The first part of the thesis surveys existing approaches to solve the algorithm selection problem and discusses techniques to analyze simulation algorithm performance. A unified categorization of existing algorithm selection techniques is worked out, as these stem from various research domains (e.g., finance, artificial intelligence).

The second part introduces a software framework for automatic simulation algorithm selection and describes its constituents, as well as their integration into the modeling and simulation framework JAMES II. The implemented selection mechanisms are able to cope with three situations:

• no prior knowledge is available
• the impact of problem features on performance is unknown
• a relationship between problem features and algorithm performance can be established empirically

An experimental evaluation of the developed methods concludes the thesis. It is shown that an automated algorithm selection may significantly increase the overall performance of a simulation system. Some of the presented mechanisms also support the research on simulation methods, as they facilitate their development and evaluation.

For the modeling of regenerative systems validation and verification are important steps. Concern of this PhD is to create appropriate visual methods in order to support these processes. For that purpose it is necessary to represent modeling data at different levels of abstraction and constrains as well as to provide appropriate interaction methods.

Furthermore, the modeling data have to be visualized in connection with wet lab results and its associated simulation data in time and space. Of particular attention is the communication of uncertainties and model errors as well as the consideration of multiple scales.

The visual analysis of models on different levels of abstraction in space and time demands for sophisticated visualization techniques.

The research of the dynamic aspects of complex networks and their visualization are the primary concern of this phd. These dynamic changes involve changes to the node and edge sets (structural changes) as well as changes of associated attributes.

There is almost no visualization approach considering all three aspects (changes to node, edge and attribute sets). This results mainly of the increasing complexity of the amount of data to be visualized. For this reason, suitable automatic methods have to be used for computing and visualizing temporal patterns and trends on the one hand (high-level visualization). On the other hand, direct access to concrete values have to be available at any time (low-level visualization)

To support modeling and simulation as process it is essential to identify tasks and workflows in the modeling and simulation domain and to integrate those in JAMES II. Not only the graphical assistance but also the documentation of the steps taken is important. Additionally the annotation of models and experiments is an integral part of the workflow integration to simplify the reuse of models as well as the reproducibility of simulation experiments. Together with the storage and explicit representation of a workflow, annotations help to assure a certain credibility and quality. A starting point would be the workflow for experimental validation of cell biological models.

The aim of the PhD-project is supporting the validation process which is an essentiel step in the M&S workflow. On the one hand, the focus lies on the formalization of the validation step, on the other hand, various validation method shall be supported and integrated in one flexible environment. State of the art methods like simulation-based model-checking are of special interest in this context. Validation experiments with cell-biological models shall show the benefits of the approach.

Current research questions from the fields of biology and medicine require an integration of data on different levels, in different formats and from different sources. Gaining insights by exploring heterogeneous data is a main challenge of Visual Analytics -- the science of analytical reasoning. However, in order to support domain experts in such an analysis scenario, the pure provision of data in a feature rich analysis system is not sufficient. The central question is how to support a user by providing orientation -- and even going a step further by actively guiding him through an analysis session. The aim of this thesis is to give an overview of the needed prerequisites for guidance (input) and the visual means to achieve it (output). This thesis is carried out within the Caleydo Project (www.caleydo.org) at the Graz University of Technology and is also associated with the dIEM oSiRiS research training group at the University of Rostock.

Different simulators shall be developed and shall be combined in a component-based model. The starting point will be the realization of different Gillespie variants. Those shall be combined with continuous simulators. Thereby, more recent approaches that allow the execution of differential equation models based on discrete event approaches shall be considered as well. The goal is an efficient and flexible simulation system for cell biological systems.

In this PhD project, methods and tools will be developed so that neural systems can be modeled simulated at different spatial and temporal scales. In systematic simulation studies, the reorganization of neuronal networks via synaptic plasticity is investigated. Regeneration in neuronal systems is explored by simulating and analyzing neurogenesis and how new neurons are integrated into existing networks via synaptic plasticity mechanisms. One of the challenges for the modeling and simulation is that the developed tools need to handle the plasticity in neuronal systems at different spatial, temporal, and structural levels.

Central to the project is the modelling and simulation of molecular diffusion processes by means of Brownian dynamics (BD). Despite recent improvements in the development of BD algorithms, they remain computationally expensive, which hampers the simulation of long periods in time and large scale systems. The aim of the project is therefore to develop and combine suitable algorithms to support an effective and efficient simulation of Brownian dynamics. Thereby, the role of Brownian dynamics in computational biology and its potential application for cell biological simulations shall be explored.

In this context BD algorithms are developed and applied to estimate bimolecular association rates. In close cooperation with Tareck Rharass, the rate of association between Nucleoredoxin and Dishevelled is computed. The redox-dependent binding of Nucleoredoxin to Dishevelled could be a new regulatory mechanismus in the wnt signaling pathway.