Jacek Gondzio

Professor Jacek Gondzio

School of Mathematics
The University of Edinburgh, Scotland, UK

J.Gondzio@ed.ac.uk

CV and Publications
Selected recent talks
Video lectures
Popular press
HOPDM IPM code
OOPS IPM code

Post-Doc Position in Large-Scale Optimization
Deadline for applications: March 1st, 2013. Job description.

Research Interests

Large scale optimization

interior point methods for LP, QP and NLP
cutting plane methods for convex optimization
simplex-type methods for LP

Sparse matrix methods in optimization

linear algebra techniques of optimization
structure exploiting techniques in dynamic, stochastic or network optimization

Parallel computing techniques in optimization

parallel methods of linear algebra
parallel decomposition methods

PhD/MPhil Students:

Marco Colombo (2003-2007: PhD) worked on theory and implementation of interior point methods;
Cathy Buchanan (2003-2007: MPhil) worked on optimal control problems;
Ghussoun Al-Jeiroudi (2004-2009: PhD) worked on iterative methods for KKT systems;
Kristian Woodsend (2005-2009: PhD) worked on optimization methods applied in data mining;
Xi Yang (2005-2010: PhD) worked on optimization problems arising in risk modelling;
Qun Zhang (2009-2010: MSc by Research) worked on quantitative analysis of structured financial products;
Pablo González (2009-) works on interior point methods for combinatorial optimization;
Kimonas Fountoulakis (2011-) works on applications of matrix-free interior point methods;
Robert Gower (2012-) works on nonlinear optimization and interior point methods.

Very Large-Scale NonLinear Optimization
Together with Andreas Grothey we solved (May 2005) a large financial planning problem, a quadratic program with 353 million rows and 1010 million columns.

Read about the Matrix-Free IPM (Oct 2009).

Survey on IPMs: IPMs 25 Years Later (Feb 2011).

Software development:
Linear, Quadratic & Convex Nonlinear Optimization with IPMs:
HOPDM (Higher Order Primal-Dual Method) is an efficient implementation of the multiple centrality correctors variant of the primal-dual method for large scale linear and convex quadratic optimization. HOPDM compares favourably with commercial LP/QP packages. Its earlier version is available for research use. The code has recently been extended to solve smooth convex nonlinear optimization problems.

Parallel Large-Scale Optimization with IPMs:
OOPS (Object-Oriented Parallel structure exploiting interior point Solver) is a new generation of software for optimization of unequalled computing power.

24 students attended the Operational Research MSc programme in 2002/2003.
27 students attended the Operational Research MSc programme in 2003/2004.
23 students attended the Operational Research MSc programme in 2004/2005.
31 students attended the Operational Research MSc programme in 2005/2006.
41 students attended the Operational Research MSc programme in 2006/2007.
46 students attended the Operational Research MSc programme in 2007/2008.
52 students attended the Operational Research MSc programme in 2008/2009.
41 students attended the Operational Research MSc programme in 2009/2010.
58 students attended the Operational Research MSc programme in 2010/2011.
55 students attended the Operational Research MSc programme in 2011/2012.
65 students attended the Operational Research MSc programme in 2012/2013.
We have started enrolling for the 2013/2014 OR MSc.
For information see Edinburgh OR MSc.

Teaching in 2012/2013:
MATH11111: Fundamentals of Optimization (Semester 1, Y4 Maths and several MSc's)
MATH11032: Large Scale Optimization (Semester 2, Y4 Maths and several MSc's)

Address and phone numbers

J. Gondzio School of Mathematics The University of Edinburgh James Clerk Maxwell Building King's Buildings, EH9 3JZ Edinburgh, UK.
Phone: +44 131 650 8574
Fax: +44 131 650 6553
E-mail: J.Gondzio@ed.ac.uk