The John von Neumann Institute for Computing (NIC) at the Research Centre Juelich, Germany is one of the leading supercomputing centres in Europe. Besides providing state-of-the-art hardware and software facilities, NIC focuses on research in scientific high-performance computing. In order to deliver excellent support to its users, it aims to gather and maintain competence in key areas of computational science. It stimulates interdisciplinary cooperation and promotes the knowledge transfer between the centre and the computational science community.

Founded as a national supercomputing centre in the mid-eighties, the NIC currently provides an increasing amount of resources to European researchers within the context of EU-funded projects or Europe-wide scientific collaborations. In summer 2004 the NIC started an initiative to partner with the new member states of the European Union. Outstanding research groups are now given the opportunity to exploit the supercomputers at NIC to boost their investigations in cutting-edge computational science.

The first part of the talk introduces a review of the organisational structure of the centre, its current supercomputer systems, and its user support. In the second part of the talk the NIC initiative is presented and the procedure of how to apply for supercomputer resources at NIC is described in detail.

back

First, I will give an overview of recent trends in the field of multiphysics multiscale simulation, as I observed it at the ICCS workshops on Simulation of Multiphysics Multiscale Systems during the past 3 years. http://staff.science.uva.nl/~valeria/smms/

Second, I will tell about my research topic: simulation of plasma chemical deposition of thin films for semiconductor electronics. A number of intertwined processes occur in such complex systems: plasma discharge processes (ionization, dissociation, excitation, attachment, etc.), convective and diffusive transport, heat transfer, homogeneous and heterogeneous chemical reactions, adsorption and desorption, film defect formation, etc. The time and length scales of these phenomena differ by many orders of magnitude. Different models and codes have been developed to take into account all the processes, along with the interfacing algorithms for their coupling on different levels.

If time permits I can tell also about the problem solving environment (Virtual Reactor) that was developed to facilitate the problem description, simulation setup and results analysis. The GUI includes various editors, pre- and postprocessors, visualization and archiving modules. The Virtual Reactor computational environment was recently modernized to allow simulation on heterogeneous Grid resources.

back

The recognition of still unknown mechanism of signal transduction in immunoglobulin meets limitations in the analysis of this very complicated process of organism defense against antibodies. One step of multiple steps process - the signal transduction from antigen binding site to complement binding site - is of particular interest because the distance between these two active sites is significantly larger than cut-off distance of any non-bonding interaction. Thus the mediation in this process seems to be necessary. The comparison of normal molecules with pathological ones based on experimental observations suggests presence of structural elements in protein molecule necessary to be responsible for particular functions.

The verification of hypotheses related to this phenomenon is possible only on the basis of in silico experiments.

Computational techniques allow simulation of processes in normal and pathological protein molecules. The aim oriented mutations – observed also in real pathological proteins – influence dynamics of protein revealing mutation-related-specificity.

The molecular dynamics simulation seems to be the best tool to recognize and differentiate particular parts of protein molecule and particular function-related structural forms.

The simulation of molecular dynamics applied to immunoglobulin molecule revealed the critical role of short Beta-structural fragment in VL domain, which correlates motions in VL and CL domains. These domains deprived of this short ordered structural element fluctuate randomly eliminating the aim orientation of structural changes in immunoglobulin molecule. Molecular dynamics simulation enabled identification of function-related structural changes revealing the mechanism of signal transduction in immunoglobulin molecule.

The problem can be classified as representing the discipline of computer aided drug design. The presence of foreign (non-protein) molecules is able to correct the malfunction of the system, what can be observed in molecular dynamics simulation of the protein-drug complex molecular dynamics simulation.

The large scale computing is necessary to solve the problem presented. The standards in simulation of molecular dynamics expect long period of time to be covered in simulation (min 10 ns with 1 fs time step). Large size of protein molecule under consideration (above 4000 of atoms + thousands of atoms of water molecules) makes the computation additionally highly time consuming.

The grid systems making large computer resources available for wide audience facilitate solutions of pharmacological projects. The EUCHINA Grid project linking the computer resources of Europe and China in one common network give an excellent opportunity to develop computer aided drug design on large scale. The problems solved on the babsis of this project will also be presented.

back

We developed a Cellular Automata based model to simulate growth and form in stony Corals (such as Pocillopora Damicornis). The goal of the research is to investigate how multi-scale/multi-physics models can assist us in understanding embryogenises at large.

Keywords: Multi-physics modelling, Cellular Automata, Growth and Form

Literature: http://www.science.uva.nl/research/scs/papers/sloot.html

back