VL Nonlinear Optimization

Contents

• News
• Lectures
• Contact
• General Information
• Registration
• Tutorial and Certificates
• Homework
• Literature

News

The lecture on tuesday afternoon starts at 14:00 sharp

The tutorial is on Monday 12:00-14:00 in Arnimallee 3 SR119

Please register to the lecture here. Non FU students may choose "Gasthörer" as Studiengang.

Lectures

Lecture	Tue. 10:15 - 11:45	Arnimallee 6, SR 025/027
	Tue. 14:00 - 15:30	Arnimallee 3, SR 005
Tutorial	Mon. 12:15 - 13:45	Arnimallee 3, SR 119

Contact

Dr. Anton Schiela	Zuse-Institut Berlin (ZIB), Takustraße 7, Raum 4305
	e-mail: schiela(at)zib.de
Sebastian Götschel	Zuse-Institut Berlin (ZIB), Takustraße 7, Raum 4307
	e-mail: goetschel(at)zib.de

General Information

Contents of the Lecture

Nonlinear optimization deals with the search for minimal points of nonlinear functions. Often there are constraints to fulfill, i.e., the minimizer has to satisfy given equalities and/or inequalities.
The numerical solution of such problems is of high practical relevance. On the one hand, there are many instances, where one looks for a "best" solution, on the other hand, many physical problems are described via minimization problems. This makes nonlinear optimization an active research topic.
In this lecture we consider the analysis of nonlinear optimization problems, but most importantly the construction of modern numerical algorithms for their solution.

Topics

nonlinear optimization problems, modeling
optimality conditions
numerical methods for unconstrained optimization(Newton-type, CG-methods, trust-region techniques)
numerical methods for equality constrained optimization
numerical methods for inequality constrained optimization

Registration

Registration

Please register electronically here.

Tutorial and Certificates

Homework:

• There is a weakly homework sheet, which is distributed during the lectures on tuesday. They are also available electronically .
• Every two weeks there is a programming exercise which should be solved using matlab or c/c++.
• All exercises may be solved in teams of at most two people.
• Homeworks are collected during the second lecture on tuesday. Deadline is the end of the lecture 15:30

Examn:

• At the end of the term there is an examn, which has to be passed to get a certificate. Condition for participation is that at least 50% of the homework credit points have been reached.
• Date and time of the examn are towards the end of the semester. The exact date is to be announced.
• Certificates have a grade, according to the grade reached at the examn.

Criteria for the certifiacte:

• 50 % of the total number of homework points
• Examn successfully passed

Sheets

Homework sheets in Portable Document Format (*.pdf) can be downloaded here . New sheets are usually uploaded Monday evening. Homework sheets in printed form are distributed during the tuesday lectures

Literature:

Nocedal/Wright: Numerical Optimization
Conn/Gould/Toint: Trust Region Methods
Jarre/Stoer: Optimierung
Bonnans/Gilbert/Lemarechal/Sagastizabal: Numerical Optimization
Geiger/Kanzow: Theorie und Numerik restringierter Optimierungsaufgaben
Fletcher: Practical methods of optimization