariants. We reduce these fourth-order theories to second order theories at the expense of introducing new scalar and tensor fields - the scalaron and the tensoron. In order to obtain junction conditions we apply the method of generalized Gibbons-Hawking boundary terms which are appended to the appropriate actions. After assuming the continuity of the scalaron and the tensoron on the brane, we recover junction conditions for such general brane universe models previously obtained by different methods. The derived junction conditions can serve studying the cosmological implications of the higher-order brane gravity models.

We discuss a general fourth-order theory of gravity on the brane. In general, the formulation of the junction conditions (except for Euler characteristics such as Gauss-Bonnet term) leads to the higher powers of the delta function and requires regularization. We suggest the way to avoid such a problem by imposing the metric and its first derivative to be regular at the brane, while the second derivative to have a kink, the third derivative of the metric to have a step function discontinuity, and no sooner as the fourth derivative of the metric to give the delta function contribution to the field equations. Alternatively, we discuss the reduction of the fourth-order gravity to the second-order theory by introducing an extra tensor field. We formulate the appropriate junction conditions on the brane. We prove the equivalence of both theories. In particular, we prove the equivalence of the junction conditions with different assumptions related to the continuity of the metric along the brane.

We discuss exotic singularities in the evolution of the universe motivated by the progress of observations in cosmology. Among them there are: Big-Rip (BR), Sudden Future Singularities (SFS), Generalized Sudden Future Singularities (GSFS), Finite Density Singularities (FD), type III, and type IV singularities. We relate some of these singularities with higher-order characteristics of expansion such as jerk and snap. We also discuss the behaviour of pointlike objects and classical strings on the approach to these singularities.

We discuss the behaviour of strings propagating in spacetimes which allow future singularities of either a sudden future or a Big-Rip type. We show that in general the invariant string size remains finite at sudden future singularities while it grows to infinity at a Big-Rip. This claim is based on the discussion of both the tensile and null strings. In conclusion, strings may survive a sudden future singularity, but not a Big-Rip where they are infinitely stretched.