Physics-Informed Neural Network Policy Iteration: Algorithms, Convergence, and Verification | Synapse