Value and Policy Function Approximations in Infinite-Horizon Optimization Problems