Research Units

SCIENTIFIC GOALS

The boundaries between the sciences are ever dissolving, and thus interdisciplinary research becomes an imperative. Here in Bonn, together with strong partners from a number of local research institutions (Max-Plank, Helmholtz and Fraunhofer institutes/centers, see the Figure) we have accumulated tremendous expertise in theoretical sciences, including in particular scientific computing, modeling and simulations. One overarching theme is the formation of complex structures from basic building blocks, covering scales from the femto-universe over scales pertinent to life up to the universe as a whole. It is the main intention of this cluster to strengthen this type of research through interdisciplinary projects across seemingly unrelated scientific fields that connect and improve the research from the respective domains, allowing for a strong and fruitful synthesis between the various sciences, including a number of emerging data-driven areas. Such type of forefront research inevitably leads to philosophical questions that will also be explored in this cluster.

The research in CASCADE will go beyond the individual areas of numerical simulation, theoretical physics, theoretical chemistry, computational life sciences and theoretical philosophy through interdisciplinary research. Most of these fields belong to the various departments of the Faculty of Mathematics and Natural Sciences (MNF) at the University of Bonn (UBonn). In addition, we include scientists from the Medical and the Philosophical Faculty. We will exemplify this interdisciplinary computationally-based approach in three research units (RUs) within the natural sciences. Moreover, the newly founded Center for Science and Thought (CST) will serve as the bracket between the natural sciences and theoretical philosophy through interdisciplinary projects.

To be more precise, the research in the cluster is organized along different length scales and methods that allow one to tackle the pertinent questions and will ultimately allow to go beyond what can be achieved in single disciplines. Correspondingly, the RUs are set up and will be briefly described now.

RU-A "Structure of complex matter", which is focused on fundamental physics, strongly interacting quantum systems and quantum chemistry, deals with scales that range from femtometers (hadrons and nuclei) to Ångström (molecules). Here, the forces underlying the formation of structure are known, namely the electromagnetic, the weak and the strong interactions of the successful Standard Model of particle physics. Nuclei further form stars and in particular, neutron stars, which play an important role in the generation of gravitational waves, see Figure. However, the formation of these structures is by far not understood and the calculation of their properties can only be achieved using high performance computing, supplemented by the developments of many-body theories and the investigation of the underlying mathematical structures. Common to quantum physics and quantum chemistry are the difficulties that arise from the treatment of strong correlations that either appear directly through the strong forces in the formation of hadrons and nuclei or through strong electronic correlations in the formation of molecules or low-dimensional materials like graphene or carbon nanotubes. These challenges can only be tackled successfully through a direct interplay with the research projects in RU-C that will provide the required computational tools. This is supplemented by investigations in fundamental physics that naturally open a bridge to philosophical questions which will be tackled in the Center for Science and Thought.

RU-B "Computational neurosciences" focuses on computational neurobiology as well as complex behavior and simulation and analysis in clinical neurodegeneration. In the biosciences, one is dealing with scales that range from molecular, cellular, organ and organism level up to entire populations. Recent years have witnessed remarkable advances in data acquisition that are  driving a major transformation in biomedical research. The field is not yet in the state to make best  use of the truly high-dimensional data and the opportunities they bring. Moreover, on the conceptual level, integration with the theoretical data sciences remains in many ways shallow. Therefore, a deep integration of data, models and computation has the potential to yield dramatic progress in biomedicine. Focusing on the two specific examples that we detailed below, we will develop concepts and frameworks that go far beyond current computational pipelines. A specific  example for the development of next generation epidemiology is depicted in the Figure.

Again, the interplay with RU-C is of particular importance in the fields of machine learning, data visualization and numerical analysis. Furthermore, direct links to the CST are given through questions concerning the notion of causality in life sciences.

RU-C "Extreme scale computing" provides the computational backbone for the projects in RU-A and RU-B. Only with this available the aforementioned challenges can be investigated. RU-C is designed in a truly multiscience way by bringing together mathematicians and computer scientists with scientists from the life and natural sciences.
Extreme is to be understood in two ways: first, in particular but not only the projects in RU-A require extreme amounts of supercomputing resources. In order to make significant progress here we will investigate numerical algorithms and their application to the specific challenges formulated in CASCADE. This will also include the scaling to future exascale computer installations. Second, extreme amounts of data need to be processed and explored. Due to the unique expertise in the cluster data, analytics methods will be tailored specifically for the science problems at hand. The feedback from science will be used to improve the understanding of machine learning methods. Of particular importance in data exploration is the visualization of data in-situ. Parallel to the investigation of novel methods and algorithms and their application RU-C will also focus on architecture development. Modular supercomputing will be developed further to pave the way towards exascale computing codesigned with input from the sciences pursued in the cluster. Modular supercomputing is particularly well suited for CASCADE because it aims to provide modules for a large variety of different application codes. In our computational science concept, RU-C is a natural connector to the research performed in RU-A and RU-B. RU-C also links naturally to research in the CST, as machine learning and artificial intelligence are also considered there.

The Center for Science and Thought: The CST is a radically interdisciplinary platform, designed
to address urgent questions that arise at the intersection of the scientific disciplines involved in CAS-
CADE. It is based on a bottom-up-epistemology: Hard questions, that emerge at the frontiers of
science and have interdisciplinary impact, will be analyzed in five subprojects clustering around re-
search topics identified through workshops and conferences.
Currently, in the wake of the computational turn, we are in a position to run simulations of models and test their theoretical adequacy even in domains where empirical testing is out of reach for the time being. This raises the central question how we conceptualize the borders of scientific knowledge and extend them into regions presently only accessible to metaphysics. Metaphysics is the discipline which deals with problems that are only accessible for theoretical evaluations. It overlaps with science at its frontiers. The CST brings science and philosophy together in order to achieve a better understanding of the current limits of scientific knowledge and to discuss the overall issue what, if any, the principle limits of scientific knowledge are. This issue is addressed within science itself in the form of case studies that arise from research in the RUs of CASCADE, see the schematic representation below.