We propose a new model of holographic dark energy with an action. It is the first time that one can derive a HDE model from the action principle. The puzzles of causality and circular logic about HDE have been completely solved in this model. The evolution of the universe only depends on the present state of the universe, clearly showing that it obeys the law of causality. Furthermore, the use of future event horizon as a present cut-off is not an input but automatically follows from equations of motion. Interestingly, this new model is very similar to the initial one of Li except a new term which may be explained as dark radiation.

We prove that, in general, the first law of black hole thermodynamics, $\delta Q=T\delta S$, is violated in $f(T)$ gravity. As a result, it is possible that there exists entropy production, which implies that the black hole thermodynamics can be in non-equilibrium even in the static spacetime. This feature is very different from that of $f(R)$ or that of other higher derivative gravity theories. We find that the violation of first law results from the lack of local Lorentz invariance in $f(T)$ gravity. By investigating two examples, we note that $f"(0)$ should be negative in order to avoid the naked singularities and superluminal motion of light. When $f"(T)$ is small, the entropy of black holes in $f(T)$ gravity is approximatively equal to $\frac{f'(T)}{4}A$.

We investigate the Hamiltonian formulation of $f(T)$ gravity and find that there are five degrees of freedom. The six first class constraints corresponding to the local Lorentz transformation in Teleparallel gravity become second class constraints in $f(T)$ gravity, which leads to the appearance of three extra degrees of freedom and the violation of the local Lorentz invariance in $f(T)$ gravity. In general, there are $D-1$ extra degrees of freedom for $f(T)$ gravity in $D$ dimensions, and this implies that the extra degrees of freedom correspond to one massive vector field or one massless vector field with one scalar field.

We find that the ambiguity term of approximate Killing vector field is responsible for the entropy production term. Without the ambiguity term, pure Einstein theory and $f(R)$ satisfy the relation of thermodynamic equilibrium. Considering such an ambiguity term of approximate Killing vector field, we can get the entropy production term and the entropy in $f(R)$ with a form defined by Jacobson. In pure Einstein theory, the shear term is the only geometric contribution of entropy production term, while in $f(R)$ it can also contribute. We believe our approach and conclusion can be generalized to other gravity theory.

Extending the holographic program of [1], we derive f(R) gravity and the Maxwell equations from the holographic principle, using time-like holographic screen. We find that to derive the Einstein equations and f(R) gravity in a natural holographic approach, the quasi-static condition is necessary. We also find the surface stress tensor and the surface electric current, surface magnetic current on the holographic screen for f(R) gravity and Maxwell's theory, respectively.

We propose a new holographic program of gravity in which we introduce a surface stress tensor. Our proposal differs from Verlinde's in several aspects. First, we use an open or a closed screen, a temperature is not necessary but a surface energy density and pressure are introduced. The surface stress tensor is proportional to the extrinsic curvature. The energy we use is Brown-York energy and the equipartiton theorem is violated by a non-vanishing surface pressure. We discuss holographic thermodynamics of a gas of weak gravity and find a chemical potential, and show that Verlinde's program does not lead to a reasonable thermodynamics. The holographic entropy is similar to the Bekenstein entropy bound.

We compute the Casimir energy of the photon field in a de Sitter space and find it to be proportional to the size of the horizon, the same form of the holographic dark energy. We suggest to make metamaterials to mimic de Sitter space in laboratory and measure the predicted Casimir energy.