Speaker
Description
We present a model-independent determination of the dark energy equation of state w_{DE}(z) and the growth index γ(z) using physics-informed neural networks (PINNs). The background expansion and the linear matter perturbation equation are imposed as soft constraints in the training loss, so that the recovered functions satisfy the governing differential equations without assuming a parametric form for either w_{DE}(z) or γ(z). We apply the framework jointly to background observations and to redshift-space distortion measurements of, whose combination breaks the degeneracy between the expansion history and the growth of structure that limits either dataset alone. Posterior uncertainties on the network parameters are obtained with Hamiltonian Monte Carlo.