Numerical Optimization

Advanced numerical optimization with emphasis on modern first- and second-order algorithms, acceleration techniques, large-scale methods, and constrained optimization. Topics include gradient, proximal, mirror, and subgradient methods; heavy-ball and Nesterov acceleration; Newton and quasi-Newton methods; incremental and stochastic gradient methods and randomized coordinate descent; projected gradient descent and interior-point methods. Theoretical foundations and practical applications to real-world problems are emphasized. Programming experience required. Prerequisites: MATH 128, MATH 225, and MATH 226.