Resumen
The spectral energy distribution of galaxies contains a convolved information on their stellar and gas content, on the star formation rate and history. It is therefore the most direct probe of galaxy properties. Each spectral range is mostly dominated by some specific emission sources or radiative processes so that only by modelling the whole spectral range it is possible to de-convolve and interpret the information contained in the SED in terms of SFR and galaxy evolution in general. The ingredients and kind of computations considered in models for the SEDs of galaxies depend on their aims. Theoretical models have the advantage of a broader interpretative and predictive power with respect to observationally calibrated semi-empirical approaches, the major drawback being a longer computational time. I summarize the main features of GRASIL, a code to compute the UV to radio SED of galaxies treating the radiative transfer and dust emission with particular care. It has been widely applied to interpret observations and to make predictions for semi-analytical galaxy formation models. I present in particular the applications in the context of galaxy models, and the new method implemented in GRASIL based on the artificial neural network algorithm to cope with the computing time for cosmological applications.
