We provide a computational economics seminar at the University of Mannheim since two years. An introduction to numerics is given in around six sessions where we present the first six technical chapters of Miranda/Fackler: Applied Computational Economics and Finance (linear and nonlinear systems, integration, approximation, optimization). Afterwards the students choose a topic and write a short paper about it within six weeks. The two general directions are numerics and economics. The numerical papers are usually about programming some numerical methods and discussing their properties (e.g. polynomial vs. finite element approximation for smooth and unsmooth functions). The economic papers are usually less concerned about programming and often use the CompEcon Matlab toolbox accompanying the book of Miranda/Fackler to discuss an economic model (e.g. approximation errors of linearized, certainty equivalent policies in a standard dynamic model). Our experience is that it is much easier to understand numerical methods when algorithms are presented together with the code implementing them. Therefore we also provide an introduction to Matlab at the beginning of the seminar and try to review code examples during the six technical sessions
