Harold Fellermann

From Santa Fe Institute Events Wiki

Harold 2004.jpg

I am a 3rd-year PhD student in Applied System Science with a genuine interest in the general system science approach and complex system studies in particular. Over the last years, I focused on mathematical modeling and computer simulation of living systems, working in the domains of systems ecology, systems biology, and currently systems chemistry in the area of artificial life.

1. What are your main interests? Feel free to include a "pie in the sky" big idea!

Currently, I am mainly interested in the modular and hierarchical organization of living systems, where distinguishing features can be found on any scale, ranging from the self-assembly of biomolecules, over cell organelles and tissue to whole organisms on the one side, and from individuals over communities, populations and metapopulations toward the whole biosphere as one interacting living system on the other side.

What causes characteristic scales? How can predominent scales in data guide model development and model simplification? How do the dynamics on one scale influence, determine, and restrict the dynamics on another scale and how can one formalize such "vertical interactions" between scales?

Following these questions, I am interested in various aggregation and dimensional reduction techniques like hierarchical aggregation, singular perturbation methods, meanfield approaches, or Mori-Zwanzig projections, and methods to combine models on different scales in multiscale modeling frameworks as well as methods developed by statistical physics to deal with high dimensional systems and their extensions toward systems with complicated interactions.

I am confident, that these methods are applicable beyond the scope of biology, and I would like to dive into other disciplines -- primarily social systems -- in the long term.

2. What sorts of expertise can you bring to the group?

I am an experienced modeler with background in mathematics, computer science, biophysics and biochemistry. I can implement and analyze mathematical models / computer simulations independently and comparably fast. I am familiar with many methods and concepts of complex systems theory. Given that I worked in multi-disciplinary projects for a long time, I am skilled in communicating with people from other scientific domains.

3. What do you hope to get out of the CSSS?

My expectations are two-fold. First, I am eager to learn concepts of complex systems theory from first hand. I hope that this will help me to move from the mere application of modelling techniques toward own work on the development of methods and concepts in complex systems research. Second, I hope to meet exciting people from other scientific domains that share a common interest in the systems science approach.

4. Do you have any possible projects in mind for the CSSS?

From my own work, I can come up with some projects that might be feasable as school projects:

i) "Plasticity, learning, and self-maintenance at the origin of life"

In order to be robust, early life must have been able to cope with unforeseen hazardous conditions in its environment. Ultimately, a reaction network's ability to recognize and classify input from its environment can be formalized by a learning algorithm. While evolutionary algorithms are the primary target of investigations in this area, it might be interesting to see, how other learning algorithms (e.g. independent component analysis, neural networks) can be implemented by math reaction kinetics. Are the requirements of such learning systems met in primordial chemistry? How is this kind of learning related to evolutionary adaptation? I started to discuss these and similar questions with Garrett Kenyon (LANL) but did not have time yet to start this collaboration. In case anybody is interested, this might be a cool thread to proceed with.

ii) "Biological information"

Information theory originated in communication theory. Although the concept of information had been numerously applied to biology (e.g. by quantifying the Shannon entropy of DNA), its assumptions are hard to pin down in a biological context -- e.g. what is the sender and receiver of a message in molecular biology? On the other hand, it is obvious that biological systems process information (e.g. a seed deciding whether to sprout now or later). It would be interesting to discuss and develop a more rigorous framework of biological information on the basis on actual chemical processes rather than by counting bits in the DNA. While this is one of my genuine interests, it might be too vague to work on this in the summer school.

iii) Improving numerical fits for power law exponents

It is well known that power law data is intriguingly hard to analyze, because preconditions of standard fit procedures (e.g. least squares) are not met by power laws. The state of the art method is maximum likelyhood estimation. While performing very well on power law data, this method has a problem when the power law data is pre-processed, e.g. when it has been grouped into an equidistant histogram. It would be a good contribution to the literature to analyze the effect of data binning on the prominent fitting procedures and quality-of-fit estimators.

In general, I am also interested to work on any project concerned with dimensional reduction and hierarchical clustering, not necessarily in the area of biology. But I do not have enough knowledge acquired to actually propose a project here.