Boundary terms and junction conditions for the DGP π-Lagrangian and galileon

In the decoupling limit of DGP, π describes the brane-bending degree of freedom. It obeys second order equations of motion, yet it is governed by a higher derivative Lagrangian. We show that, analogously to the Einstein-Hilbert action for GR, the π-Lagrangian requires Gibbons-Hawking-York type boundary terms to render the variational principle well-posed. These terms are important if there are other boundaries present besides the DGP brane, such as in higher dimensional cascading DGP models. We derive the necessary boundary terms in two ways. First, we derive them directly from the brane-localized π-Lagrangian by demanding well-posedness of the action. Second, we calculate them directly from the bulk, taking into account the Gibbons-Hawking-York terms in the bulk Einstein-Hilbert action. As an application, we use the new boundary terms to derive Israel junction conditions for π across a sheet-like source. In addition, we calculate boundary terms and junction conditions for the galileons which generalize the DGP π-Lagrangian, showing that the boundary term for the n-th order galileon is the (n−1)-th order galileon.