Ferroelectric materials have received lots of attention thanks to their intriguing properties such as the piezoelectric and pyroelectric effects, as well as the large dielectric constants and the spontaneous polarization which can potentially be used for information storage. In particular, perovskite crystal has a very simple unit cell structure yet a very rich phase transition diagram, which makes it one of the most intensively studied ferroelectric materials. In this dissertation, we use effective Hamiltonian, a first-principles-based computational technique to study the finite-temperature properties of ferroelectric perovskites. We studied temperature-graded (BaxSr1-x )TiO3 (BST) bulk alloys as well as the dynamics of nanodomain walls (nanowalls) in Pb(ZrxTi1-x )O3 (PZT) ultra-thin films under the driving force of an AC field. Our computations suggest that, for the temperature-graded BST, the polarization responds to the temperature gradient (TG), with the "up" and "down" offset observed in polarization components along the direction of TG, in agreement with the findings from experiments. For the nanowalls in PZT, the dynamics can be described by the damped-harmonic-oscillator model, and we observed a size-driven transition from resonance to relaxational dynamics at a critical thickness of 7.2 nm. The transition originates from the change in the effective mass of a nanowall as a film thickness increases. Some of the findings may find potential applications in various devices, such as thermal sensors, energy converters, or novel memory units.