An intuitive picture of a granular gas is that of a molecular gas where particles collide inelastically. However, there are important differences between molecules and grains. The temperature, for instance, plays no fundamental role in the behaviour of a granular gas, since grains are usually macroscopic particles. In addition, and due to the loss of energy in collision, a state of a granular gas in a real experiment needs for an energy injection. Consequently, phase transitions, non-equilibrium instabilities, spatiotemporal pattern formation, and segregation, among others, usually occur. Therefore, granular gases may be regarded as outstanding examples of complex systems. In this talk, It will be explained how traditional tools of non-equilibrium statistical mechanics, as well as numerical simulations methods, have been adapted to understand, predict, and explain a variety of states of a granular gas, as well as experimental observations. A special situation of a granular gas where energy flow is due only to a density gradient will be considered.