Solvers

Exact solvers, computing the quantity a learning algorithm can only estimate. They are used to obtain the ground truth an experiment is compared against: the optimal value function of a finite MDP, the optimal gain of an LQR problem, and the optimal Q-function of the car-on-hill problem by exhaustive search.

value_iteration	Value iteration algorithm to solve a dynamic programming problem.
policy_iteration	Policy iteration algorithm to solve a dynamic programming problem.
solve_car_on_hill	Solver of the Car-On-Hill environment.
compute_lqr_feedback_gain	Computes the optimal gain matrix K.

• Dynamic programming
• Car on hill solver
• LQR solver