| Article ID: | iaor200973304 |
| Volume: | 58 |
| Issue: | 3 |
| Start Page Number: | 305 |
| End Page Number: | 318 |
| Publication Date: | Apr 2009 |
| Journal: | Optimization |
| Authors: | Li D, Sun X L, Niu S F |
| Keywords: | programming: integer, programming: branch and bound |
We consider in this article a factor model in portfolio selection with roundlot constraints. Mathematically, this model leads to a quadratic integer programming problem. We exploit the separable structure of the model in order to derive Lagrangian bounds. A branch-and-bound algorithm based on Lagrangian relaxation and continuous relaxation is then developed for solving this model. Computational results are reported for test problems with up to 150 securities.