| Abstract: | One of the main challenges in modeling massive stars to the onset of core collapse is the computational bottleneck of nucleosynthesis during late burning stages. The large number of isotopes formed makes the simulations computationally intensive and prone to numerical instability. To overcome this barrier, we designed a nuclear neural network (NNN) framework to emulate nucleosynthesis in massive stars following oxygen depletion in the core. We find that the NNN successfully predicts the results obtained with large nuclear networks, which are crucial for multidimensional simulations of supernovae, at a computational cost comparable to that of the small commonly used networks. While further work is needed to integrate NNN trained models into stellar evolution codes, this approach is promising for facilitating a large-scale generation of supernova progenitors with higher physical fidelity, thus advancing our understanding of the explosion mechanism, the evolution of gravitational wave progenitors, and the role of massive binaries in chemical enrichment of galaxies. |