Professor emeritus Dr. Robert Giegerich

Dear colleagues and friends,
as you may know, I have retired from Bielefeld University in October 2014.

I am now a private scholar.

I am still associated with the Faculty of Technology at Bielefeld University as a professor emeritus, and also with the international research training group (DFG Graduiertenkolleg) Diversity and Dynamics of Genomes. Here is a link to my former group web pages.

No longer am I available for reviewing manuscripts, nor for writing or evaluating proposals, and I do not accept new PhD students. However, if you seek advice or dispute on one of the topics I have recently worked on, feel free to contact me.

• Inverse coupled rewrite systems -- ICOREs
  ICOREs are a dynamic programming framework for which I claim a unique combination of elegance and power. Dynamic programming problems are specified by simple term rewrite rules, which rewrite a solution of an optimization problem to the problem input(s). This inverse direction is much easier described than constructing an (optimal) solution in forward direction. Yet, all the dynamic programming machinery to solve the problem can be generated automatically from the ICORE. See Modeling Dynamic Programming Problems over Sequences and Trees with Inverse Coupled Rewrite Systems. The most recent conference presentation on ICOREs is I can Only read Equations.
  Note that ICOREs have not been implemented, posing quite a few challenges. Let me know if you embark on a project in this direction -- I'll be pleased to help.
• Pareto optimization in Dynamic Programming
  Pareto optimization performs combinatorial optimization under several independent objectives. It avoids an artificial amalgamation of objective functions, such as pseudo-energies or ad-hoc probabilities. It computes the set of solutions not dominated in all objectives by any other, including a significant portion of "ghost solutions" that cannot be made visible with amalgamated objectives. Pareto optimization is always mathematically correct in (algebraic) dynamic programming. For the idea, see Pareto optimization in algebraic dynamic programming. When you use an algebraic dynamic programming system such as Bellman's GAP, its implementation comes for free! See Integrating Pareto Optimization into Dynamic Programming.
• The BRaliBase Dent
  For a decade, many of us working on RNA structure prediction have been puzzled by a lapse of performance of our algorithms in the "twilight zone" below 60% sequence identity. Many clever algorithms were invented to fight the inglorious "BRaliBase Dent". Finally, we could explain that the dent is a feature of the benchmark data set rather than one of our algorithms. See The BRaliBase dent -- a tale of benchmark design and interpretation

My Bielefeld University e-mail address will go out of service in the not-so-far future, and my new address will not be posted here. In case of need, contact any of my recent co-authors in the above publications, they will know about the new address.

Take care

Robert Giegerich