Title:  New Frontiers in Primal-Dual Algorithms for Large-Scale Convex
Optimization

Levent Tuncel


Abstract: As the availability of big data sets and the interest in solving much larger
scale convex optimization problems increased in the last twenty years, the first-order
algorithms came back to the forefront. We will focus on modelling and algorithms
which utilize duality in a significant way and connect first-order algorithms to
second-order algorithms in a natural way.  We will start with a general discussion
of utilization of the dual problem in modelling and algorithms for convex optimization,
have a journey through variable metric methods in first-order algorithms, self-concordant
barrier functions and then finally conclude with review and report of some exciting
recent work on first-order and second-order primal-dual interior-point algorithms.