**Invited** and **contributed talks** with **poster session**.

Invited keynote speakers:

Invited speakers:

In UQ problems, one ultimately wants to analyze probability distributions of certain quantities of interest (QoI). In models used to describe the problem, randomness is often introduced in a somewhat ad-hoc fashion. This leaves much room for variation and raises various questions from the modelling perspective which we will discuss in this section, e.g.: How does the choice of model effect the probability law of the QoI? What are the computational advantages/disadvantages of different stochastic models? Are there effective models which can be rigorously justified by ab initio deductions from laws of physics? Can the modelling experience from neighbouring scientific fields such as statistical physics or geostatistics provide promising insights for the improvement of currently used approaches in UQ?

Invited keynote speakers:

Homogenization is a method of studying partial differential equations with rapidly oscillating coefficients where one approximates the original problem by an effective model obtained from asymptotic analysis. This model is often called 'homogenization formula' and is achieved by sending the length scale characteristic of the oscillations to zero. If the variation of the coefficients on small scales is random, a stochastic variant of homogenization theory can be applied. Here again one can obtain formulas for homogenized coefficients using an averaging procedure. More detailed questions concern fluctuations around the deterministic limit which is achieved through averaging. The distribution of the fluctuations encode uncertainty in a similar way as is the case for traditional UQ problems.

Invited keynote speakers:

- Jean-Christophe Mourrat
- Felix Otto (Slides)

An integral part of solving many UQ problems is the simulation of infinite or high dimensional stochastic systems. Methodically, one encounters the same challenge when studying lattice quantum field theory via path integrals. We wish to facilitate exchange between experts in these two fields in order to compare the current state of knowledge, the best algorithms available and challenges encountered.

Invited keynote speakers:

Many methods of UQ rely on (exponentiated) Gaussian processes, e.g. as a model for conductivity in an inhomogeneous porous medium. The crucial input to these methods is the covariance structure of the Gaussian process. In this hands-on tutorial based on the statistical computing environment R, the geostatistical estimation of covariances and semivariograms is demonstrated and relevant R-extensions (packages) are introduced. We also demonstrate how to estimate the uncertainty that originates from the statistical error in estimation. The tutorial is limited to 12 participants.

Open for all members of the GAMM Activity Group Uncertainty Quantification (www.tu-chemnitz.de/gamm-uq) and those interested in participating.

A more detailled program of the workshop will be announced in due time.

TU Dortmund

Fakultät für Mathematik

Lehrstuhl IX

Vogelpothsweg 87

44227 Dortmund

You find us on the 6th floor of the Math tower.

Janine Textor (room M 620)

Tel.: (0231) 755-3063

Fax: (0231) 755-5219

Mail: janine.textor@tu-dortmund.de