The DGP model is given by the action
The equations of motion read  calculated on the brane and is the energy-momentum tensor of localized matter. Since , the continuity equation follows from Eq. (13.48). The length scale is defined by
If we consider the flat FLRW brane (), we obtain the modified Friedmann equation [201, 203]
In the DGP model the modification of gravity comes from a scalar-field degree of freedom, usually called , which is identified as a brane bending mode in the bulk. Then one may wonder if such a field mediates a fifth force incompatible with local gravity constraints. However, this is not the case, as the Vainshtein mechanism is at work in the DGP model for the length scale smaller than the Vainshtein radius , where is the Schwarzschild radius of a source. The model can evade solar system constraints at least under some range of conditions on the energy-momentum tensor [204, 285, 496]. The Vainshtein mechanism in the DGP model originates from a non-linear self-interaction of the scalar-field degree of freedom.
Although the DGP model is appealing and elegant, it is also plagued by some problems. The first one is that, although the model does not possess ghosts on asymptotically flat manifolds, at the quantum level, it does have the problem of strong coupling for typical distances smaller than 1000 km, so that the theory is not easily under control . Besides the model typically possesses superluminal modes. This may not directly violate causality, but it implies a non-trivial non-Lorentzian UV completion of the theory . Also, on scales relevant for structure formation (between cluster scales and the Hubble radius), a quasi-static approximation to linear cosmological perturbations shows that the DGP model contains a ghost mode . This linear analysis is valid as long as the Vainshtein radius is smaller than the cluster scales.
The original DGP model has been tested by using a number of observational data at the background level [525, 238, 405, 9, 549]. The joint constraints from the data of SN Ia, BAO, and the CMB shift parameter show that the flat DGP model is under strong observational pressure, while the open DGP model gives a slightly better fit [405, 549]. Xia  showed that the parameter in the modified Friedmann equation  is constrained to be (68% confidence level) by using the data of SN Ia, BAO, CMB, gamma ray bursts, and the linear growth factor of matter perturbations. Hence the flat DGP model () is not compatible with current observations.
On the sub-horizon scales larger than the Vainshtein radius, the equation for linear matter perturbations in the DGP model was derived in [400, 369] under a quasi-static approximation:
The index of the growth rate is approximated by . This is quite different from the value for the CDM model. If the future imaging survey of galaxies can constrain within 20%, it will be possible to distinguish the CDM model from the DGP model observationally . We recall that in metric f (R) gravity the growth index today can be as small as because of the enhanced growth rate, which is very different from the value in the DGP model.
Comparing Eq. (13.53) with the effective gravitational constant (10.42) in BD theory with a massless limit (or the absence of the field potential), we find that the parameter has the following relation with :. There is a claim that the ghost may disappear for the Vainshtein radius of the order of , because the linear perturbation theory is no longer applicable . In fact, a ghost does not appear in a Minkowski brane in the DGP model. In  it was shown that the Vainshtein radius in the early universe is much smaller than the one in the Minkowski background, while in the self accelerating universe they agree with each other. Hence the perturbative approach seems to be still possible for the weak gravity regime beyond the Vainshtein radius.
There have been studies regarding a possible regularization in order to avoid the ghost/strong coupling limit. Some of these studies have focused on smoothing out the delta profile of the Ricci scalar on the brane, by coupling the Ricci scalar to some other scalar field with a given profile [363, 362]. In  the authors included the brane and bulk cosmological constants in addition to the scalar curvature in the action for the brane and showed that the effective equation of state of dark energy can be smaller than . A monopole in seven dimensions generated by a SO(3) invariant matter Lagrangian is able to change the gravitational law at its core, leading to a lower dimensional gravitational law. This is a first approach to an explanation of trapping of gravitons, due to topological defects in classical field theory [508, 184]. Other studies have focused on re-using the delta function profile but in a higher-dimensional brane [334, 333, 197]. There is also an interesting work about the possibility of self-acceleration in the normal DGP branch [ in Eq. (13.50)] by considering an f (R) term on the brane action  (see also ). All these attempts indeed point to the direction that some mechanism, if not exactly DGP rather similar to it, may avoid a number of problems associated with the original DGP model.
This work is licensed under a Creative Commons License.