Building bridges between nuclear reactions and statistical models

In modern nuclear data evaluations, nuclear reactions and statistical methods cannot be separated. Whereas the first one is continuously improved for many decades, the second one is now benefiting from large computer power. Ahead of its time, our colleague Eric Bauge had understood the advantage of linking them together. He developed modern Bayesian methods, and helped many of us to move in this direction. This short paper will present two examples of the work that we did together, following his vision: the application of BFMC, and the evidence of correlation between nubar, chi and fission cross section. Finally, he was not only a bridge builder, he was also able to jump from fundamental physics to very applied aspects, making him a frontier crosser.