|
Today complexity is challenging
us in many areas of science, such as biology, neuroscience,
computer science, economy, and sociology. Nervous systems,
distributed computing systems, financial markets and social
systems are all instances of complex systems. A complex system in
general is a system consisting of a great number of interacting
units or constitutes. While some of the above areas are rather
concerned with investigation, others focus on design and
engineering. The design-oriented perspective on complex systems especially
applies to synthetic biology and computer science. Synthetic
biology addresses the design and construction of biological
devices and systems that for example process information,
fabricate materials and structures, or maintain and enhance human
health. This task is intrinsically complex, since scores of
interactions between various molecular processes have to be
organized. Due to a continuous trend of miniaturization,
expressively reflected in Moore's law, and the advance of
massively parallel processing, already now, computer science has
to deal with complexity beyond the imagination of system
designers. Miniaturization soon comes along with a loss of both
determinism and detailed insight into systems at the microscopic
level. These trends and new areas of research in information
technology, such as molecular computing or quantum computing,
lead to the expectation that as a future trend computer science
and synthetic biology will converge to each other. The challenge
for both is to organize systems of interacting processes in a
way, which guarantees the desired behavior of the system on a
global level.
For this purpose, organic computing suggests a goal-oriented
system design based on self-organized processes. This requires a
system to be describable as a set of goals and subgoals such that
each subgoal can be implemented as a self-organized subsystems.
Hence, an organic system design implies the structuring of
systems in a number of hierarchically ordered functional
subsystems corresponding to a hierarchy of goals, as depicted
below.
This
raises the question how to define goals. Within the dynamical system
approch goal-orientation corresponds to convergence and the goals
are given by the attractors of the system. While the interpretation
of complex systems as computational systems allows for the
numerical simulation of their dynamics, it leads to an
explanatory problem, since the semantics of goal-orientation is
not reflected in a sequence of computational states. For this
reason, our dynamical system approach towards an organic system
design is characterized by the focus on semantical aspects of
dynamical systems that arise from specific topological structures
associated with the state variables. In this view dynamical
systems appear in a self-explanatory way and goals can be defined as
patterns within the underlying topological space.
Our approach was successfully applied to the problem of
scheduling medium-grained parallel computations in distributed
computer systems. In this context the desired patterns that
represent the goal can simply be described as clusters of well-connected
nodes. Using this goal representation we designed a dynamical
system which is illustrated in a computer
animation based on a numerical simulation of the system. Now, our
research also addresses applications in the broad area of
synthetic biology. Future results will be presented on this site.
Comments,
questions and suggestions are welcome. Send them to Alexander Sinsel.
|
