Optimization and Systems Theory Seminar
Friday, May 11, 2001, 11.00-12.00, Room 3721, Lindstedtsvägen 25

Jörgen Blomvall
Division of Optimization
Department of Mathematics
Linköping University
E-mail: johan@mai.liu.se

Optimization of financial decisions using a new stochastic programming method

Stochastic programming is an emerging area in optimization where one can determine optimal decisions under uncertainty. This type of problems has received increasing attention since more realistic problems can be solved today. We have developed a primal interior point algorithm that can solve multistage stochastic programs with nonlinear convex objective and global constraints. We have used the algorithm to solve problems with up to 5.8 million scenarios and 100 million variables. The algorithm has also been implemented on a pc-cluster. The achieved speedup is close to perfectly linear. We have also used the algorithm to optimize a stock and option portfolio. By using optimization we could increase the portfolio return substantially. The tests were made over a ten-year period.
Calendar of seminars
Last update: April 2, 2001 by Anders Forsgren, anders.forsgren@math.kth.se.