p-div-gnn.This work proposes P-DivGNN, a divergence regularized physics-informed graph neural network tailored for accurately reconstructing mechanical fields in materials undergoing large deformations using elastic and hyperelastic constitutive models.