Optimization is ubiquitous in science, engineering and economics. Optimization appears in parameter estimation and data-fitting, optimal control, model predictive control, machine learning, modern signal processing, image deblurring, tomography etc. Most real-world optimization problems cannot be solved analytically but must be solved numerically.
In the Scientific Computing section at DTU Informatics we teach and do research in numerical algorithms for continuous optimization. We have a strong interest in modern convex optimization including conic optimization as well linear and quadratic programming tailored for specific engineering applications such as for instance model predictive control.
The research projects within numerical optimization at the Scientific Computing section at DTU Informatics involve development of new algorithms and software for numerical optimization. Some of this development is done in Matlab while high performance implementations are conducted in C/C++ and Fortran. We implement algorithms for constrained and unconstrained optimization, convex and non-convex optimization, and usually tailor these algorithms for specific engineering applications.