|
The brochure of the John von Neumann Institute for Computing
is available in English and in German. It can be ordered at the NIC secretariat
(nic@fz-juelich.de).
deutsche Broschüre (pdf) | English brochure (pdf)
 Other Fields of Application
The following examples provide an additional impression of the large
variety of issues explored
at NIC. Especially, the number of applications in the life sciences
has increased significantly in
the last few years.
Epilepsy manifests itself not only in visible seizures, but also
in characteristic patterns in the
electroencephalogram (EEG). It is thus an important field of application for modern methods of
nonlinear time series analysis. In collaboration with the Department of Epileptology at the
University of Bonn, we are pursuing two main aims. On the one hand, we develop methods for
localizing epileptic foci in patients who are possible candidates for surgery.
The figure shows regions
with elevated (green) and high (red) likelihood of being a focus in a
retrospective analysis. Also
shown is the region which has actually been removed, based on other analyses (black). In the
case illustrated, the primary focus had been correctly recognized and
removed, and the patient is
now free of seizures. An advantage of our method over previous ones is that it uses only data
from seizure-free epochs, and that their predictions do not always agree with those of other
methods. In some rare cases where the patient is not completely free of seizures, our analysis
would suggest that there is indeed a secondary focus or that the primary focus has not been
removed entirely.
A second - and much more ambitious - aim of our research is to predict seizures. Finally, our
work is also aimed at the development of new methods of nonlinear time series analysis which
will then be applied to other fields in interdisciplinary collaborations.
(Ralph Gregor Andrzejak, Thomas Kreuz, Alexander Kraskov,
Peter Grassberger, NIC Research Group "Complex Systems", Jülich;
Florian Mormann, Klaus Lehnertz, Christian E. Elger,
Department of Epileptology, University of Bonn; and Peter David,
Physics Department, University of Bonn)
The figure shows results obtained by synchronization tomography. Magnetoencephalography is
used to record the magnetic fields produced by the working brain.
The underlying brain currents
are estimated. Synchronization between different brain areas as well
as between brain areas and
muscular activity is determined for each time slice. This enables the
anatomical localization of
synchronization processes in the brain, for example, the synchronization
between the extensor
muscle of the right index finger during slow self-paced finger movements
and all brain volume
elements in the corresponding anatomical cross-sections as shown in the
figure. Synchronization
tomography revealed that synchronization processes between brain areas
may change drastically
during changes of motor control strategies, whereas the amount of
activation of all participating
brain areas may remain unchanged.
(Peter Tass, Institute of Medicine, Research Centre Jülich)
Ion channels are responsible for the transport of ions through the cell
membrane and, therefore,
they are crucial for signal propagation along nerve cells. The dynamics of two interacting
populations of ion channels embedded in a flat biomembrane near a narrow cleft and in the
presence of an ion-density gradient across the membrane is the first example of a new class of
nonlinear pattern-forming systems.
The picture shows a snapshot of the spatial deviation of the density
from its constant mean value
for one of the two ion-channel populations. The major characteristic of this system, which
distinguishes it from many other pattern-forming systems, is, on the one hand, the locally
oscillatory dynamics and, on the other hand, a global conservation law, i.e. ion channels are
neither created nor annihilated during the dynamical evolution of the
pattern. Besides intermittent
spatio-temporal chaos, the system also exhibits previously unknown und surprising coarsening
phenomena.
(Markus Hilt, Walter Zimmermann, Theoretical Physics, University of Saabrücken)
Energetic ion beams are used for a variety of important applications,
such as isotope production,
or hadron therapy in the treatment of certain cancers. The large
mass of ions compared to electrons or
photons allows them to penetrate long distances into solid matter,
depositing their energy exactly
where it is needed. Recent advances in short pulse laser technology
have ushered in a new class
of compact "tabletop" ion sources, which exploit the extremely
high electric field strengths (~GVcm-1) produced at
the surface of irradiated solid targets to accelerate ions to multi-MeV
energies. Despite these experimental successes, knowledge of the
underlying acceleration physics
is still sketchy and even controversial. At ZAM we are tackling
this problem using a new mesh-
free plasma simulation technique, in which Coulomb forces
for 106 - 107 particles are computed
directly using a parallel tree code. The example illustrated
here shows a 1 µm-radius "wire"
irradiated by a 50 terawatt laser (from the left). After
about 500 fs, the entire mid-section of the
wire has been pushed out, forming a beamlet with a mean
ion energy of 3 MeV. By performing
several series of such simulations with various laser and
target parameters, we can deduce scaling
laws with which to optimize the ion beam properties and improve the source design.
(Paul Gibbon, NIC-ZAM, Jülich)
Recognizing and quantifying common features of objects is fundamental for all sciences. An
important measure of such similarities or dependencies is the
so-called mutual information. Its
numerical estimation from actual data is often difficult, but
improved methods have been
developed in our group - together with methods for clustering
objects based on them. There are
numerous applications for these methods. They range from phylogenetic
trees (here for 32
mammal species, based on their mitochondrial DNA) to the
decomposition of signals into least-dependent components.
In the lower figure, the latter is
applied to the electrocardiogram (ECG)
of a pregnant woman. After decomposing the signals, the
components corresponding to the
mother and to the fetus were added up again separately.
In this way practically noise-free fetal
ECGs can be obtained.
(Harald Stögbauer, Alexander Kraskov, Ralph Gregor Andrzejak,
Peter Grassberger, NIC Research Group "Complex Systems", Jülich)
The regulation and transduction of signals between cells and their environment are mediated in
many cases by channels and pores residing in cell membranes. They are an important class of
biomolecular machines. They work by selecting specific ions and by catalyzing the passive
diffusion through the selectivity filter. The picture shows the potassium
channel KcsA residing
inside a lipid membrane surrounded by water molecules. Controlled molecular dynamics
simulations were conducted with this model system that consists of about
30,000 atoms. The
simulations permit the collective motion of ions and water molecules to
be monitored through the
narrow selectivity filter. The simulations reveal that the high conductivity is based on the
cooperative diffusion of ions and water molecules mediated by the flexible carbonyl groups
lining the selectivity filter. The schematic diagram on the right hand
side shows four successive
configurations of ions and water molecules passing through the filter.
(Jean-Fang Gwan, Artur Baumgärtner, Institute of
Solid State Research, Research Center Jülich)
|
