Summer Schools

Joint Summerschool ISAM, IGDK 1754 and TopMath on

"Variational Inequalities and Optimization"

Technische Universität München

July 20 - 24, 2015.

Lecturers:

Prof. Dr. Giuseppe Buttazzo (Universita di Pisa) - Optimization Problems in Mass Transport Theory

	The theory of optimal transport starts in 1781 by Gaspard Monge, who proposed a model to describe the best way to move a given mass density into a target configuration, with a minimal total cost. The Monge model is very flexible and suitable to generalizations that can be used in a very large variety of optimization problems. During the course ``Optimization problems in mass transport theory'' some of them will be presented. A possible schedule is the following; due to the lack of time, some of the topics will be only outlined.The theory of optimal transport starts in 1781 by Gaspard Monge, who proposed a model to describe the best way to move a given mass density into a target configuration, with a minimal total cost. The Monge model is very flexible and suitable to generalizations that can be used in a very large variety of optimization problems. During the course ``Optimization problems in mass transport theory'' some of them will be presented. A possible schedule is the following; due to the lack of time, some of the topics will be only outlined.

The theory of optimal transport starts in 1781 by Gaspard Monge, who proposed a model to describe the best way to move a given mass density into a target configuration, with a minimal total cost. The Monge model is very flexible and suitable to generalizations that can be used in a very large variety of optimization problems. During the course ``Optimization problems in mass transport theory'' some of them will be presented. A possible schedule is the following; due to the lack of time, some of the topics will be only outlined.The theory of optimal transport starts in 1781 by Gaspard Monge, who proposed a model to describe the best way to move a given mass density into a target configuration, with a minimal total cost. The Monge model is very flexible and suitable to generalizations that can be used in a very large variety of optimization problems. During the course ``Optimization problems in mass transport theory'' some of them will be presented. A possible schedule is the following; due to the lack of time, some of the topics will be only outlined.

Introduction to optimal transport theory.
Optimization problems in urban planning.
Location, irrigation, average distance problems.
Shape optimization problems via optimal transport.
Optimal Dirichlet regions for elliptic equations.
Sampling patterns for control problems.
Reducing congestion in transport areas.

Prof. Dr. Michael Hintermüller (Humboldt-Universität Berlin) - Quasi-Variational Inequalities

	The lecture series presents existence and regularity results for quasi/variational inequalities with monotone / elliptic and parabolic operators. In particular, constraints of obstacle and gradient type will be of interest as well as nonlinear partial differential operators such as the p-Laplacian. Motivating applications come from contact problems, superconductivity, plasticity (with thermal effects), or electrostatics. For the numerical treatment, path-following semismooth Newton methods in function space will be presented and numerical aspects, including the balancing of path- and discretization parameters will be addressed.

The lecture series presents existence and regularity results for quasi/variational inequalities with monotone / elliptic and parabolic operators. In particular, constraints of obstacle and gradient type will be of interest as well as nonlinear partial differential operators such as the p-Laplacian. Motivating applications come from contact problems, superconductivity, plasticity (with thermal effects), or electrostatics. For the numerical treatment, path-following semismooth Newton methods in function space will be presented and numerical aspects, including the balancing of path- and discretization parameters will be addressed.

Prof. Dr. Marc Pfetsch (TU Darmstadt) - Mixed Integer Nonlinear Optimization

	We deal with optimization problems that involve both nonlinear constraints and requirements that some variables take only integer values. Such problems are very hard to solve, but are of huge practical importance; we present some examples. The goal is introduce methods that guarantee the computation of global optimal solutions. Starting from the convex case, we introduce methods like nonlinear branch-and-bound and outer approximation and prove their correctness. For the general case, one also needs to compute relaxations which are refined using so-called spatial branching. This involves convex underestimators and concave overestimators.

We deal with optimization problems that involve both nonlinear constraints and requirements that some variables take only integer values. Such problems are very hard to solve, but are of huge practical importance; we present some examples. The goal is introduce methods that guarantee the computation of global optimal solutions. Starting from the convex case, we introduce methods like nonlinear branch-and-bound and outer approximation and prove their correctness. For the general case, one also needs to compute relaxations which are refined using so-called spatial branching. This involves convex underestimators and concave overestimators.

Programme and timetable

Download of the Agenda

Conference venue

Room H.E.008
Leibniz-Rechenzentrum (LRZ)
Boltzmannstraße 1
85748 Garching by Munich

Getting thereGetting there

Travelling by public transport:
It takes 25 min by Underground from Marienplatz (town center of Munich) to the Garching campus
(No. U6 to the terminal Garching-Forschungszentrum (= Garching research center)).
Coming from Munich Airport:
- Take the S-Bahn (suburban railway) S1 to Neufahrn, then switch to the bus No. 690
  or
- take the S8 as far as Ismaning, change then to bus No. 230.

Registration

The deadline for on-line registration has already passed.

Poster

Download of the summerschools's poster.

Contact

Graduate Office
SB-S-MA
Technische Universität München
Boltzmannstraße 3
D-85748 Garching

Mail: gradoffice emat

ma.tum.de

Room: MI 03.012.018
Phone: +49 (0)89 289-17046 or +49 (0)89 289-17043

Organizers

AccommodationAccommodation

We recommend the following hotels.

Motel One Garching, Daimlerstraße 5a, 85748 Garching Hochbrück.
59 EUR single room, including breakfast. 69 EUR double room, including breakfast.
Please indicate the keyword "TUM / Bund 2015" when booking.

Hotel Hoyacker Hof, Garching, Freisinger Landstraße 9a, 85748 Garching.
75 EUR single room, including breakfast.
Please indicate the keyword "TUM Summerschool" when booking.

Acknowledgement

We acknowledge the support of Leibniz Supercomputing Center (LRZ)

Material

Impressum | Datenschutzerklärung | AnregungenCopyright Technische Universität München