6. Conclusion

Simulation in materials science has reached a degree of maturity that enables it to be used for predictive purposes, even at industrial level. However, it is still subject to limitations that prevent it from being considered as an autonomous tool. In particular, the fact that a model may be entirely satisfactory (i.e. capable of accurately representing observed phenomena) does not necessarily prove that the scientific hypotheses formulated are correct. In fact, the question is "not whether a model is correct or not", but rather how to assess its contribution to a process of progressive understanding of the system and its behaviours, motivated by a set of initially explicit questions.

The model's digital laboratory is often too complex to access the real mechanisms of a simulated phenomenon. Under these conditions, the model doesn't shed any light on the nature of...