Full-Newton step interior-point methods for conic optimization
More Info
expand_more
expand_more
Abstract
In the theory of polynomial-time interior-point methods (IPMs) two important classes of methods are distinguished: small-update and large-update methods, respectively. Small-update IPMs have the best theoretical iteration bound and IPMs with full-Newton steps belong to this class of methods. Within each of these classes one has feasible and infeasible interior-point methods (IIPMs). In this thesis we first deal with full-Newton step IIPMs, and we consider feasible full-Newton step IPMs.