Optimization for Learning and Control