Schedule
Tuesdays 4:30 pm at SAMSI, Research Triangle Park, NC, beginning Tuesday, January 17, 2017
(No class March 14, 2017 for Spring Break)
Last class: Tuesday, April 25, 2017
Instructor: Ekkehard Sachs, University of Trier
Course Description:
The main topic of this class is the numerical solution of constrained optimization problems. We review the classical schemes with a special emphasis on new and modern developments in this area. We also analyze how the algorithms can be especially adapted to optimization problem with constraints that come from ordinary, partial or even stochastic differential equations. Another important topic in this course will be applications in optimization which come from problems in statistics and which often exhibit special structure that can be exploited to reduced computation time and storage.
Specific topics to be covered include:
– first order methods like gradient projection, conditional gradient methods, etc
– second order methods such as SQP methods
– augmented Lagrangian methods- interior point methods- efficient computation of gradients via an adjoint calculus
– special optimization methods for problems including Monte Carlo simulations
– an outlook towards machine learning and support vector machines
On special occasions, we will invite guest lectures who are specialists in these topics.
Registration for this course is being processed through your respective university:
- UNC-CH: STOR 894.001/MATH 892.002
- Duke: 790-01
- NCSU: MA 810.002, ST 810.005 (NCSU Course Web Page)
Questions: email [email protected]