f(R) Cosmology revisited

We consider a class of metric f(R) modified gravity theories, analyze them in the context of a Friedmann-Robertson-Walker cosmology and confront the results with some of the known constraints imposed by observations. In particular, we focus in correctly reproducing the matter and effective cosmological constant eras, the age of the Universe, and supernovae data. Our analysis differs in many respects from previous studies. First, we avoid any transformation to a scalar-tensor theory in order to be exempted of any potential pathologies (e.g. multivalued scalar potentials) and also to evade any unnecessary discussion regarding frames (i.e. Einstein vs Jordan). Second, based on a robust approach, we recast the cosmology equations as an initial value problem subject to a modified Hamiltonian constraint. Third, we solve the equations numerically where the Ricci scalar itself is one of the variables, and use the constraint equation to monitor the accuracy of the solutions. We compute the "equation of state" (EOS) associated with the modifications of gravity using several inequivalent definitions that have been proposed in the past and analyze it in detail. We argue that one of these definitions has the best features. In particular, we present the EOS around the so called "phantom divide" boundary and compare it with previous findings.