The objective of this internship is to characterize the role of microstructural defects on pore germination. Several questions arise about the cavitation threshold, the role of compressibility, the influence of defects (vacancies, dislocations, grain boundaries) on this initiation and the consequence on the pore size distributions generated. However, monitoring nanometric phenomena is impossible to achieve experimentally in real time. Post-mortem, numerous events are superimposed on germination, it becomes difficult to extract the phenomenon of interest. This study, on the other hand, is potentially affordable via simulations working at the micrometric scale. Thanks to the new architectures of supercomputers, and the development of adapted codes, classical molecular dynamics simulations can process a few billion atoms, compatible with the scales of defects sought. However, these simulations generate a considerable amount of data and their exploitation remains a challenge. These computer tools are entirely to be developed for small-scale studies (a few dozen atoms).