Towards the complete understanding of water by a first-principles computational approach

The current state-of-the-art in theoretical investigations of water is reviewed. Emphasis is on first-principles approaches, i.e., those not using any experimental information to make predictions. These approaches have substantially advanced in the past few years, to the point of providing the currently most accurate description of the various forms of water, more accurate than empirical methods which fit parameters of water models to observations. The advantage of a first-principles approach is that it allows a deeper physical insight in the properties of water and therefore achieves a better understanding of this unusual medium.