A Boson-informed neural network for Ground States of Many-Body Schr" o dinger equations
A symmetry-preserving Boson-informed neural network (BINN) was developed for variational Monte Carlo computation of interacting bosons in continuous space. It separates physical scales using a cell Jastrow factor for microscopic interactions and a potential-aware envelope for macroscopic confinement. The method achieves high accuracy, approaching Hartree and Gross-Pitaevskii limits, with results matching Diffusion Monte Carlo within 0.06% error.