lifeClipper2 is an Augmented Reality project. It offers an audiovisual walking experience in a virtually extended public space.

Nowadays, for a web site to reach peak popularity it must present the latest information, combined from various sources, to give an interactive, customizable impression. Embedded content and functionality from a range of specialist Þelds has led to a signiÞcant improvement in web site quality. However, until now the capacity of a web site has been deÞned at the time of creation; extension of this capacity has only been possible with considerable additional effort. The aim of this thesis is to present a software architecture that allows users to personalize a web site themselves, with capabilities taken from the immense resources of the World Wide Web. Recent web sites are analyzed and categorized according to their customization potential. The results of this analysis are then related to patterns in the Þeld of software engineering and from these results, a general conclusion is drawn about the requirements of an application architecture to support these patterns. A theoretical concept of such an architecture is proposed and described in detail. The empirical part of the study includes an implementation of the proposal and a demonstration of the assembly of capabilities found in the World Wide Web. This implementation is based on established technologies but applies them to a new, specially-designed structure. It allows users to add selected facilities to arbitrary web sites simply by calling a speciÞc web address. This gives the user the potential to adapt the appearance and function of web sites to his or her personal needs. An in-depth analysis of the challenges and restrictions of the soft- ware design completes the proposed architecture. Practical examples of behavior patterns show possible implementations in a range of Þelds. Finally, a vision developed from the results presented in this thesis is outlined and sub jects for future research are examined.

This project will join forces between IBM, the High Performance Web Computing group at the University of Basel and the ITIS foundation (ETH Zurich) to develop a Cell-based mainframe system z that will host high performance programs to model, simulate and optimize biomedical hyperthermia cancer treatment planning tools. It will use Cell BE coprocessors to handle message-passing and physics simulations. The Cell Broadband Engine (BE) processor is a multi-core chip comprised of a 64-bit Power Architecture processor core and eight synergistic processor cores, capable of massive floating point processing, optimized for compute-intensive workloads and broadband rich media applications.

Hyperthermia cancer treatment an promising therapeutical option in oncology. By heating the tumor usually with electromagnetic energy, the tumor tissue is sensitized to radiotherapy and chemotherapy. It is therefore used in combination with either radiotherapy or chemotherapy.

Modelling and simulation of the thermal behavior of the tissue exposed to the non-ionizing radiation is an essential component of hyperthermia treatment planning. Finding optimal antenna parameters, given the patient geometry, leads to a large-scale PDE-constrained optimization problem. The most efficient methods for solving these nonlinear nonconvex programming problems are primal-dual interior point methods. These resulting optimization problems are computationally demanding and require special algorithmic solution schemes and high performance computing solutions. The ultimate goal is an interactive real-time optimization for an in-vivo patient hyperthermia planning tool.

In today's highly competitive industrial environment, it is imperative to minimize production cost. This project with its innovative graphics-based approach speeds up the calculation process with automatic feature detection and production sequence definition. It allows users to systematically identify the cost optimized manufacturing process using discrete optimization algorithms while ensuring full cost transparency and result reproducibility.

Before starting the optimization defining an appropriate mathematical model of the whole manufacturing process is a challenging task: Specify certain decision variables and the objective function(s) and thinking about possible constraints.

Typically, discrete optimization problems are NP-hard so that solving the problem exactly could be very difficult. Normally, existing algorithms need a preprocessing phase and a kind of a branch-and-bound algorithm combined with cutting planes techniques. Also, heuristics such as genetic algorithms are adapted to the solving method.

In the project a mapping of the cost calculator in a optimization model and the concluding optimal cost computation are the major and challenging tasks.

A thorough understanding of the physics of the human body is an essential prerequisite for practising medicine. Due to pressure of time, however, it is hardly possible for medical students to carry out experiments in this area during their studies. This is where the Physica pro medicis project comes in: e-learning modules simulate physical phenomena and trials that cannot be presented so clearly by means of traditional methods, using medical examples to illustrate and explain the physical fundamentals.

Physica pro medicis plans a combination of face-to-face teaching and self-studying on the Internet. The modules will mainly be used in the first year course, and be employed again in later years for the detailed discussion of individual medico-physical concepts such as X-rays and MRI.

Basel and Bern Universities have already worked together to design the content of the new "Physics for Doctors" lectures. This means that there is a considerable supply of copyright-free modules available for further use.