Description

The goal of this software practical is to develop a description of a moderately general case of an optimum experimental design problem, together with software that can convert this description into running code. This is done using tools that are well suited to this task, making the practical accessible to beginner or advanced levels.

Background

Modern applications of optimization often require complex descriptions of problems, and this complexity makes it difficult to keep the applicability of code general. A specification of such a problem may include one or more models, different ways of defining controls of these models, different output quantities, and different approximations of elements that are not easy to compute.

A successful outcome of this practical would be a classification of some of these possibilities, together with a program that can read it and reformulate it in terms of components that, at least in theory, can be executed. This program should be written in Common Lisp to take advantage of some new software that has been written in the group. This language is also particularly well suited for this task.

Your gain

• Constructive experience with the difficulties of modern optimization technology, and the challenges and opportunities it presents
• A good insight into the power of metaprogramming in the context of complex optimization software
• A good overview of research subjects available at the AG Simulation und Optimierung, with a possibility to continue with a Bachelor, Master, or Diploma thesis.

Classification of the project

This project is suited for one or two students as a beginner's or advanced software practical. The demanded effort will be scaled accordingly.

Requirements

• Basic numerical analysis (e.g., course Numerik 0).
• Basic understanding of ordinary differential equations
• Interest in learning a new programming language, or better still, experience with Common Lisp

Further reading

• The book "Practical common lisp", by Peter Seibel.
• Further literature can be requested in case of interest.

Contact

Dr. Mario S. Mommer
Interdisciplinary Center for Scientific Computing (IWR)
Im Neuenheimer Feld 368
Universität Heidelberg

e-mail: mario.mommer@iwr.uni-heidelberg.de
Office: INF 368 (IWR), R 406