LULEÅ UNIVERSITY OF TECHNOLOGY


Research

My research interests are in scientific computing and numerical analysis. More specificly algorithms for robust regression, piecewise quadratic techniques to solve linear programming problems, and ortogonal factorization techniques for sparse matrices. I am doing this in cooperation with Håkan Ekblom at Luleå University of Technology, and Kaj Madsen and Hans Bruun Nielsen at the Technical University of Denmark.

The 9:th of September 1999 I defended my Ph.D. thesis, Solution of Linear Programming and Non-Linear Regression Problems Using Linear M-Estimation Methods. It can be downloaded by clicking here.

The MATLAB code that was used in the examples in Articles I and II of the thesis can be downloaded here.

spLQ - Sparse LQ-factorization and up/down-dating

Finally I am releasing the software presented in

Ove Edlund, 2002, A Software Package for Sparse Orthogonal Factorization and Updating, ACM Trans. Math. Softw., 28, 4, p. 448-482.

Download it here.

Constrained M-estimation for regession

This is a cooperation with Olcay Arslan and Håkan Ekblom. The algorithm is described in general terms in

Arslan O., Edlund O. & Ekblom H. (2002), Algorithms to compute CM- and S-estimates for regression, Metrika 55, 37-51.

The Matlab implementation has recently been submitted to the Journal of Statistical Software along with a paper:

The new paper contains some important information on the choice of parameters for CM estimates.