Corrected gravity

Researchers trained a neural network first on the standard cosmological model and then on models that included massive neutrinos, corrected gravity, and primitive non-Gaussianity. The method saved expensive simulations, but in some cases caused “transfer”