The University of Hamburg which excels in both the natural sciences and in computer science provides an excellent environment for interdisciplinary research and teaching. Information technology can now be found in almost all areas of social life, in the natural sciences it has a very long-standing tradition. Not only technologically sophisticated experiments are dependent on information technology, also scientific data management and global communication are cornerstones in the natural sciences. Of central importance, however, is the modeling of scientific phenomena. Computer simulations thus supply major contributions to a better comprehension of nature.
In the modern life sciences, computer science is omnipresent. Computers brought about a revolution in genomics a few years ago. For the very first time due to advanced laboratory technology and computer science methods, it became possible to decipher the human genome to a length of 3.1 billion base pairs. Today, the genomes of many mammals, plants, bacteria and viruses are available for research. Through the advancement of technology, individual genetic information can be used to promote a better understanding of the processes in living organisms. This particularly benefits modern medical research.
Modelling, Simulation and Visualization
Using state-of-the-art spectroscopy methods, we can penetrate cosmic structure down to the most minuscule detail. Many thousands of currently available structures, in particular those of essential function owners in the living organism such as proteins form the foundation for modern molecular biological research. Computers play an important role not only in the elucidation of structure, they are also widely used for modeling, simulation and visualization. For example, computer models can characterize the flexibility and stability of molecules, thereby predicting reciprocal action. Particularly in pharmaceutical research, new drugs are developed by means of computer-based active ingredients.
Scientific Computing and Visualization
The high-resolution simulation of time-dependent phenomena requires not only computers whose power demand can only be met by high-level parallelization and the use of special processors. For data analysis and visualization of the resulting strong increase in data results, innovative methods are required that do not interfere with the massive parallel calculation. For example, techniques for the extraction of 3D graphics from spatially distributed raw data – e.g. temperature, humidity and currents – are developed and carried out in conjunction with the simulation on a supercomputer. The researcher may then access the prepared results via a streaming process as a 3D movie in a virtual reality system, enabling an interactive exploration of complex interrelationships.
These days all sciences accumulate huge amounts of data. They can be found in tomography, sequencers, microscopes and other equipment and or as a data result of numerical simulations. Especially in the field of climate research, this data is kept for decades. It is stocked in extensive disk storage and tape libraries that need to be managed efficiently. Computer science concepts allow worldwide access to this data, which serve not only to gain knowledge in the field of Earth system science, but also, for example, to be merged with data from the social sciences in order to arrive at entirely new insights. This is called data-intensive science. It is a new, stand-alone component in Computing in Science. Computer science enables efficient storage, analysis and graphical representation of data and transforms it into information and ultimately into new scientific knowledge.
The field of Machine Learning assists people in the evaluation of ever larger and more complex data sets. In many applications in science and industry, huge amounts of data are compiled, collected and stored. In order to use this data and evaluate it, there is a tremendous need for methods that can pick out from the flood of data interesting information. The emergent discipline Machine Learning is dedicated to this objective. Here algorithms are designed to analyze - on a basic level - large amounts of complex data, aiming to discover interrelationships or to clarify specific issues. Machine Learning algorithms may be deployed in numerous applications, for example in image processing or in automatic language processing. In addition, more and more scientific disciplines, for example biology, neuro-science, physics or medicine are discovering the potential of Machine Learning for evaluating their empirical data.